article
stringlengths 0
745k
| abstract
stringlengths 0
92.7k
|
|---|---|
this paper explores a basic case of what we believe is a general connection between exact lagrangian submanifolds in the complement to an ample divisor @xmath5 in a complex calabi yau manifold @xmath6we view @xmath7 as an exact symplectic manifold and coherent sheaves on a scheme _ defined over @xmath3 _ , the ` mirror ' to @xmath7 .
we take @xmath6 to be an elliptic curve ; its complex structure is irrelevant , so it is really a 2-torus @xmath8 .
we take @xmath5 to be a point @xmath9 .
the mirror is the weierstrass cubic @xmath10 , the restriction to @xmath11 of the tate curve @xmath12 .
kontsevich s 1994 homological mirror symmetry ( hms ) conjecture @xcite claims that the fukaya @xmath13-category @xmath14 of a polarized calabi
yau manifold should have a formal enlargement precisely formulated a little later as the closure @xmath15 under taking mapping cones and passing to idempotent summands which is @xmath13-quasi - equivalent to a dg enhancement for the derived category of coherent sheaves on the ` mirror ' @xmath16 , a calabi
yau variety over the field of complex novikov series .
are more subtle than those claimed by our one - sentence prcis of kontsevich s conjecture . ]
the hms conjecture has inspired a great deal of work in symplectic geometry , algebraic geometry and mathematical physics ; the hms paradigm has been adapted so as to apply not only to varieties whose canonical bundle @xmath17 is trivial , but also to those where either @xmath18 or @xmath17 is ample , with such varieties playing either symplectic or algebro - geometric roles .
meanwhile , progress on the original case of calabi yau manifolds has been slow .
there are currently complete mirror - symmetric descriptions of the fukaya category only for the 2-torus @xmath19 and of the square 4-torus @xmath20 @xcite .
the case of calabi yau hypersurfaces in projective space has been solved up to a certain ambiguity in identifying the mirror variety @xcite .
there are significant partial results for linear symplectic tori of arbitrary dimension @xcite .
our contention is that even in the solved cases , there is more to be said about hms .
the fukaya category for the 2-torus has a natural model which is defined over @xmath21 , a subring of the complex novikov field .
this model has a mirror - symmetric description as the perfect complexes on the tate curve @xmath22 over @xmath21 .
the symplectic geometry of the torus is thereby connected to the arithmetic algebraic geometry of @xmath22 .
establishing this connection is the task of this article .
experts have certainly been aware that , in principle , homological mirror symmetry should have an arithmetic - geometric dimension ( cf .
lecture @xcite , for instance ) , but we believe that this article is the first to treat this idea in detail . whilst refining existing proofs of hms for the 2-torus might be a viable option , our method is also new : we identify a generating subalgebra @xmath23 of the fukaya category , and show that weierstrass cubic curves precisely parametrize the possible @xmath13-structures on it ( theorem [ ainf comparison ] ) .
the mirror to @xmath24 is then the unique weierstrass curve corresponding to the @xmath13-structure belonging to the fukaya category .
our identification of this mirror parallels an argument of gross @xcite but also has a novel aspect , relating the multiplication rules for theta - functions on the tate curve to _ counts of lattice points _ in triangles ( not areas of triangles ) .
our identification of the _ wrapped _ fukaya category of the punctured torus with coherent complexes on @xmath25 appears to be a basic case of an unexplored aspect of mirror symmetry for calabi
yau manifolds .
let @xmath8 be a closed , orientable surface of genus 1 ; @xmath26 a symplectic form on @xmath8 ; @xmath27 a basepoint ; @xmath28 ; and @xmath29 a primitive for @xmath26 on @xmath30 .
fix also a grading for the symplectic manifold @xmath8 , that is , an unoriented line - field @xmath31 .
these data suffice to specify the relative fukaya category @xmath32 up to quasi - isomorphism .
it is an @xmath13-category linear over @xmath21 whose objects are embedded circles @xmath33 which are exact ( @xmath34 ) and are equipped with orientations , double covers @xmath35 and gradings ( a grading is a homotopy from @xmath36 to @xmath37 in @xmath38 ) .
let @xmath12 denote the tate curve , the cubic curve in @xmath39 with equation @xmath40 where @xmath41 ( note that @xmath42 is always divisible by 12 ) .
let @xmath43 denote the @xmath21-linear differential graded ( dg ) category whose objects are locally free sheaves of finite rank over @xmath22 , and whose morphism spaces are ech complexes with respect to a fixed affine open cover : @xmath44 .
[ mainth ] a choice of basis @xmath45 for @xmath46 , with @xmath47 , determines , canonically up to an overall shift and up to natural quasi - equivalence , a @xmath21-linear @xmath13-functor @xmath48 from the relative fukaya category to the dg category of twisted complexes in @xmath43 .
moreover , 1 .
the functor @xmath49 maps an object @xmath50 representing @xmath51 to the structure sheaf @xmath52 .
it maps an object @xmath53 representing @xmath54 to the complex @xmath55 $ ] , where @xmath56 $ ] is the section at infinity of @xmath22 , and the map is the inclusion .
( this complex is quasi - isomorphic to the skyscraper sheaf @xmath57}$ ] at the section at infinity . )
it is an embedding on the full subcategory @xmath23 on @xmath58 ; and is characterized , up to natural equivalence , by its restriction to @xmath23 . see figure [ hmscorrespondence ] .
@xmath49 extends to an equivalence @xmath59 from the idempotent - closed derived fukaya category to the triangulated category of perfect complexes on @xmath22 .
3 . the specialization of @xmath49 to @xmath11 is a @xmath1-linear functor @xmath60 from the exact fukaya category of @xmath61 to the category of perfect complexes on the central fiber of the tate curve , inducing an equivalence on derived categories @xmath62 ( both of these derived categories are already idempotent - closed ) .
@xmath63 extends to an equivalence of triangulated categories @xmath64 from the derived wrapped fukaya category to the bounded derived category of coherent sheaves on @xmath25 ( these derived categories are again idempotent - closed ) .
the functor @xmath49 has an additional property , which is that it is ` trace - preserving ' , in a sense to be discussed later .
clause ( ii ) has the following corollary : there is an @xmath13 quasi - equivalence @xmath65 from the category of cohomologically unital @xmath32-modules to a dg enhancement of the derived category of unbounded quasi - coherent complexes on the tate curve .
indeed , @xmath66 is quasi - equivalent to @xmath67 as an instance of the general theory of @xcite or @xcite . at 195 10 at -20 180 @xmath68 at 185 180 @xmath9 at 60 60 at 420 240 at 420 200 at 420 160 at 300 240 at 300 230 at 200 295 at -30 300 at -30 290 and the mirror correspondence @xmath49 , for one possible choice of the line field @xmath31.,title="fig:",scaledwidth=60.0% ] [ [ comparison - to - the - standard - formulation .
] ] comparison to the standard formulation .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + the @xmath13-structure in the ` relative ' fukaya category @xmath32 is based on counting holomorphic polygons weighted by powers @xmath69 , where @xmath70 counts how many times the polygon passes through the basepoint @xmath9 . the ` absolute ' fukaya category @xmath71 , in the version most popular for mirror symmetry , has as objects lagrangian branes @xmath72 in @xmath8 equipped with @xmath73 local systems @xmath74 . in the latter version ,
holomorphic polygons are weighted by @xmath75 .
the coefficient - ring for @xmath71 is usually taken to be @xmath76 , the field of complex novikov series @xmath77 : here @xmath78 , @xmath79 , and @xmath80 . to explain the relation between the relative and absolute versions , note first that there is an equation of currents @xmath81 , where @xmath82 is a 1-current .
we take @xmath29 to be the ( smooth ) restriction of @xmath82 to @xmath83 .
there is a fully faithful ` inclusion ' functor @xmath84 } \lambda_{\c } \to \euf(t),\ ] ] linear over @xmath85 and acting as the identity on objects . for each exact lagrangian @xmath86 ,
select a function @xmath87 such that @xmath88 . then define @xmath89 on morphism - spaces @xmath90 by @xmath91 where @xmath92 is the symplectic action , defined via the @xmath93 , and @xmath94 is the exact symplectomorphism used to obtain transversality .
the higher @xmath13-terms for @xmath89 are identically zero .
the symplectic action is defined as follows . for a path @xmath95 ; 0 , 1 ) \to ( m ; l_0,l_1)$ ] ( for instance , a constant path at an intersection point ) we put @xmath96 for any disc @xmath97 , we have @xmath98 similarly , if @xmath99 is a polygon attached to a sequence of lagrangians @xmath100 ( where @xmath101 ) at corners @xmath102 , then @xmath103 from this it follows that @xmath104 , which proves that @xmath89 is a functor .
note that the perturbations that are used to define hom - spaces in @xmath32 serve equally well in @xmath71 .
it is clear that @xmath89 is fully faithful .
the ` standard ' statement of mirror symmetry is as follows .
let @xmath105 } \lambda_\c$ ] ; it is an elliptic curve over the field @xmath76 .
when @xmath26 is normalized so that @xmath106 , there is a functor @xmath107 where @xmath108 is the unique dg enhancement of the bounded derived category @xmath109 @xcite , inducing a derived equivalence ; and that this functor is again canonically determined by a choice of basis for @xmath46 : see @xcite for one proof ; @xcite for an expository account of another , occasionally missing technical details ( e.g. certain signs ) ; and @xcite for yet another .
our result is an arithmetic refinement of this standard one : [ comparing functors ] the diagram @xmath110_{e } \ar[r]^{\psi\otimes 1 } & \tw \vb(\eut ) \otimes\lambda_\c\ar^{i}[d ] \\
\euf(t ) \ar[r]^{\phi } & \widetilde{\der}^b\coh(\eut_{\lambda_\c } ) . }
\ ] ] is homotopy - commutative under composition of @xmath13-functors . since @xmath111 }
\lambda_\c$ ] is a non - singular variety over the field @xmath76 , we may take @xmath112 as our dg enhancement of @xmath113 .
then @xmath114 is the base - change functor @xmath115 . for this theorem to make sense , @xmath49 and @xmath116
must be set up so that @xmath117 and @xmath118 agree precisely ( not just up to quasi - isomorphism ) on objects .
useful references for this material include @xcite .
the _ tate curve _ is the plane projective curve @xmath22 over @xmath119 whose affine equation is the weierstrass cubic @xmath120 where @xmath121 and @xmath122 are as at ( [ tate coeffs ] ) .
so @xmath22 is a projective curve in @xmath39 . like any weierstrass curve @xmath123
, @xmath22 comes with a canonical differential with poles at the singularities , @xmath124 notation : @xmath125)$}.\ ] ] the analytic significance of the tate curve is the following .
consider the riemann surface @xmath126 , where @xmath127 .
the exponential map @xmath128 identifies @xmath129 with @xmath130 . as @xmath131 varies over the punctured unit disc @xmath132 , the riemann surfaces @xmath130 form a holomorphic family @xmath133 .
the weierstrass function @xmath134 for the modular parameter @xmath131 defines an embedding @xmath135 , q).\ ] ] this embedding is cut out by an equation @xmath136 , which is a weierstrass cubic in @xmath137 varying holomorphically with @xmath131 .
the functions @xmath138 and @xmath139 are holomorphic at @xmath11 , and moreover are defined over @xmath140 \series{q}$ ] ( making this so is the purpose of the powers of @xmath141 in the definition of the embedding ) . we can change coordinates , writing @xmath142 and @xmath143 , so as to put the equation in the form @xmath144 .
the benefit of the coordinate - change is that the coefficients now lie in @xmath21 .
the series @xmath121 and @xmath122 are those given above
so the algebraic curve @xmath144 is the tate curve @xmath22 .
we conclude , then , that the specialized tate curve @xmath145 is an elliptic curve , analytically isomorphic over @xmath146 to the family @xmath147 when @xmath148 .
its integrality is one interesting feature of @xmath22 , but another is that the absence of negative powers of @xmath131 .
one can therefore specialize @xmath22 to @xmath11 .
the result is the curve @xmath149 in @xmath150 given by @xmath151 we can characterize this weierstrass curve as follows : [ tate at q=0 ] the curve @xmath152 has a section @xmath153 $ ] which is a node of @xmath154 , the mod @xmath155 reduction of @xmath156 , for every prime @xmath155 .
any weierstrass curve @xmath157 possessing a section @xmath70 with this property can be transformed by integral changes of variable to @xmath156 .
consider a general weierstrass curve @xmath158 $ ] , given as the projective closure of @xmath159 integral points of @xmath160 , other than @xmath161 $ ] , can represented as rational points on the affine curve . the point @xmath161 $ ] is regular over any field , and is the unique point of @xmath162 with @xmath163 .
suppose @xmath164 $ ] is an integral point that is nodal mod @xmath155 for all primes @xmath155
. then @xmath165 must be non - zero mod @xmath155 for every prime @xmath155 , and hence @xmath165 is a unit of @xmath1 .
consider the @xmath1-point @xmath166 of the affine curve .
the partial derivatives vanish , since they vanish mod @xmath155 for all @xmath155 : @xmath167 the nodal condition is that the hessian is non - singular , that is , @xmath168 ( we note in passing that conditions ( [ crit ] , [ nodal ] ) hold for the point @xmath169 $ ] of @xmath170 at all primes @xmath155 . ) since ( [ nodal ] ) holds for all @xmath155 , we have @xmath171 we shall use the criterion ( [ criterion ] ) to make three changes of variable , successively making @xmath172 , @xmath173 and @xmath174 equal to their counterparts for @xmath156 . first , ( [ criterion ] ) tells us that @xmath172 is odd .
hence by a change of variable @xmath175 , @xmath176 , we may assume that @xmath177 , whereupon @xmath178 is either @xmath179 or @xmath180 .
the latter possibility is absurd , so @xmath181 .
being divisible by 3 , @xmath173 can be removed altogether by a change of variable @xmath182 , @xmath183 without interfering with @xmath172 .
thus we can assume additionally that @xmath184 .
we now find from ( [ criterion ] ) that @xmath185 .
hence @xmath186 , so @xmath174 is even .
it follows that @xmath174 can be set to zero by a change of variable @xmath175 , @xmath187 , leaving @xmath172 and @xmath173 untouched .
equations ( [ crit ] ) now tell us that @xmath188 , while the equation ( [ weier ] ) for @xmath162 tells us that @xmath189 . more abstractly , if we define a curve @xmath190 by taking @xmath191 and identifying the sections @xmath192 $ ] and @xmath193 $ ] , so as to make every geometric fiber nodal , then the parametrization @xmath194 given by @xmath195 \mapsto [ st(s - t ) : s(s - t)^2 : t^3]$ ] identifies @xmath162 with @xmath156 .
[ [ outline - of - method - and - algebraic - results . ] ] outline of method and algebraic results .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + this article is long partly because it contains rather more than a single proof of theorem [ mainth ] , and partly because working over @xmath1 presents significant technicalities beyond those that would be present if one worked over fields ( or in some cases , of fields in which 6 is invertible ) .
part i a large chunk is purely algebraic ; it refines and elaborates the method of @xcite .
the basic point is that for any weierstrass curve @xmath162 , one has a 2-object subcategory @xmath196 of @xmath197the dg category of perfect complexes of coherent sheaves with objects @xmath52 ( the structure sheaf ) and @xmath198 ( the skyscraper sheaf at the point at infinity ) , and this subcategory split - generates @xmath197 . the cohomology category
@xmath199 is independent of @xmath162 , but the dg structure of @xmath196 knows @xmath162 .
one can transfer the dg structure to a minimal @xmath13-structure on @xmath200 .
this procedure defines a functor from the category of weierstrass curves to the category of minimal @xmath13-structures on @xmath200 .
we prove in theorem [ ainf comparison ] that this functor is an equivalence .
a slightly coarsened statement of theorem [ ainf comparison ] is as follows : let @xmath201 be an integral domain which is either noetherian and normal of characteristic zero , or an arbitrary field .
let @xmath202 be an @xmath201-linear @xmath13-category together with a calabi yau structure of dimension 1 .
assume that @xmath203 is minimal , has just two objects @xmath204 and @xmath205 , both spherical of dimension 1 and forming an @xmath206-chain ( i.e. @xmath207 ) \cong \hom(b , b)$ ] as graded @xmath201-algebras ; and @xmath208 , @xmath209 $ ] as graded @xmath201-modules ; and @xmath210 ) .
then @xmath203 is trace - preservingly quasi - equivalent to @xmath196 for a unique weierstrass curve @xmath211 , where @xmath196 has the calabi
yau structure arising from its weierstrass differential @xmath212 .
the proof of theorem [ ainf comparison ] invokes the hochschild cohomology @xmath213 .
we computed this cohomology additively in @xcite , but here we give a complete calculation , as a gerstenhaber algebra , by interpreting @xmath213 as the hochschild cohomology @xmath214 of a cuspidal weierstrass curve @xmath215 ( theorem [ hh for cusp curve ] ) . in part ii , we identify the unique curve @xmath216 for which @xmath217 is quasi - isomorphic to the 2-object subcategory @xmath218 of the fukaya category @xmath32 on objects of slopes @xmath179 and @xmath219 , equipped with non - trivial double coverings . in @xcite , we used abouzaid s plumbing model @xcite to prove that @xmath220 is not formal , which implies that @xmath216 is not cuspidal . here
we identify @xmath216 precisely .
in fact , we identify the specialization @xmath221 in three independent ways : ( i ) by eliminating the possibility that @xmath216 is smooth or cuspidal after reduction to an arbitrary prime @xmath155 , by means of the ` closed open string map ' from symplectic cohomology to hochschild cohomology of the fukaya category ; ( ii ) by calculating `` seidel s mirror map '' @xcite , or more precisely , by determining the affine coordinate ring of @xmath221 via a calculation in the exact fukaya category ; and ( iii ) via theta - functions .
the third proof extends to a proof of mirror symmetry for @xmath32 , not just its restriction to @xmath11 .
we use an intrinsic model for the tate curve , and the integral theta - functions for this curve which played a major role in gross s proof @xcite .
the nub is the multiplication rule for these theta - functions and its relation to counts of lattice - points in triangles .
the proof of mirror symmetry for the wrapped category is a rather formal extension of that for the exact category .
we should perhaps make one more remark about exposition .
the authors background is in symplectic topology .
we imagine that typical readers will have an interest in mirror symmetry , perhaps with a bias towards the symplectic , algebro - geometric or physical aspects , but , like us , will not be expert in arithmetic geometry
. we would be delighted to have readers who _ do _ come from an arithmetic geometry background , but ask for their patience in an exposition which we fear belabors what is obvious to them and rushes through what is not .
[ [ higher - dimensions ] ] higher dimensions ?
+ + + + + + + + + + + + + + + + + + we believe that there should be an arithmetic refinement to homological mirror symmetry for calabi
yau manifolds in higher dimensions , but will leave the formulation of such conjectures for elsewhere ; the 2-torus is , we think , far from being an isolated case . the case of 2-tori with several basepoints can be treated rather straightforwardly starting from the one - pointed case , but we shall also leave that for another article . [ [ acknowledgements . ] ] acknowledgements .
+ + + + + + + + + + + + + + + + + yl was partly supported by the herchel smith fund and by marie curie grant eu - fp7 - 268389 ; tp by nsf grant dms 1049313 .
paul seidel provided notes outlining the algebro - geometric approach to computing hochschild cohomology for @xmath200 .
conversations with several mathematicians proved useful as this work evolved , and we would particularly like to thank mohammed abouzaid , david ben - zvi , mirela iperiani , brian conrad , kevin costello , and paul seidel .
we thank the simons center for geometry and physics for its generous hospitality .
our conventions and definitions are those of @xcite ; see @xcite for an informal introduction . for now , we work over a ground _ field _ @xmath222 ( commutative and unital ) , but we shall discuss presently more general ground rings .
all our @xmath13-categories and functors are cohomologically unital .
[ [ triangulated - envelopes . ] ] triangulated envelopes .
+ + + + + + + + + + + + + + + + + + + + + + + any @xmath13-category @xmath223 has a triangulated envelope , a minimal formal enlargement that is a triangulated @xmath13-category , i.e. , every morphism in @xmath223 has a mapping cone in @xmath223 .
the _ twisted complexes _
@xmath224 of an @xmath13-category @xmath223 form a model for the triangulated envelope . the cohomological category @xmath225 is known as the derived category and denoted @xmath226 . [ [ split - closure . ] ] split closure .
+ + + + + + + + + + + + + + one can formally enlarge @xmath227 further to another triangulated @xmath13-category @xmath228 which is additionally _ split - closed _
( also known as idempotent - closed or karoubi complete ) .
idempotent _ in the @xmath13-category @xmath227 is defined to be an @xmath13-functor @xmath229 from the trivial @xmath13-category @xmath222 , which has one object @xmath230 and @xmath231 ( the ground field ) .
for example , if the object @xmath6 is the direct sum of objects @xmath232 and @xmath233 , meaning that @xmath234 in the category of @xmath224-modules , then @xmath232 defines an idempotent @xmath235 in @xmath236 , with @xmath237 and , on morphisms , @xmath238 .
the module @xmath239 is actually intrinsic to the idempotent @xmath235 ( it can be constructed as the ` abstract image of @xmath235 ' ( * ? ? ?
* chapter 1 , ( 4b ) ) ) ; the object @xmath232 _ represents _ the abstract image .
split - closed means that the abstract image of an arbitrary idempotent is represented by an object .
we write @xmath240 for the triangulated category @xmath241 .
it is useful to note that @xmath227 is split - closed as an @xmath13-category if and only if @xmath242 is split - closed as an ordinary @xmath222-linear category .
[ [ thomasons - theorem . ] ] thomason s theorem .
+ + + + + + + + + + + + + + + + + + + by @xcite , a necessary and sufficient condition for an @xmath13-functor which is a quasi - embedding to be a quasi - equivalence is that ( a ) it should induce a quasi - isomorphism after split - closure , and ( b ) that it should induce an isomorphism of grothendieck groups @xmath243 .
thus , clauses ( iii ) and ( iv ) from theorem [ mainth ] , which assert derived equivalence without split - closure , are partly statements about @xmath243 .
[ [ a_infty - categories - over - rings . ] ] @xmath13-categories over rings .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + our fukaya categories will be @xmath13-categories over unital commutative rings @xmath244 .
the usual definition of an @xmath13-category @xmath223 makes sense over such rings : the morphism spaces are arbitrary graded @xmath244-modules .
let s call such an object a _
naive @xmath13-category_. the basic notions carry through .
for instance , the twisted complexes @xmath224 , defined as usual ( the multiplicity spaces are finite - rank free modules ) , form a triangulated envelope for @xmath223 , as in @xcite . however , some of the naive constructions do not have the homotopical significance one might wish for .
an example is that the hochschild cohomology @xmath213 of a @xmath244-algebra @xmath200 , defined through the bar complex , does not compute the bimodule - ext module @xmath245 , but rather , relative @xmath246 for the map of @xmath244-algebras @xmath247 @xcite . over fields ,
@xmath13-constructions are automatically ` derived ' . to retain this property
, we define a _ projective _ @xmath13-category to be a naive @xmath13-category in which the morphism spaces are projective graded @xmath244-modules .
fukaya categories are projective because the hom - spaces come with finite bases .
projective graded modules satisfy @xmath248 and @xmath249 .
the naive definitions of @xmath13-functors and their natural transformations , and of hochschild homology and cohomology , work well for projective @xmath13-categories .
[ [ dg - categories - over - rings . ] ] dg categories over rings .
+ + + + + + + + + + + + + + + + + + + + + + + + + differential graded ( dg ) categories over commutative rings have been well studied @xcite , and the theory does not depend on such ad hoc arrangements as having projective hom - spaces .
there is a self - contained theory in which derived categories are defined via localization , not via twisted complexes .
[ [ calabiyau - structures . ] ] calabi yau structures .
+ + + + + + + + + + + + + + + + + + + + + + when @xmath222 is a field , a _ serre functor _ of a @xmath222-linear category @xmath223 with finite - dimensional hom - spaces is an equivalence @xmath250 , together with isomorphisms @xmath251 natural in both inputs , such that @xmath252 as maps @xmath253 @xcite .
a serre functor @xmath254 for the ( bounded ) derived category @xmath255 of a smooth projective variety @xmath6 over a field @xmath222 is given by @xmath256 $ ] .
the maps @xmath257 are serre duality isomorphisms .
a _ calabi yau ( cy ) structure _ of dimension @xmath258 on @xmath223 is a serre functor @xmath259 in which @xmath260 is the shift functor @xmath261 $ ] . if @xmath6 is a smooth projective calabi
yau variety , equipped with an @xmath258-form @xmath262 trivializing @xmath263 , then its derived category has a cy structure induced by the isomorphism @xmath264 .
the role of @xmath262 is to normalize the cy structure .
the cohomological ( not derived ) fukaya category @xmath265 of a ( compact or exact ) symplectic manifold @xmath266 , with coefficients in a field @xmath222 , comes with a natural calabi yau structure : @xmath267 is the floer - theoretic poincar duality isomorphism @xmath268 $ ] .
it is a subtler matter to obtain a calabi yau structure on the _ derived _ fukaya category @xmath269 .
it is expected that such a structure does exist , and is canonical , and arises from a cyclic symmetry defined on the @xmath13-level ( see @xcite for a construction of such a cyclic symmetry over @xmath270 , and @xcite for an account of the relevant homological algebra ) , but this more refined structure will play no role in our considerations .
a cy structure gives a canonical ` trace map ' @xmath271 . from the trace maps
, one can reconstruct all the maps @xmath257 . in this article
we think of cy structures in terms of their trace maps ; a functor preserving cy structures will be called _ trace - preserving_. we shall need to say what we mean by a cy structure for a category over a commutative ring @xmath244 .
the categories in question are of form @xmath272 , where @xmath223 is an @xmath13-category , and this permits us to make an expedient ( but not fully satisfactory ) definition : a _ cy structure _ consists on the @xmath244-linear @xmath13-category @xmath223 consists of cochain - level maps @xmath273)^\vee\ ] ] such that the induced maps on cohomology @xmath274 \colon \hom_{h^0(\euc\times_{\mathbb{l } } \mathbb{f})}(a , b ) \simeq \hom_{h^0(\euc\times_{\mathbb{l } } \mathbb{f})}(b , a[n])^\vee\ ] ] form a cy structure for each residue field @xmath275 .
( note that since the hom - space in @xmath223 are projective modules , they are also flat , so tensoring them with @xmath276 commutes with @xmath277 . )
if @xmath223 and @xmath278 have cy structures , an @xmath13-functor @xmath279 is called _ trace - preserving _ if the induced functors @xmath280 are all trace - preserving . with this definition ,
fukaya categories have cy structures over arbitrary rings @xmath244 , since poincar duality is defined at cochain level but our demands on the maps are all at cohomology - level . [
[ perfect - complexes . ] ] perfect complexes .
+ + + + + + + + + + + + + + + + + + let @xmath6 be a scheme .
a _ strictly perfect complex _ is a bounded complex of locally free , finite rank @xmath281-modules .
perfect complex _ is a cohomologically bounded complex @xmath282 of coherent sheaves of @xmath281-modules which is locally quasi - isomorphic to a strictly perfect complex . inside the bounded derived category of coherent sheaves
@xmath283 , one has a full triangulated subcategory @xmath284 of perfect complexes .
we will need to consider dg enhancements of @xmath285 ; that is , we want a pre - triangulated dg category @xmath223 and an equivalence of triangulated categories @xmath286 .
when @xmath6 is a projective scheme over a field @xmath222 , @xmath285 has a dg enhancement @xmath287 which is unique : if @xmath288 is another then there is a _ quasi - functor _ @xmath289 such that @xmath290 @xcite . since we wish to work over more general base rings , and for computational purposes , we specify a dg enhancement of @xmath285 , valid for @xmath6 a projective noetherian scheme , as follows .
assume @xmath6 is separated and noetherian .
fix an affine open covering @xmath291 of @xmath6 .
define a dg category @xmath292 whose objects are locally free sheaves ( = vector bundles ) of finite rank , and whose hom - spaces , denoted @xmath293 , are ech complexes : @xmath294 with @xmath295 the ech differential .
the cohomology of the ech complex is @xmath296 by ( * ? ? ? * theorem iii.4.5 ) and the fact that @xmath297 .
composition combines the shuffle product of ech cochains with the composition of sheaf - morphisms . whilst @xmath292 depends on the open covering ,
different choices lead to quasi - isomorphic dg categories ( take the union of the two coverings ) .
we now pass to the pre - triangulated dg category @xmath298 of twisted complexes .
there is an embedding @xmath299 , mapping a twisted complex to its total complex .
this embedding is a quasi - equivalence , because every perfect complex is quasi - isomorphic to a strictly perfect complex ( * ? ? ? * prop .
2.3.1(d ) ) .
another approach to dg enhancement is to use injective resolutions ; the equivalence of the injective and ech approaches is shown in ( * ? ? ?
* lemma 5.1 ) , over fields ; the proof remains valid over rings .
grothendieck serre duality defines a cy structure for @xmath298 when @xmath6 is equipped with a trivialization of the relative dualizing sheaf @xmath300 . in theorem
[ mainth ] , the functor @xmath49 is trace - preserving
. we shall need to work with curves over the rings @xmath1 and @xmath21 , and to this end we note some terminology for curves over schemes ( cf . for example @xcite ) .
a _ curve _ over a noetherian scheme @xmath260 is a morphism of schemes @xmath301 that is separated , flat and finitely presented , such that for every closed point @xmath302 the fiber @xmath303 is non - empty of pure dimension 1 .
the euler characteristic @xmath304 is then locally constant ; when it is constant and equal to @xmath305 , and the geometric fibers are connected , we say that @xmath162 has arithmetic genus @xmath306 .
we shall always apply the restrictions that curves are to be proper , and that the fibres @xmath303 are cohen macaulay .
this implies that one has a dualizing sheaf @xmath307 , and where @xmath308 is regular it coincides with the sheaf of differentials @xmath309 . a reminder on duality @xcite
: there is an intrinsic residue isomorphism of sheaves on @xmath260 @xmath310 with the yoneda ( composition ) product @xmath311 , this defines the serre duality pairing , @xmath312 for any coherent sheaf @xmath313 .
a curve has arithmetic genus one if and only if @xmath314 , i.e. , if and only if @xmath315 is a dualizing sheaf . if @xmath316 is an isomorphism then it composes with the residue map to give an isomorphism @xmath317 and a serre duality pairing @xmath318 which induces a perfect pairing on stalks at any closed point @xmath319 .
an _ abstract weierstrass curve _ @xmath320 over @xmath260 is a curve @xmath308 of arithmetic genus one , such that each geometric fiber @xmath303 is irreducible , equipped with a section @xmath321 of @xmath235 and a specific isomorphism @xmath316 .
isomorphism _ of abstract weierstrass curves @xmath322 and @xmath323 over @xmath260 is an isomorphism @xmath324 of @xmath260-schemes such that @xmath325 , and such that the map @xmath326 coincides with @xmath327 under the identifications @xmath328 and @xmath329 induced by @xmath330 . an _ embedded weierstrass curve _ over @xmath331 is a curve @xmath332 embedded as a cubic @xmath333 such a curve comes with its point at infinity @xmath334 $ ] , which defines a section @xmath335 of @xmath308 .
it also comes with a standard differential @xmath26 , possibly with poles at the singular points : writing the cubic equation as @xmath336 , one has @xmath337 at points where @xmath338 , and @xmath339 at points where @xmath340 .
assume that @xmath201 is a _
normal _ ring ( i.e. , @xmath201 is reduced and integrally closed in its total quotient ring ) .
then @xmath26 defines a section of the dualizing sheaf @xmath341 .
let @xmath342 = r[a_1,a_2,a_3,a_4,a_6]$]another normal ring .
it will suffice to prove the assertion for the ` universal weierstrass curve ' @xmath343 $ ] defined by ( [ weierstrass equation ] ) , since the formation of the dualizing sheaf is compatible with the specialization to particular values of the @xmath344 .
the scheme @xmath223 is normal : in the open set @xmath345 where ( [ weierstrass equation ] ) is valid , @xmath122 is a function of the other variables , so projection @xmath346 $ ] is an isomorphism , and @xmath347 $ ] is normal . along the section at infinity @xmath56
$ ] , the fibers of @xmath155 are regular , and the base normal , so the total space is normal . the relative dualizing sheaf @xmath348}$ ] is an invertible sheaf , since all its fibers are gorenstein ( being local complete intersections ) .
the locus where the fibers of @xmath155 are singular is defined by @xmath349 .
this locus has codimension 2 in @xmath223 : it maps to the codimension 1 locus @xmath350 $ ] defined by the vanishing of the discriminant , and it has codimension 1 in each fiber . since @xmath26 is a section of @xmath351}$ ] defined outside a codimension 2 subset of a normal scheme , it extends to a global section , by the algebraic counterpart to hartogs s theorem . of a field @xmath352 is the intersection of the valuation rings in @xmath352 which contain @xmath200 @xcite . ]
thus an embedded weierstrass curve functorially defines an abstract weierstrass curve @xmath353 . by riemann
roch , every abstract weierstrass curve is isomorphic to an embedded one . to specify the embedding into @xmath354
, one must give a basis of @xmath355 of the form @xmath356 , where @xmath357 is the regular function with value 1 , and @xmath358 .
the denominator - free form of the argument is given at @xcite , for instance .
the algebraic group @xmath359 of elements which preserve weierstrass form consists of matrices ( up to scale ) of the shape @xmath360,\quad u \in \mathbb{g}_m.\ ] ] we shall call @xmath361 the _ reparametrization group _ for embedded weierstrass curves .
it acts on embedded weierstrass curves via the substitutions @xmath362 the effects of a substitution on the weierstrass coefficients are listed in @xcite or @xcite : @xmath363 the unipotent subgroup @xmath364 of elements where @xmath365 is the subgroup which preserves the differential @xmath26 . thus if @xmath366 then @xmath367 is an isomorphism of abstract weierstrass curves .
[ [ the - lie - algebra . ] ] the lie algebra .
+ + + + + + + + + + + + + + + + the lie algebra @xmath368 of @xmath361 is spanned by four vectors : @xmath369,\ ; \partial_r:= \left [ \begin{array } { ccc } 0 & 0 & 0 \\ 0 & 0 & 1 \\ 0 & 0 & 0 \end{array}\right],\ ; \partial_t:= \left [ \begin{array } { ccc } 0 & 0 & 1 \\ 0 & 0 & 0 \\ 0 & 0 & 0 \end{array}\right],\ ] ] ( these three span the lie algebra @xmath370 of @xmath345 ) and @xmath371.\ ] ] the derivative of the @xmath361-action on @xmath372\ ] ] is an action of @xmath368 on @xmath373 by a lie algebra homomorphism @xmath374 which we think of as a map @xmath375 .
the partial derivative @xmath376 makes @xmath373 a @xmath377-module .
we can form a differential graded lie algebra ( dgla ) concentrated in degrees @xmath179 and @xmath357 , @xmath378 whose bracket combines the lie bracket of @xmath368 with the module structure of @xmath373 .
thus @xmath379 captures the truncation of @xmath380 where we only work in a first - order neighborhood of @xmath381 .
there are @xmath382-actions on @xmath368 and on @xmath373 , intertwined by @xmath383 .
the action on @xmath373 is given by @xmath384 ; that on @xmath368 by @xmath385 , @xmath386 , @xmath387 , @xmath388 .
thus @xmath373 and @xmath368 are graded @xmath222-modules . explicitly , taking @xmath389 as basis for @xmath368 , and @xmath390 as coordinates for @xmath373
, one has @xmath391\ ] ] and @xmath392 } \oplus \stackrel{a_2}{\frac{\bbk}{(3 ) } [ 2 ] } \oplus \stackrel{a_3}{\frac{\bbk}{(2 ) } [ 3 ] } \oplus\stackrel{a_4}{\bbk [ 4 ] } \oplus \stackrel{a_6}{\bbk [ 6 ] } , \\ \label{ker d } \ker d & = \stackrel{\partial_u}{\bbk } \ ; \oplus \ ; \stackrel{\partial_s}{\frac{\bbk}{(2)}[1 ] } \ ; \oplus \ ; \stackrel{\partial_r}{\frac{\bbk}{(3)}[2 ] } \ ; \oplus \ ; \stackrel{\partial_t}{\frac{\bbk}{(2)}[3]}. \end{aligned}\ ] ] the cuspidal weierstrass curve @xmath393 will play a special role in our story , stemming from the fact that the full subcategory of its derived category whose objects are the structure sheaf and the skyscraper at infinity is _ formal_. let @xmath394 , and let @xmath395 be the @xmath222-point @xmath192 $ ] .
let @xmath396 denote the standard affine coordinate @xmath397 , @xmath398 $ ] .
one has the structure sheaf @xmath281 , and inside it the sheaf @xmath399 of functions @xmath330 such that @xmath400 ( that is , in terms of the local coordinate @xmath9 , functions @xmath401 ) .
let @xmath402 denote the scheme @xmath403 .
[ abstract cusp ] the abstract weierstrass curve underlying @xmath215 is isomorphic to @xmath404 the normalization of @xmath215 is a non - singular rational curve @xmath405 with a distinguished point @xmath406 which maps to the cusp under the normalization map @xmath407 .
we fix an isomorphism @xmath408 mapping @xmath155 to @xmath406 .
the map @xmath409 is a homeomorphism in the zariski topology , and so defines a scheme - theoretic isomorphism @xmath410 .
one has @xmath411 : the local model near the cusp is the map of @xmath222-algebras @xmath412/(y^2-x^3 ) \to \bbk[z]$ ] given by @xmath413 and @xmath414 , whose image is @xmath415 $ ] .
the @xmath416-module of differentials @xmath417 is given by the submodule of @xmath418 ( meromorphic differentials on @xmath6 with a double pole at @xmath155 ) formed by the differentials with vanishing residue at @xmath155 . in terms of the affine
coordinate @xmath9 near @xmath155 , the differential of a function @xmath419 is @xmath420 .
the weierstrass differential @xmath26 is given by @xmath421 ( in characteristics 2 and 3 only one of these expressions makes sense ) . in terms of @xmath9 ,
one has @xmath422 ; this makes global sense because @xmath423 .
in this subsection we explain how to pass from a weierstrass curve @xmath211 to a two - object dg category @xmath196 with standard cohomology .
consider a genus - one curve @xmath162 over a noetherian affine scheme @xmath260 .
it has a dg category @xmath424 , defined via an affine open covering , linear over the ring @xmath425 .
the dg category @xmath426 for an abstract weierstrass curve @xmath353 over @xmath427 has extra structure in the form of a trace pairing @xmath428 , as described in the introduction .
it also has distinguished split - generators , namely , the structure sheaf and the skyscraper @xmath429 at @xmath335 ( more properly , its locally - free resolution @xmath430 ) : [ gen perf ] for a weierstrass curve @xmath211 over a noetherian affine scheme , one has @xmath431 here @xmath432 denotes the smallest dg subcategory of @xmath426 closed under quasi - isomorphisms , shifts , mapping cones and passing to idempotents .
we claim first that , if @xmath433 is a very ample line bundle on @xmath162 , then @xmath426 is split - generated ( i.e. , generated under quasi - isomorphisms , shifts , mapping cones and passing to idempotents ) by the twists @xmath434 .
the argument is as in ( * ? ? ?
* lemma 5.4 ) , which seidel attributes to kontsevich .
take a locally free sheaf @xmath435 on @xmath162 . by serre s theorem that very ample implies ample ( * ? ? ?
* theorem ii.5.17 ) , which is valid for noetherian projective schemes , one can find an epimorphism @xmath436 for some @xmath437 .
iteratively , one can find for each @xmath438 a left resolution @xmath439 there results an exact triangle in @xmath440 @xmath441 \stackrel{+}{\to } .\ ] ] now , @xmath442 , and if we take @xmath443 , this ext - module must vanish .
consequently , the exact triangle splits and defines a quasi - isomorphism @xmath444 .\ ] ] thus @xmath435 is a direct summand in the object on the lhs .
note also that @xmath445 is a perfect complex , because it is the mapping cone of a map of perfect complexes ; therefore , @xmath445 is quasi - isomorphic to a strictly perfect complex .
this proves that every locally free sheaf lies in the split - closure of the collection @xmath434 .
it follows that the same is true of every object of @xmath426 .
note next that @xmath446 includes @xmath447 for each @xmath448 , by a straightforward induction .
but @xmath449 is a very ample line bundle , so now the claim completes the proof .
[ [ a - two - object - subcategory . ] ] a two - object subcategory .
+ + + + + + + + + + + + + + + + + + + + + + + + + let @xmath450 denote the full dg subcategory of @xmath426 with the two objects @xmath451 and @xmath452 and with the trace map @xmath453 .
it is defined up to quasi - isomorphisms acting trivially on cohomology . to be precise
, we shall define @xmath196 using the ech complexes associated with an affine open covering @xmath291 .
if we pick two coverings @xmath454 and @xmath455 , we get dg categories @xmath456 and @xmath457 , and a zigzag of quasi - isomorphisms @xmath458 the cohomology category @xmath459 is truly canonical defined up to canonical isomorphism .
an isomorphism @xmath460 of abstract weierstrass curves is a homeomorphism @xmath306 , together with a local isomorphism of sheaves of @xmath260-modules @xmath461 , respecting the sections and differentials .
that means , first , that @xmath462 is the composite @xmath463 ; this implies a canonical isomorphism @xmath464 .
the isomorphism @xmath306 induces isomorphisms between the abelian categories of coherent sheaves on @xmath465 and @xmath466 , preserving the objects @xmath52 and @xmath467 .
this naturally extends to an isomorphism of dg categories @xmath468 , provided that we use an open covering @xmath291 for @xmath465 and @xmath469 for @xmath466 .
thus , if we have @xmath470 and @xmath471 with composite @xmath472 , then the composite isomorphism @xmath473 coincides with @xmath474 , provided again that we use the coverings @xmath291 , @xmath475 and @xmath476 . if we do not , then we get instead the formal composite of chains of quasi - isomorphisms : @xmath477\ar[d ] & & \eub_{g_{23}(\euu_2 ) \cup \euu_3}\ar[d]\ar[dr ] \\
\eub_{\euu_1 } \ar[r ] & \eub_{g_{12}(\euu_1 ) } & \eub_{\euu_2 } \ar[r ] & \eub_{g_{23}(\euu_2 ) } & \eub_{\euu_3 } } \ ] ] [ [ the - cohomology - category . ] ] the cohomology category .
+ + + + + + + + + + + + + + + + + + + + + + + + we shall be interested in the map which assigns to each weierstrass curve @xmath478 a graded - linear cohomology category @xmath479 and a dg category - with - trace , @xmath480 with an isomorphism @xmath481 , defined up to quasi - isomorphisms acting trivially on @xmath479 .
the category @xmath479 is independent of the abstract weierstrass curve .
precisely : there is an @xmath425-linear graded category @xmath200 with two objects @xmath52 and @xmath467 , equipped with a trace map @xmath428 , such that the following holds : for any abstract weierstrass curve @xmath482 the cohomology category @xmath483 is trace - preservingly isomorphic to @xmath200 in such a way that if @xmath484 is any isomorphism of weierstrass curves then the resulting map @xmath485 is the identity . in other words ,
the category of ext - modules between @xmath315 and @xmath486 is independent of @xmath353 as a graded @xmath222-linear category with trace . to prove the proposition
, we examine the structure of @xmath459 . writing @xmath487 and @xmath488
, one has canonical isomorphisms @xmath489 thus both endomorphism spaces are 2-dimensional and sit in degrees @xmath179 and @xmath357 .
the trace isomorphisms @xmath490 are , in the first case , the one that we have discussed ( the composite of @xmath26 and the residue pairing ) and in the second case the pullback by @xmath335 of the composite @xmath491 where @xmath492 is the map defined by evaluating @xmath26viewed as a differential on tangent vectors at @xmath335 .
one has @xmath493 finally , one has isomorphisms @xmath494 we now describe the category @xmath200 demanded by the proposition above .
let @xmath495 and @xmath496 .
we have seen how to use @xmath26 to obtain algebra isomorphisms @xmath497 ) \cong \end(y)\ ] ] such that the trace maps correspond to the identity map of @xmath425 .
we also have exhibited isomorphisms @xmath498 and @xmath499 $ ] .
the composition maps are mostly dictated by the requirements of grading and unitality .
the interesting ones are @xmath500 these are both given by the multiplication of functions @xmath501 the objects @xmath6 and @xmath502 and their morphisms form a graded - linear cy category @xmath503 , independent of @xmath162 . [ [ differential - graded - structure . ] ] differential graded structure .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + while the cohomology category @xmath504 is independent of @xmath162 , the dg structure of @xmath196 is fully sensitive to the curve @xmath162 : [ dg comparison ] work over a field @xmath222 . 1 .
let @xmath203 be a dg category with trace such that @xmath505 .
then there exist an abstract weierstrass curve @xmath478 and a trace - preserving @xmath13-quasi - isomorphism @xmath506 .
if @xmath478 and @xmath507 give rise to quasi - isomorphic dg categories with trace , i.e. , @xmath196 is related to @xmath508 by a zig - zag of trace - preserving isomorphisms , then @xmath509 .
we state this result now so as to indicate our aims .
however , we will establish it as a corollary of a more detailed statement , theorem [ ainf comparison ] , and it will in fact be the latter result which we use , not theorem [ dg comparison ] .
the proofs will be given later , but we offer two hints . for the uniqueness clause ,
the point is that there is a construction which assigns to any such category @xmath203 a sequence @xmath510 of twisted complexes in a uniform manner .
when @xmath511 , one has @xmath512 . one further constructs multiplication maps @xmath513
when @xmath511 , these reproduce the multiplication @xmath514 .
thus the coordinate ring of the affine curve @xmath515 , the open complement of @xmath516 , is determined by @xmath196 . the existence clause ( 1 )
is plausible because one has @xmath517 and @xmath518 . as a result
, @xmath519 has no poisson deformations and no non - commutative deformations , and it is reasonable to expect all deformations of @xmath520 to be geometric .
the cuspidal cubic @xmath521 has the following special property , which already appeared in @xcite : [ formality ] the dga @xmath522 is formal .
we may transfer the dg structure of @xmath523 to a minimal @xmath13-structure on @xmath524 .
the transfer of dg structure will be described in detail in the proof of lemma [ splittings ] .
the goal , then , is to prove that the @xmath13 structure maps @xmath525 vanish for @xmath526 .
@xmath215 is the curve @xmath527 .
the multiplicative group @xmath528 acts on @xmath215 by @xmath529 , preserving the point @xmath56 $ ] , and therefore acts on @xmath523 .
the action of @xmath528 on @xmath523 induces an action on the cohomology @xmath200 , and the transfer will be set up equivariantly so that the resulting @xmath13-structure has the property that @xmath530 .
a short computation leads to the following conclusion : the weight of the @xmath528-action on a hom - space @xmath531 ( where @xmath6 is @xmath52 or @xmath467 , ditto @xmath532 ) is equal to the degree @xmath438 .
now take @xmath533 a sequence of objects ( @xmath52 or @xmath467 ) , and take @xmath534 . for
the equation @xmath530 to hold , one must have @xmath535 i.e. , @xmath536 . when we come to prove theorem [ mainth ] , clause ( iii ) , we will need to apply thomason s theorem about grothendieck groups @xcite , and for that we shall need to know @xmath537 .
we think of @xmath156 , the central fiber of the tate curve , as the curve over @xmath3 obtained from @xmath538 by identifying @xmath539 $ ] and @xmath540 $ ] . by definition
, @xmath537 is the grothendieck group of the abelian category of vector bundles ( locally free sheaves of finite rank ) on @xmath156 .
it can also be thought of as @xmath541 , the grothendieck group of the triangulated category of perfect complexes.=[a]+[c]$ ] for each distinguished triangle @xmath542 $ ] . ]
the proof of the following lemma is more substantial than one might expect . as partial justification , we point out that @xmath537 is an _ absolute _ invariant of the scheme @xmath156it is not defined ` relative to @xmath543'and that @xmath156 is _
2-dimensional _ as a scheme .
[ k theory ] the map @xmath544 is an isomorphism .
thus a vector bundle on @xmath156 with trivial determinant is stably trivial .
let @xmath545 denote the grothendieck group of finitely - generated projective modules over the commutative ring @xmath201 .
we also have the reduced group @xmath546 and the group of ` stable endomorphisms ' @xmath547 . for the following standard results in k - theory we refer to the text @xcite ( see in particular the ` fundamental theorem for @xmath548 ' ( 3.6 ) ) .
we have @xmath549 since @xmath1 is a pid .
the units @xmath550 are always a subgroup of @xmath547 , and one has @xmath551 . since @xmath1 is a regular ring , the inclusion - induced maps @xmath552)$ ] and @xmath553)$ ] are isomorphisms .
one has a split injection @xmath554 ) \to k_1(\z[t , t^{-1}])$ ] , induced by the natural map @xmath555\to \z[t , t^{-1}]$ ] , whose cokernel is @xmath556 . hence @xmath557)\cong \z \oplus \z^\times$ ] .
there is a group @xmath558}{(t)})$ ] of complexes of f.g .
projective @xmath555$]-modules whose cohomology is bounded and supported on the ideal @xmath559 ; and an exact sequence @xmath560 ) \xrightarrow { } k_1(\z[t , t^{-1 } ] ) \xrightarrow{\partial } k_0(\on{\z[t]}{(t ) } ) \xrightarrow { } \widehat{k}_0(\z[t])\ ] ] ( * ? ? ?
* ( 5.1 ) ) . from our discussion
, we see that the first map has cokernel @xmath1 and that @xmath561)=0 $ ] , whence @xmath558}{(t ) } ) \cong \z$ ] .
now consider the ` node ' @xmath562/(xy)$ ] , and the normalization map @xmath563 \oplus \z[t]$ ] , namely , @xmath564 .
one has @xmath565
\oplus \bbk[t])=0 $ ] .
the kernel of @xmath566 \oplus \z[t])$ ] is also zero , because a f.g .
projective @xmath201-module @xmath83 is determined by @xmath567which is given by a pair of f.g .
projective @xmath555$]-modules @xmath568 and @xmath569and ` descent data ' , an isomorphism @xmath570 . by stabilizing
, we may assume that @xmath568 and @xmath569 are free , and choosing appropriate bases we can make @xmath29 the identity matrix . hence @xmath571 .
we have ( from ( * ? ? ?
* ( 5.1 ) ) again ) a commutative diagram with exact rows @xmath572\ar^{\nu_*}[d ] & k_1(r[x^{-1},y^{-1 } ] ) \ar[r]\ar_{\cong}^{\nu_*}[d ] & k_0(\on{r}{(x , y ) } ) \ar[r]\ar^{\nu_*}[d ] & 0 \\ 0\ar[r ] & \bigoplus^2{k_1(\z[t ] ) } \ar[r ] & \bigoplus^2{k_1(\z[t , t^{-1}])}\ar[r ] & \bigoplus^2{k_0(\on{\z[t]}{(t ) } ) } \ar[r ] & 0 } \ ] ] the middle vertical arrow is an isomorphism , since it is induced by a ring isomorphism .
note that @xmath547 contains the units @xmath573 .
chasing the diagram , we see that @xmath574 , where the summand @xmath575 mapped isomorphically by @xmath576 to @xmath577}{(t)})}$ ] , and @xmath260 is either @xmath573 or 0 .
we now switch from rings to schemes , referring to @xcite for foundational matters .
if @xmath165 is a closed subscheme of @xmath6 , @xmath578 is the grothendieck group of the triangulated category of perfect complexes on @xmath156 whose cohomology sheaves are coherent and supported on @xmath165 .
the map @xmath579 is surjective apply it to a line bundle .
we must prove its injectivity .
let @xmath580 be the normalization map .
let @xmath165 be the closure of the image of the nodal section of @xmath156 , and @xmath581 the inclusion of the open complement of @xmath165 .
there is a commutative diagram with exact rows @xmath582\ar^{=}_{\nu^*}[d ] & k_0(\on{\eut_0}{z } ) \ar^{\mu}[r ] \ar_{\nu^*}[d ] & \wh{k}_0(\eut_0 ) \ar^{j^*}[r]\ar_{\nu^*}[d ] & \wh{k}_0(u ) \ar^{=}_{\nu^*}[d ] , \\
k_1(u ) \ar^{\partial'}[r ] & k_0(\on{\mathbb{p}^1}{\overline{\ { p , q \ } } } ) \ar^{\mu'}[r ] & \wh{k}_0(\mathbb{p}^1 ) \ar[r ] & \wh{k}_0(u ) } .\ ] ] the groups @xmath578 have an excision property ( * ? ? ? * ( 3.19 ) ) , which tells us for instance that restriction induces an isomorphism @xmath583 , where @xmath435 is any open neighborhood of @xmath165 .
further , it tells us that @xmath584 , where @xmath585 is the completion of @xmath435 along @xmath165 , and @xmath586 the image of @xmath165 .
similarly , @xmath587}{(t)})}.\ ] ] now , @xmath585 is isomorphic to @xmath588 , and by naturality of normalization , the map @xmath589 corresponds to the normalization map @xmath590 , which is itself the completion of @xmath591/(xy ) \to \spec \bigoplus^2{\z[t]}$ ] .
the upshot is that the map @xmath592 in ( [ second k diagram ] ) can be identified with the arrow @xmath593}{(t ) } ) } $ ] in ( [ first k diagram ] ) . hence @xmath594
= s$ ] . now take @xmath595 with @xmath596 .
then @xmath597 , and @xmath598 ( * ? ? ?
* ( 4.1 ) ) . hence @xmath599 .
thus @xmath89 is the image @xmath600 of some @xmath601 , and @xmath602 maps to zero in @xmath603 .
so @xmath604 for some @xmath605
. then @xmath606 , and so @xmath607 .
let @xmath608 be the line - bundle obtained from the trivial line bundle on @xmath538 by gluing the fibers over the @xmath192 $ ] and @xmath609 $ ] by the map @xmath610 .
let @xmath611 - 1 \in \wh{k}_0(\eut_0)$ ] .
then @xmath612 , so @xmath613 . we have @xmath614 , so @xmath615 , say . also @xmath616 ; hence @xmath617 , and so @xmath618 , say , and @xmath619 .
if @xmath260 were zero , we would have @xmath620 .
hence @xmath260 is not zero , which we have seen implies that @xmath621 , and therefore @xmath622 .
so our class @xmath89 from the previous paragraph must be zero .
in the previous section we set up a two - object , graded linear category @xmath200 , the cohomology @xmath623 of a dg category associated with an arbitrary weierstrass curve . we can view ( the sum of direct sum of the hom - spaces in ) @xmath200 as a @xmath222-algebra . in this section
we compute its hochschild cohomology @xmath624 as a bigraded algebra ; or more precisely , the truncated version @xmath625 relevant to non - curved @xmath13 deformations @xcite . later ( theorem [ bracket ] ) we shall complete the picture by determining the gerstenhaber bracket . our main result is as follows : [ hh for cusp curve ] let @xmath222 be any field .
introduce the commutative graded @xmath222-algebra @xmath626/(y^2-x^3 , 2y , -3x^2),\ ] ] concentrated in degree zero .
make it a bigraded algebra by assigning the following internal degrees @xmath70 to the generators : @xmath627 introduce also the free graded - commutative graded algebra @xmath628 = \bbk[\beta]\otimes \lambda[\gamma ] , \quad \deg\beta=2,\quad\deg \gamma=1,\ ] ] made a bigraded algebra by assigning internal degrees @xmath629 define the bigraded algebra @xmath630 .
let @xmath631 denote the subalgebra of @xmath632 spanned by those classes with @xmath633 . then there is a canonical isomorphism of bigraded algebras @xmath634 ( the bihomogeneous classes of @xmath632 with @xmath635i.e .
, the bihomogeneous classes omitted in @xmath636are spanned by @xmath637 and @xmath638 when @xmath639 ; by @xmath637 , @xmath640 and @xmath641 when @xmath642 ; and by @xmath637 , @xmath643 , @xmath644 and @xmath645 when @xmath646 . ) by definition , @xmath647)$ ] , the @xmath648th derived homomorphism @xmath649 $ ] in the category of graded @xmath650-bimodules .
composition of bimodule - homomorphisms yields , on the derived level , an associative product making @xmath624 a graded @xmath222-algebra with an additional internal grading @xmath70 .
the product is graded - commutative with respect to the cohomological grading @xmath651 @xcite . the truncation @xmath625 , in which @xmath70 is required to be non - positive , is a subalgebra .
theorem [ hh for cusp curve ] does _ not _ make any claims about the untruncated hochschild algebra .
the rank of @xmath652 was computed by the authors in @xcite by a different method .
another approach has been found , in characteristic @xmath179 , by fisette @xcite , who also makes the link between @xmath13-structures and elliptic curves . although the additive result is the only part that is essential , we choose to present here this more complete calculation , proved via algebraic geometry on a cuspidal cubic curve @xmath215 , by a method explained to us by paul seidel .
let @xmath215 be a cuspidal weierstrass curve over the field @xmath222
. then @xmath653 .
this lemma is restated as part of prop .
[ hh for weierstrass ] and proved there .
one can compute @xmath654 using sheaf theory ; the heart of the calculation is that of @xmath655 , where @xmath656/(y^2-x^3)$ ] .
the latter calculation can be done as for any complete intersection singularity by using a koszul resolution to replace @xmath201 by a smooth affine dg manifold , for which a version of the hochschild kostant rosenberg theorem is available .
one virtue of the method is that it adapts easily to yield a computation of the hochschild cohomology of @xmath200 with a non - trivial @xmath13-structure arising from a weierstrass curve .
we shall carry out the computation for the case of a nodal curve , which we will be able to identify with the symplectic cohomology @xmath657 for the punctured torus .
_ note _ : the ideas that we review in ( [ hh for var ] ) have undergone a lengthy evolution , which we have not attempted to trace in detail .
our citations do not necessarily reflect priority .
let @xmath6 be a quasi - projective scheme of dimension @xmath383 over a field @xmath222 .
its hochschild cohomology @xmath658 is defined as @xmath659 where @xmath660 is the diagonal map @xcite .
when @xmath6 is smooth , there is a natural morphism of complexes of sheaves @xmath661\ ] ] ( trivial differential in the complex on the right ) , the hochschild kostant rosenberg ( hkr ) map ; see @xcite or the prcis in @xcite .
when @xmath662 is invertible in @xmath222which means that @xmath663 is either @xmath179 or @xmath664the hkr map is a quasi - isomorphism .
, this map may not be a quasi - isomorphism but it is nevertheless true ( ( * ? ? ?
9.1.3 ) , ) that @xmath665 when @xmath666 .
] from the hkr map and the adjunction @xmath667 , one obtains , under this assumption on @xmath222 , an isomorphism of graded @xmath222-vector spaces @xmath668 [ [ the - hodge - spectral - sequence . ] ] the hodge spectral sequence .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + for @xmath6 quasi - projective over @xmath222 , there is a sheaf of graded algebras @xmath669 on @xmath6 , the sheafification of a natural presheaf whose sections are @xmath670 .
there is by @xcite a local - to - global ( or ` hodge ' ) spectral sequence @xmath671 converging to @xmath672 , with @xmath673 when @xmath6 is smooth of dimension @xmath383 , and @xmath662 is invertible in @xmath222 , one has @xmath674 by the hkr isomorphism . comparing dimensions of @xmath675 and @xmath676
, one sees that the spectral sequence degenerates at @xmath677 .
one therefore has @xmath678 as a sheaf of graded algebras , @xmath679 has cohomology spaces which form a bigraded algebra @xmath680 . like any local - to - global spectral sequence computing self - exts
, the hodge spectral sequence is multiplicative . from this one
sees : [ edge ] for @xmath6 a quasi - projective @xmath222-scheme , the edge - map @xmath681 is an algebra homomorphism .
let @xmath682 denote a dg enhancement for the unbounded derived category of quasi - coherent sheaves on the @xmath222-scheme @xmath6 ( e.g. @xcite ) .
thus one has an equivalence of @xmath683 with the unbounded derived category .
there is a natural isomorphism of rings @xcite @xmath684 where by the latter we mean the hochschild cohomology of the dg category @xmath685 ( i.e. , self - ext of the identity functor ) .
let @xmath686 be the full dg subcategory of @xmath682 of objects which map to perfect complexes in the unbounded derived category . since @xmath686 is a full subcategory of @xmath682 ,
there is a restriction ( ring ) map @xmath687 and this too is an isomorphism when @xmath6 is quasi - projective , because @xmath682 is the _ ind - completion _ of @xmath686 @xcite .
when @xmath6 is quasi - projective , we also have @xmath688 .
putting these facts together , we obtain the following : [ perf computes hochschild ] for any quasi - projective @xmath222-scheme @xmath6 , one has a canonical algebra - isomorphism @xmath689 if @xmath8 is a split - generator for @xmath690 , the morita invariance of hochschild cohomology ( e.g. @xcite ) implies that the restriction map @xmath691 is an isomorphism .
hence we have @xmath692 , and consequently [ hh for weierstrass ] take an abstract weierstrass curve @xmath353 and let @xmath693 .
then there is a canonical algebra - isomorphism @xmath694 in particular , by lemma [ formality ] , @xmath695 note that the latter isomorphism respects bigradings .
the internal grading of @xmath624 comes from the grading of @xmath200 .
the internal grading of @xmath654 arises because of the @xmath382-action on @xmath215 discussed in the proof of ( [ formality ] ) , which gives rise to an action on @xmath696 , hence on hochschild cohomology .
they agree because the action on @xmath215 induces the grading of @xmath697 ( cf .
the proof of ( [ formality ] ) ) .
we now specialize to quasi - projective curves @xmath162 over @xmath222 .
hkr reads as follows : [ hkr curves ] for a non - singular curve @xmath162 over an arbitrary field @xmath222 , one has an hkr isomorphism of @xmath222-modules @xmath698 for more general curves , we have the following [ curve ses ] for any quasi - projective curve @xmath162 , one has short exact sequences @xmath699 since curves have cohomological dimension 1 , the hodge spectral sequence ( [ hodge ] ) is supported at @xmath677 in two adjacent columns @xmath700 , and therefore degenerates . the edge - maps then give rise to these short exact sequences
. one has @xmath701 , while @xmath702 , the sheaf of derivations of @xmath315 , also known as the tangent sheaf @xmath703 . to go further
, we suppose that @xmath162 is given with a point @xmath704 , and that @xmath705 is non - singular .
we take an affine open cover @xmath706 of @xmath162 such that @xmath707 .
we compute sheaf cohomology @xmath708 using the ech complex @xmath709 . to validate such a computation , we must check that @xmath710 when @xmath711 , @xmath435 or @xmath712 and @xmath713 . as coherent sheaves , @xmath714 and @xmath715
have no higher cohomology on affine open subsets . when @xmath716 , @xmath717 is supported at @xmath70 , by ( [ hkr curves ] ) .
it therefore has no higher cohomology .
we now proceed with the computation .
since @xmath435 is non - singular , ( [ hkr curves ] ) implies that @xmath718 , @xmath719 and @xmath720 for @xmath716 .
the same applies over @xmath712 .
so the ech complex is @xmath721 supported in degrees 0 and 1 . hence : [ sheaf coh of hh ] for @xmath716 , one has @xmath722 therefore , for @xmath723 , @xmath724 in low degrees , we have an isomorphism @xmath725 and two short exact sequences @xmath726 in summary , _ @xmath727 coincides with @xmath728 with corrections in degrees @xmath729 from the cohomology of the functions and of the vector fields on @xmath162 . _
if @xmath656/(f)$ ] is a plane curve singularity , over a field @xmath222 , one can compute @xmath655 via koszul resolutions and a version of hkr .
we have been informed that such calculations go back to quillen @xcite , but an explicit recipe , valid for complete intersections , is explained by kontsevich in @xcite .
the result is that @xmath730 where @xmath731 is a certain dga , namely , the supercommutative @xmath222-algebra @xmath732 , \quad \deg \beta=2,\quad \deg x^*=\deg y^ * = 1\ ] ] and @xmath733 , @xmath734 , @xmath735 , @xmath736 .
thus @xmath737 with the isomorphism ( [ kontsevich ] ) understood , one immediately reads off the even hochschild cohomology of @xmath201 .
let @xmath738 . in the literature @xmath8
is often called the _
tjurina algebra _ of this isolated hypersurface singularity ; it parameterizes a miniversal deformation of the singularity ; see for instance ( * ? ? ?
* theorem 14.1 ) for the case of curve singularities .
[ tjurina ] assume that @xmath739 and @xmath740 are not both zero .
then the map @xmath741 \mapsto [ \alpha \beta^n]\ ] ] is an isomorphism for each @xmath742 . hence @xmath743 $ ] .
let @xmath83 be the @xmath201-submodule of @xmath744 of pairs @xmath745 with @xmath746 .
it has a submodule @xmath747 generated by @xmath748 .
there is a skew - symmetric pairing @xmath749\ ] ] such that @xmath750 .
when @xmath330 is irreducible and @xmath751 , one has @xmath752 .
indeed , @xmath201 is then an integral domain , and the matrix @xmath753 , \ ] ] over @xmath201 , has a non - trivial kernel and therefore vanishing determinant .
observe also that the @xmath201-module @xmath754 is actually a @xmath8-module .
one has a surjective map @xmath755\ ] ] the kernel is @xmath179 when @xmath756 and is @xmath747 when @xmath742 .
the proof is an easy check , in light of ( [ kontsevich ] ) .
note that @xmath757 , the derivations of @xmath201 ; the isomorphism @xmath758 induced by ( [ kontsevich ] ) is the map @xmath759 .
we deduce : [ hh for r ] there is a canonical map of algebras @xmath760 which is an isomorphism in degrees @xmath761 . here
\otimes \left(t \oplus \frac{m}{n}[-1]\right),\quad \deg t=0,\quad \deg\beta=2.\ ] ] the product in @xmath632 combines the algebra structure of @xmath8 , the left and right @xmath8-module structures of @xmath754 , and the skew pairing @xmath26 : @xmath763 the following observation , whose proof is immediate , is helpful in computing @xmath754 : [ koszul ] @xmath754 is the middle homology @xmath764 of the koszul complex @xmath765 , i.e. , the chain complex @xmath766 } r^2 \xrightarrow{\left[\begin{array}{cc } f_x & f_y \end{array}\right ] } r .\ ] ] @xmath765 ( notation from serre ( * ? ? ?
iv ) ) is the tensor product @xmath767 of koszul complexes for the elements @xmath739 and @xmath740 . here
@xmath768 denotes the complex @xmath769 where the map multiplication by @xmath204maps degree 1 to degree 0 .
we take @xmath770 the tjurina algebra @xmath8 ( [ tjurina ] ) is as follows : @xmath771/(x^2 ) & \text{if } 6\neq 0,\\ \bbk[x]/(x^3 ) & \text{if } 3=0,\\ \bbk[x , y]/ ( x^2,y^2 ) & \text{if } 2=0 .
\end{cases}\ ] ] ( the isomorphism takes @xmath772 to @xmath637 , and takes @xmath773 to @xmath179 in the first two cases and to @xmath643 in the third . )
[ m mod n ] there are isomorphisms of @xmath201-modules @xmath774 and @xmath775 & \text{if } 6 \neq 0,\\ [ t,0 ] & \text{if } 3 = 0,\\ [ 0 , t ] & \text{if } 2 = 0 .
\end{cases}\ ] ] by lemma [ koszul ] , @xmath754 is equal as an @xmath201-module to @xmath776 , and hence isomorphic to @xmath777 . if @xmath778 then the element @xmath779 is not a zero - divisor in @xmath201 .
therefore the complex @xmath780 has homology only in degree zero , and @xmath781 .
projection @xmath782 defines a quasi - isomorphism @xmath783 .
hence @xmath784 , and the latter complex is @xmath785 where the differential maps degree 1 to degree 0 .
so @xmath764 is isomorphic as an @xmath201-module ( hence also as a @xmath8-module ) to the annihilator of @xmath739 in @xmath786 .
explicitly , the annihilator is @xmath787/(x^3,y)$ ] if @xmath788 and @xmath412/(x^3,y)$ ] if @xmath789 . if @xmath646 then the element @xmath790 is not a zero - divisor in @xmath201 .
we can then run the same argument with the roles of @xmath739 and @xmath740 interchanged , with a similar outcome .
here @xmath791 is @xmath412/(x^2,y^2)$ ] , and @xmath792 . at this stage , we can easily see that @xmath793 .
the given map @xmath794 is injective , by another easy check , hence an isomorphism .
since @xmath330 is irreducible and @xmath751 , we have the [ skew ] the skew pairing @xmath795 is zero . collating results ( prop .
[ hh for r ] and lemmas [ m mod n ] , [ skew ] ) we obtain [ hh for cusp singularity ] consider the free graded - commutative algebra @xmath796 $ ] , where @xmath797 and @xmath798 .
one then has a map of graded @xmath222-algebras @xmath799 which is an isomorphism except in degrees 0 and 1 . in light of prop .
[ hh for r ] , we need only see that @xmath800 $ ] agrees with the algebra named @xmath632 there . in view of the isomorphism @xmath801 , and the vanishing of @xmath26 ,
the latter algebra is @xmath802\otimes ( t\oplus t[-1 ] ) \cong \bbk[\beta,\gamma]\otimes
t,\quad \deg \gamma=1.\ ] ] we can be more explicit .
we know that @xmath803 .
the map @xmath804 is @xmath201-linear .
it maps @xmath51 to @xmath51 .
it maps a certain class @xmath805 \in h^1(d)$ ] to @xmath806 .
the coefficients are given by @xmath807 when @xmath808 ; @xmath809 when @xmath642 ; and @xmath810 when @xmath646 .
[ derivations ] the lie algebra @xmath811 contains linearly independent vector fields @xmath812 , @xmath813 which restrict to @xmath345 as the vector fields @xmath814 when @xmath639 , @xmath812 and @xmath813 span . when @xmath642 , @xmath811 is 3-dimensional , spanned by @xmath812 and @xmath813 together with @xmath815 when @xmath646 , it is 4-dimensional , spanned by @xmath812 , @xmath813 and @xmath816 ( we will see presently that the subscript @xmath70 of @xmath817 represents the internal degree . ) as shown to us by seidel , the calculation of @xmath818 is straightforward when one takes the ` abstract ' view of @xmath215 ( lemma [ abstract cusp ] ) , as @xmath538 with the marked point @xmath406 , and non - standard structure sheaf @xmath819 . thus @xmath820 is the sheaf of meromorphic vector fields @xmath29 on @xmath538 , with a pole only at @xmath406 , which preserve @xmath819 .
the latter condition forces @xmath29 to be regular at @xmath406 except when @xmath821 .
when @xmath29 is regular at @xmath406 , it must vanish there , except in characteristic 2 .
one has @xmath822 ( the lie algebra of the automorphism group of @xmath538 ) , spanned by @xmath823 , @xmath824 and @xmath825 . in characteristic 2 , all three lie in @xmath826 , while when @xmath827 , only @xmath812 and @xmath813 do ( they span the borel subalgebra @xmath828 of upper triangular @xmath829 matrices modulo scalars ) . when @xmath642 , there is a additional derivation @xmath830 ; when @xmath646 , there is again one additional derivation , @xmath831 . to interpret these vector fields as derivations of @xmath201
, we observe that the normalization map @xmath832 corresponds to the map of rings @xmath656/(y^2-x^3)\to \bbk[z]$ ] , @xmath413 , @xmath414 .
using this we compute that the @xmath833 restrict to @xmath345 in the way stated above .
[ h1 vanishes ] one has @xmath834 .
consider @xmath538 with the cusp modifying its structure sheaf at @xmath835 $ ] .
let @xmath836 $ ] .
we use the affine coordinate @xmath837\mapsto z$ ] . take a derivation @xmath29 of @xmath838 ( i.e. , a vector field ) over @xmath839 .
we must show that it is the difference @xmath840 of vector fields @xmath841 on @xmath842 and @xmath843 on @xmath844 .
we can extend @xmath29 to a meromorphic vector field on @xmath538 .
we proceed by induction on the order @xmath383 of the pole of @xmath29 at @xmath205 . if @xmath29 is regular at @xmath205 ( i.e. , @xmath845 ) then we take @xmath846 and @xmath847 . for the inductive step ,
say @xmath848 at @xmath205 , where @xmath849 and @xmath850 .
let @xmath851 .
then @xmath841 defines a derivation of @xmath819 near @xmath406 , because @xmath850 .
moreover , @xmath852 has a pole of order @xmath853 at @xmath205 , so by induction we can write @xmath854 for @xmath855 on @xmath345 and @xmath856 on @xmath435 ; then @xmath857 and we are done . at this stage
it becomes useful to bring internal gradings into play .
the hochschild cohomology @xmath654 carries an internal grading @xmath70 , arising from the @xmath382-action on @xmath215 .
equivalently , this is the internal grading of @xmath624 arising from the grading of @xmath200 .
there is also an action on @xmath858 , hence an internal grading on @xmath859 . under the @xmath382-action on @xmath215 ,
the functions @xmath6 , @xmath502 and @xmath165 on @xmath215 have respective weights @xmath860 , @xmath861 and @xmath862 . hence on the affine part , the functions @xmath863 and @xmath864 have weights @xmath865 and @xmath866 .
the dga @xmath867 inherits a @xmath382-action , i.e. , a grading @xmath70 , in which the weights of @xmath637 and @xmath643 are @xmath868 and @xmath869 .
the variables @xmath870 and @xmath871 have weights @xmath872 and @xmath873 .
it then follows that @xmath874 . in lemma [ derivations ]
, one should understand @xmath875 to have weight @xmath876 and @xmath877 weight @xmath878 .
then the vector field @xmath879 has weight @xmath438 .
introduce the internal grading on @xmath632 as in the statement of theorem [ hh for cusp curve ] .
it is is set up so that the homomorphism @xmath880 respects it .
let @xmath631 be the subalgebra of @xmath632 where the internal grading is non - positive .
similarly , define @xmath881 .
[ hh to h0 ] the map @xmath882 is an isomorphism .
the kernel of the surjective map @xmath883 is spanned by @xmath884 in degree 2this vanishes by lemma [ h1 vanishes]and @xmath885 in degree @xmath357 .
the action of @xmath886 has weight @xmath357 on @xmath885 .
hence the restricted map @xmath887 is injective , and so an isomorphism .
one has an isomorphism of graded algebras @xmath888 there is a restriction map @xmath889 . by lemmas [ sheaf coh of hh ] and [ h1 vanishes ] , @xmath890 is an isomorphism in degrees @xmath761 . in degree 1 , it is the restriction map @xmath891 , and therefore again injective .
the composite @xmath892 of @xmath890 with the map from lemma [ hh for cusp singularity ] , restricted so as to map between the non - positively graded subalgebras , is the sought - for map .
it is certainly an isomorphism in degrees @xmath893 .
it is also an isomorphism in degree @xmath179 , where both sides reduce to @xmath222 .
the non - positively graded part of @xmath894 has dimension @xmath895 , by lemma [ derivations ] : it is spanned by @xmath812 , together with @xmath896 in characteristic 3 , and @xmath897 , @xmath898 in characteristic 2 .
one also has @xmath899 , and @xmath900 .
we have identified @xmath901 as derivations of @xmath201 , hence as elements of @xmath902 .
one has @xmath903 , in such a way that the map @xmath904 corresponds to the quotient map @xmath905 . to show that @xmath906 is injective , one need only show that the relevant elements @xmath901 are linearly independent in @xmath754 . but as elements of @xmath83 , one has @xmath907 $ ] ; @xmath908 $ ] ; @xmath909 $ ] ; and @xmath910 $ ] .
linear independence in @xmath754 is easily seen .
the isomorphism we want is the composite of the isomorphisms described in the last two lemmas : @xmath911 before leaving @xmath215 , it will be worthwhile to spell out our findings about @xmath912 in a clean form .
they are as follows : [ hh1 ] 1 .
the canonical map @xmath913 is an isomorphism .
2 . let @xmath914 be as at ( [ ker d ] ) ; it is the lie subalgebra of @xmath915 of those vector fields on @xmath916 which preserve @xmath215 .
thus there is a canonical map @xmath917 natural in @xmath222 .
the latter map is an isomorphism .
only the second clause has not already been proved . as a vector field on @xmath354 , we have @xmath918 . similarly , @xmath919 ; @xmath920 ; and @xmath921 .
now , @xmath922 is tangent to @xmath215 , and the restriction of @xmath922 to @xmath215 is the vector field @xmath812 . when @xmath646 , @xmath923 and @xmath924 are also tangent to @xmath215 , and they restrict to @xmath215 as the respective vector fields @xmath897 and @xmath898 .
when @xmath642 , @xmath925 is tangent to @xmath215 and restricts to @xmath215 as @xmath896 .
thus , by lemma [ derivations ] , the map @xmath926 is an isomorphism in every case .
_ note : this subsection is not used elsewhere in the paper .
_ let @xmath152 be the central fiber of the tate curve , defined by @xmath123 where @xmath927 : the proposed mirror to the punctured 2-torus @xmath30 .
the hochschild cohomology @xmath928 is also the hochschild cohomology of @xmath929 . since hochschild cohomology is invariant under @xmath13 quasi - equivalences , theorem [ mainth ]
says that @xmath930 .
so , taking that theorem for granted for now , one can regard this subsection as a computation of @xmath931 .
in general , there is a natural map to the hochschild cohomology of the exact fukaya category from the _ symplectic cohomology _
@xcite of the manifolds , and in certain cases @xcite this is expected to be an isomorphism .
the graded ring we compute here is indeed isomorphic to @xmath932 , though we do not check that the map is an isomorphism .
[ hh nodal ] over any field @xmath222 , there is an isomorphism of graded algebras @xmath933}{\left(\gamma_1\gamma_2,\beta ( \gamma_1 - \gamma_2)\right)}\ ] ] where @xmath934 and @xmath935 $ ] is a free supercommutative algebra .
the computation follows similar lines to that for the cuspidal cubic , and we shall be terse . taking @xmath936/(w)$ ] ,
one readily checks the following : [ tjurina nodal ] the tjurina algebra @xmath8 is reduced to @xmath222 .
its module @xmath754 is 1-dimensional , and therefore the skew pairing @xmath937 is zero .
hence there is a surjective map of algebras @xmath938,\quad \deg \beta=2,\quad \deg \gamma=1,\ ] ] with kernel concentrated in degree 1 . the normalization of @xmath156 is isomorphic to @xmath538 ; the normalization map @xmath939 carries two points , @xmath155 and @xmath131 , say , to the node .
the pullback @xmath940 is contained in @xmath838 as the sheaf of functions @xmath330 such that @xmath941 .
similarly , @xmath942 is contained in @xmath943 as the sheaf of vector fields @xmath843 with @xmath944 .
hence @xmath945 by riemann roch for @xmath538 , @xmath946 in ech terms
, this means that if we cover @xmath538 by @xmath947 and @xmath948 , then every vector field on @xmath712 can be expressed as the difference @xmath840 of vector fields on @xmath345 and @xmath435 with descend to @xmath949 and @xmath950 respectively .
hence @xmath951 this vanishing result gives us the following lemma : [ surj ] the map @xmath952 is surjective with kernel @xmath953 $ ] .
we know that @xmath954 in degrees @xmath955 .
we have @xmath956 the composite of the maps @xmath957 ( the restriction map and the surjection from lemma [ surj ] ) is an isomorphism . we know that @xmath958 , the space of global vector fields , has dimension 1 , so
we need only show that the composite map is non - zero .
the composite @xmath959 $ ] is a map of algebras .
moreover , by lemma [ tjurina nodal ] , @xmath960 is generated in degrees 1 and 2 .
if the map on @xmath958 were zero , the same would be true of the map @xmath961 ; yet we know that the latter map is an isomorphism .
the previous lemma implies that @xmath962\ ] ] is an isomorphism .
now consider the surjective map @xmath963 its kernel is @xmath964 \oplus \h^1(\eut_0,\euo)[-1]$ ] , and the first summand vanishes .
thus we have a surjection of algebras @xmath965\ ] ] with kernel @xmath953 $ ] .
since this map is an isomorphism in degrees @xmath761 , we write @xmath51 to denote a class in @xmath966 , etc .
let @xmath967 be a generator for the 1-dimensional image of @xmath885 in @xmath968 .
we claim that @xmath969 . to see this , observe that the @xmath970-page of the hodge spectral sequence is an algebra isomorphic to @xmath971 , the associated graded algebra for the filtration giving rise to the spectral sequence .
the @xmath970-page here is supported along the @xmath179th row and the @xmath179th column ( in non - negative total degrees ) .
hence @xmath972 for @xmath973 .
thus , to show that @xmath969 , it suffices to show that it vanishes in @xmath974 , which is obvious since this module vanishes .
now let @xmath975 be any lift of @xmath806 to @xmath968 , and let @xmath976 .
then @xmath977 is a basis of @xmath968 .
moreover , @xmath978 , and @xmath979 is zero since it maps to @xmath980 $ ] .
the result now follows .
in this section we shall prove our ` dg comparison theorem ' [ dg comparison ] , and refinements of it .
our plan is to reformulate theorem [ dg comparison ] in terms of minimal @xmath13-structures on the fixed algebra @xmath200 , and prove it in sharper form in that language .
to do so , we need homological perturbation theory .
let @xmath981 be a cochain complex over a commutative ring @xmath201 , with an action of a group @xmath982 by automorphisms .
a _ splitting _ for @xmath983 is an internal direct sum decomposition @xmath984 for each @xmath438 , with @xmath982-invariant summands .
equivalently , it is a @xmath982-equivariant linear map @xmath985 such that @xmath986 ( for given given the direct sum decomposition we put @xmath987 and @xmath988 , while given such an @xmath70 we have @xmath989 \oplus \im\delta } _ { \ker \delta } \oplus \im ( s\circ \delta)$ ] ) .
the set of splittings will be denoted by @xmath990 . when @xmath23 is a @xmath982-equivariant @xmath13-algebra , with cohomology @xmath200 , a splitting for @xmath23 as a cochain complex gives rise , via the homological perturbation lemma @xcite , to a canonical , @xmath982-equivariant @xmath13-structure @xmath991 on @xmath200 , together with equivariant @xmath13-homomorphisms @xmath992 and @xmath993 such that @xmath114 and @xmath155 induce the identity map @xmath994 on cohomology ; equivariance follows from the naturality of the construction .
if one merely splits the cocycles , writing @xmath995 , but does not complement the cocycles , _ the conclusion is the same except that one does not get the map @xmath155 _ @xcite .
one can find a splitting of the cocycles ( not necessarily equivariant ) whenever @xmath200 is a projective module .
let @xmath996 $ ] , and and let @xmath223 be the universal embedded weierstrass curve @xmath997 over @xmath998 .
we are interested in cochain models @xmath999 for the endomorphism algebra @xmath1000 , where @xmath1001 over the curve @xmath223 .
a technical irritation is that @xmath467 , though a perfect complex , is not locally free .
however , there is an autoequivalence @xmath1002 of @xmath1003a twist along the spherical object @xmath52 @xcite such that @xmath1004 and @xmath1005 .
there is an autoequivalence @xmath1006 which is inverse to @xmath1002 in that @xmath1007 and @xmath1008 are both naturally isomorphic to the identity functor .
there is an induced isomorphism @xmath1009 \mapsto [ \tau'\circ h\circ \tau ] \in \ext^\bullet(t',t')$ ] , where @xmath1010 .
we shall set up a cochain complex @xmath1011 which computes @xmath1012 , and use @xmath1002 and @xmath1006 to obtain from it a complex @xmath999 which computes @xmath1000 .
we observe that a splitting for the former complex will transfer to one for the latter . our dg model for @xmath1011 will be the ech complex @xmath1013 where @xmath1014 is the complement of the point at infinity @xmath56 $ ] and @xmath1015 .
the multiplicative group @xmath886 acts on @xmath162 , covering its an action on @xmath373 .
the action preserves each of the two sets in the covering @xmath291 ; hence @xmath886 acts on @xmath1011 by automorphisms
a _ strict _ action ( * ? ? ? * ( 10b ) ) .
[ splittings ] the set @xmath1016 of @xmath886-equivariant retractions is non - empty ; there exists a distinguished splitting @xmath890 . the sheaf @xmath1017 has four line - bundle summands ( ` matrix entries ' ) : @xmath52 ( twice ) , @xmath1018 , and @xmath1019 .
the complex @xmath1020 is the direct sum of the ech complexes for these line bundles , so it suffices to handle them separately .
evidently , we must put @xmath1021 and @xmath1022 .
our tasks are to identify @xmath1023 and @xmath1024 for the line bundles @xmath52 , @xmath1018 and @xmath1019 . * 1 .
* we consider endomorphisms of @xmath52 or of @xmath1019 .
we have @xmath1025 . to describe @xmath1026 , take a function @xmath1027 .
if it is regular at @xmath335 then @xmath1028 . if @xmath306 has a pole of order @xmath1029 at @xmath335 then we can find a function @xmath1030 on @xmath345 such that @xmath1031 has a pole of order @xmath1032 at @xmath335 . on the other hand , if @xmath1033 has a simple pole at @xmath335 , with @xmath1034 with respect to a local uniformizer @xmath70 at @xmath335 , then @xmath1035\cdot \bbk[w]$ ] .
( the restriction of @xmath306 to the locus @xmath1036[\delta^{-1}]$ ] where the discriminant @xmath1037 is non - zero can not be a coboundary , since there is no degree 1 rational map defined on an elliptic curve .
it is of no consequence to our argument whether @xmath306 is a coboundary when we restrict to @xmath1038 . )
we take @xmath1039 , which is indeed regular on @xmath1040 .
one has a local uniformizer @xmath1041 at @xmath335 , and @xmath162 is cut out formally as the graph of @xmath1042 \series{s}$ ] with @xmath1043 ; and @xmath1044 , as required .
thus , we put @xmath1045\cdot g$ ] , and we then have @xmath1046 .
( _ note : _
@xmath306 restricts to the cuspidal fibres as a ech coboundary , but that does not affect our argument . ) turning to the degree 0 part , @xmath1047 consists of pairs @xmath1048 with @xmath406 constant . a complement @xmath1024 is given by @xmath1049 * 2 . *
we consider @xmath1050 .
one has @xmath1051 so @xmath1052 ; and @xmath1053 thus @xmath1054 where @xmath1055 consists of sections over @xmath435 which vanish at @xmath335 .
we consider @xmath1056 .
there are no global sections , so we put @xmath1057 we have @xmath1058 where @xmath1059 \cdot g$ ] , for similar reasons to those explained in case 1 .
we now come to a key point in the construction : take @xmath890 from lemma [ splittings ] . by applying the homological lemma to ( @xmath1060 , we obtain a minimal @xmath13 structure @xmath1061 , linear over @xmath1036 $ ] .
said another way , we obtain a family of minimal @xmath13-structures @xmath1062 parametrized by @xmath1063 , whose structure coefficients depend polynomially on @xmath1064 .
the @xmath382-equivariance of @xmath890 implies equivariance of @xmath1065 ; precisely , for @xmath1066 ( the @xmath438th graded part ) , we have @xmath1067 the dg comparison theorem was stated in terms of abstract weierstrass curves , but the @xmath13-version will be formulated using embedded weierstrass curves .
first we set up the relevant categories : 1 .
the groupoid @xmath1068 whose objects are embedded weierstrass curves , thought of as elements @xmath1063 .
the group of reparametrizations @xmath361 , from ( [ reparam ] ) , acts on @xmath1069 ; we set @xmath1070 2 .
the groupoid @xmath1071 of minimal @xmath13-structures @xmath23 on @xmath200 .
let @xmath1072 denote the part of the @xmath890th hochschild cochain group in cohomological degree @xmath1073 and internal degree @xmath70 , @xmath1074}(a^{\otimes r},a).\ ] ] and write @xmath1075 : these are the truncated hochschild cochains , and they govern deformations of @xmath13-structures @xcite .
let @xmath1076 denote the group of ` rescalings and gauge transformations ' , namely , the group of @xmath1077 whose leading term @xmath1078 is @xmath1079 for some @xmath1080 .
then @xmath1076 acts on the objects of @xmath1071 , @xmath1081 .
we think of @xmath841 as an @xmath13-functor @xmath1082 acting trivially on objects .
the morphisms in @xmath1071 are given by @xmath1083 , where @xmath1084 is the equivalence relation which identifies homotopic functors . `
homotopy ' has the following meaning : say @xmath1085 and @xmath1086 are @xmath13-functors @xmath1087 with the same action on objects .
their difference @xmath1088 then defines a natural transformation , i.e. , a morphism @xmath1089 , where @xmath1090 is the @xmath13-category of non - unital functors @xmath1091 , satisfying @xmath1092 .
one puts @xmath1093 and @xmath1094 for @xmath1095 .
a _ homotopy _ from @xmath1085 to @xmath1086 is a @xmath1096 , with @xmath1097 , such that @xmath1098 .
functors are called _ homotopic _ if a homotopy exists .
[ ainf comparison ] let @xmath222 be a normal commutative ring .
the passage from weierstrass curves over @xmath222 to @xmath222-linear minimal @xmath13-structures on @xmath200 defines a functor @xmath1099 .
precisely : 1 .
each embedded weierstrass curve @xmath1100 gives rise to a minimal @xmath13-structure @xmath1101 , by applying the homological perturbation lemma to @xmath196 , defined via the open cover @xmath291 and the splitting @xmath890 .
these @xmath13-structures depend polynomially on @xmath1063 , and are @xmath382-equivariant , meaning that @xmath1102 .
in particular , the cuspidal cubic @xmath1103 gives rise to the minimal @xmath13-structure @xmath1104 with no higher products : @xmath1105 for @xmath1106 .
2 . each pair @xmath1107 , where @xmath361 is the group of projective transformations acting on weierstrass curves , gives rise to an element @xmath1108 such that @xmath1109 .
moroever , @xmath1110 for @xmath1111 .
one has @xmath1112 . for any @xmath1113
, there exists a homotopy @xmath1114 depending algebraically on @xmath1064
. the functor @xmath313 commutes with base - change ( i.e. , the operation @xmath1115 when @xmath1116 is a ring homomorphism . if @xmath222 is either an integrally closed noetherian domain of characteristic zero , or a field then @xmath313 is an equivalence of categories .
the equivalence clause says that the following three properties hold : * _ essential surjectivity : _ every minimal @xmath13-structure on @xmath200 is isomorphic in @xmath1071 to one of the form @xmath1117 . *
_ : if @xmath1118 , and if @xmath1119 , @xmath1120 are homotopic functors , then @xmath1121 .
* _ fullness : _ if @xmath1122 and @xmath1123 then @xmath1124 is homotopic to @xmath1125 for some @xmath1126 such that @xmath1127 .
the normality condition is there because that is the condition under which we know that the weierstrass differential @xmath26 on the universal weierstrass curve @xmath162 defines a section of the dualizing sheaf .
over a field @xmath222 , every dg structure @xmath203 can be transferred to a quasi - isomorphic @xmath13-structure @xmath23 on the cohomology @xmath1128 . by theorem [ ainf comparison ] ,
@xmath23 is gauge - equivalent to @xmath1117 for some @xmath1064 , which in turn is quasi - isomorphic to the dg category @xmath1129 .
hence @xmath1130 in the @xmath13-sense .
furthermore , if @xmath1131 , realize @xmath1132 as an embedded weierstrass curve @xmath1133 ; so @xmath1134 , and hence @xmath1135 . pick a gauge - equivalence @xmath1136 , and then use the theorem to replace @xmath841 by a homotopic gauge - transformation @xmath1137 , where @xmath1127 . hence @xmath1138 .
[ [ coherence . ] ] coherence .
+ + + + + + + + + + one can ask whether the homotopies @xmath1139 from @xmath1140 to @xmath1141 can be chosen coherently in the sense of ( * ? ? ? * ( 10b ) ) .
let @xmath1142 be the category of minimal @xmath13-structures on @xmath200 , over @xmath222 , in which morphisms @xmath1087 are rescaled gauge transformations @xmath1122 such that @xmath1143 .
let @xmath1144 ; it is an abelian group under composition ( by an eckmann hilton argument ) , and a @xmath1145-module .
the obstruction to coherence is a group cohomology class @xmath1146 .
if one works over a field @xmath222 and asks that the homotopies to be continuous in @xmath1147 and @xmath138 the obstruction lies in the continuous group cohomology @xmath1148 ; if wants them algebraic , it lies in an algebraic version of group cohomology .
we do not pursue these obstructions in detail , but content ourselves with an easy case .
work over @xmath146 , and let @xmath1149 denote @xmath1150 with its analytic ( not zariski ) topology .
we ask that the homotopies be continuous on @xmath1151 .
if we restrict @xmath1147 and @xmath138 to the uni - triangular normal subgroup @xmath1152 then the obstruction lies in @xmath1153 . but @xmath1154 is contractible , hence @xmath1155 .
thus we _ can _ make our homotopies coherent for @xmath1154 .
the lyndon hochschild serre spectral sequence for @xmath1152 is concentrated in a single row , which tells us that @xmath1156 .
so @xmath1157 is where to find the obstruction to extending the coherent homotopies from @xmath1154 to @xmath1149 .
the proof of the equivalence clause of theorem [ ainf comparison ] will be given in the next section ; for now , we shall set up the functor .
we have already set up the functor on objects namely , we have constructed @xmath1158 . now take @xmath1126 and @xmath1063 , and notice that they define an isomorphism @xmath1159 .
we have a diagram of @xmath13 quasi - isomorphisms @xmath1160_{f}\ar[dr]^{f ' } & \\
\eub_{w,\euu } \ar^{g_*}[r ] \ar_{p_w}[d ] & \eub_{g(w),g(\euu ) } \ar@{.>}@<-1ex>_s[ur ] & & \eub_{g(w),\euu}\ar_{p_{g(w)}}[d ] \\
\eua_{w}\ar@<-1ex>_{i_w}[u ] & & & \eua_{g(w)}\ar@<-1ex>_{i_{g(w)}}[u ] } \ ] ] here @xmath330 and @xmath1161 are maps of dga ; they forget one of the three open sets in the covering @xmath1162 .
the dotted arrow marked @xmath70 is an @xmath13-homomorphism which is inverse to @xmath330 , up to homotopy ; @xmath70 is still to be constructed .
the @xmath13-maps @xmath1163 and @xmath1164 are mutual inverses up to homotopy ; they are associated with the splitting @xmath890 , via the homological perturbation lemma ( * ? ? ?
* remark 1.13 ) .
once @xmath70 has been constructed , we shall define @xmath1165 as the @xmath13-composite @xmath1166 . for composition of @xmath13-functors ,
see ( * ? ? ?
* ( 1e ) ) .
to obtain @xmath70 , we consider the following picture ( cf .
* cor . 1.14 ) ) : @xmath1167 & \eub_{g(w),g(\euu ) \cup \euu } \ar^{f}[l ] \\
g^*\eua_w\ar^{v^{-1}}[r ] & \tilde{\eua}_{g(w ) } \ar@<1ex>[l]^{v } \ar_{\tilde{i}_{g(w ) } } [ u ] } \ ] ] the left vertical arrow is the pullback by @xmath306 of the @xmath13-morphism @xmath1168 , produced by means of the splitting @xmath890 and the homological perturbation lemma . _
claim : _ the @xmath1036$]-cochain complex @xmath1169 admits a @xmath382-invariant splitting for its cocycles .
the proof of the claim will be given below .
the splitting gives rise to a minimal @xmath13-structure @xmath1170 on its cohomology @xmath200 , and the @xmath13-quasi - isomorphism @xmath1171 .
the composite @xmath1172 is an @xmath13-morphism inducing the identity map @xmath485 on cohomology .
thus it is a gauge transformation @xmath1173 . as such
, it has a strict inverse @xmath1174 .
we put @xmath1175 .
then @xmath1176 and @xmath1177 are homotopic to identity maps .
( to prove this , use the fact that composition with a fixed functor preserves homotopy and observe that therefore , since @xmath1178 , one has @xmath1179 . )
we now consider the existence of homotopies @xmath1180 .
for this , contemplate the diagram @xmath1181[d]\ar@[cyan][dr ] \\ & \eub_{g(w ) , g(\euu)\cup \euu}\ar@[blue][d]\ar@[red][dr ] \ar@<-1ex>_{h_*}[r ] & \eub_{hg(w ) , hg(\euu)\cup h(\euu)}\ar@[blue][l ] & \eub_{hg(w ) , h(\euu)\cup \euu}\ar@[magenta][dr]\ar[d ] \\ \eub_{w,\euu } \ar@<-1ex>_{g_*}@[red][r ] \ar_{p_w}[d ] \ar@{.>}@/^3pc/[uurr ] & \eub_{g(w),g(\euu)}\ar@[blue][l ] \ar@{.>}@<-1ex>@[red][u ] & \eub_{g(w),\euu } \ar_{p_{g(w)}}@[red][d ] \ar@<-1ex>@[magenta]_{h_*}[r ] & \eub_{hg(w),h(\euu)}\ar[l ] \ar@{.>}@<-1ex>@[magenta][u ] & \eub_{hg(w),\euu}\ar_{p_{hg(w)}}@[magenta][d ] \\ \eua_{w } \ar@[red]@<-1ex>_{i_w}[u ] & & \eua_{g(w)}\ar@<-1ex>@[magenta]_{i_{g(w)}}[u ] & & \eua_{hg(w)}\ar@<-1ex>_{i_{hg(w)}}[u ] } \ ] ] in the lower part of the diagram we see the juxtaposition of the maps that go into the definitions of @xmath1165 ( red arrows ) and @xmath1182 ( magenta arrows ) . in the top row
is the ech complex associated with a 4-set open cover .
the long , curved arrow pointing to it is a homotopy - inverse to the four - step composite formed by the blue arrows ; it must be constructed .
the arrows marked @xmath1183 or @xmath1184 have strict inverses @xmath1185 and @xmath1186 . now consider the arrow pointing down and right from the top of the diagram , shown in cyan .
it is homotopic to the composite of five arrows going the other way round the hexagonal region .
moreover , the composite of the four blue arrows in ( [ big diagram ] ) is equal to the forgetful map @xmath1187 followed by @xmath1188 . as a result
, we see that @xmath1189 is homotopic to the map @xmath1190 which factors through the curving and cyan arrows : @xmath1191\ar@[cyan][dr ] \\
\eub_{w,\euu } \ar^{(hg)_*}[r]\ar@{.>}@/^2pc/[urr ] & \eub_{hg(w),hg(\euu)}\ar@<-1ex>@{.>}[ur ] & & \eub_{hg(w),\euu}\ar_{p_{hg(w)}}[d ] \\ \eua_{w}\ar@<-1ex>_{i_w}[u ] & & & \eua_{hg(w ) } } \ ] ] therefore it will suffice to construct a homotopy - inverse to the latter forgetful map ( this homotopy - inverse is indicated by the straight dotted arrow in ( [ small diagram ] ) ) . for this
, it is sufficient to prove the _ claim : _ the @xmath1036$]-cochain complex @xmath1192 admits a @xmath382-invariant splitting for its cocycles .
the @xmath13-homomorphism @xmath1193 indicated by ( [ small diagram ] ) is homotopic to @xmath1194 .
that is because the solid arrows make the following diagram commutative ; hence the dotted homotopy - inverses form a homotopy - commutative diagram : @xmath1195\ar[ddr]\ar[d ] \\ & \eub_{hg(w ) , hg(\euu ) \cup \euu}\ar[dl]\ar[dr ] \\
\eub_{hg(w),hg(\euu)}\ar@<-.5ex>@{.>}[uur ] \ar@<-.5ex>@{.>}[ur ] & & \eub_{hg(w),\euu } } \ ] ] hence @xmath1196 .
it remains to prove the two claims highlighted above .
recall that we constructed a splitting for @xmath1197 by constructing one for @xmath1198 , meaning that we considered @xmath1199 instead of @xmath1200 .
we shall do the same here . and as before , it suffices to consider the line bundles @xmath52 , @xmath1018 and @xmath1019 which form the matrix entries for @xmath1199
. these sheaves have cohomology in any given degree which is either zero ( in which case the splitting of the cocycles is trivial ) or is a free module of rank 1 , which means that one can complement @xmath1201 in @xmath1202 by choosing any representative for the generator of cohomology and the resulting splitting will be @xmath382-invariant .
this establishes the two claims , and thereby completes the construction of the functor @xmath313 .
introduce the shifted hochschild cochain complex @xmath1203 , given by @xmath1204.\ ] ] a minimal @xmath13-structure on the algebra @xmath200 is a sequence of maps @xmath1205 for @xmath1206 such that @xmath1207 is the multiplication for @xmath200 .
we can view the structure as a truncated hochschild cochain @xmath1208 .
thus the functor @xmath313 is defined , on objects , by a map @xmath1209 .
pick a @xmath1210-equivariant splitting of the cocyles for @xmath1211 , where @xmath223 is the universal weierstrass curve over @xmath1036 $ ] ; then we obtain , for each @xmath1063 , a minimal @xmath13-structure @xmath1212 on @xmath200 .
[ homogeneous ] pick @xmath1213 and @xmath1214 .
the map @xmath1215 given by @xmath1216 is zero except when @xmath1217 is a multiple of @xmath383 .
it is linear if @xmath1218 , and in general is homogeneous of degree @xmath1219 .
the map @xmath155 is a polynomial function , equivariant under @xmath382 , meaning that @xmath1220 , and is therefore homogeneous , of degree @xmath1219 .
we now consider the derivative at @xmath179 of the map @xmath1221 , @xmath1222 take @xmath1223 . by definition , @xmath1224 is a sequence @xmath1225 where @xmath1226 . by lemma [ homogeneous ] ,
@xmath1227 except when @xmath1218 .
thus , we consider @xmath1228 as a map @xmath1229 it is the map which assigns to @xmath1064 its _ primary deformation cocycle _ ( see e.g. @xcite ) . because @xmath1230 for @xmath1231 , we have @xmath1232 . the effect of @xmath313 on morphisms is encoded in a map @xmath1233 which has a partial derivative @xmath1234 we define @xmath1235 when @xmath1236 , the only non - vanishing component of @xmath1237 lies in @xmath1238 .
the vertical maps @xmath1239 in the diagram @xmath1240 & \frakg \ar^{d}[r]\ar_{\kappa^0}[d ] & w \ar[r]\ar^{\kappa^1=\lambda}[d ] & 0\ar[d]^{\kappa^2}\\ 0 \ar[r ] & \eud^{-1 } \ar^{\delta}[r ] & \eud^0 \ar^{\delta}[r]&\eud^1 \ar^{\delta}[r ] & \eud^2 } \ ] ] define a map @xmath1241 of cochain complexes .
we have seen that @xmath1242 . to prove that @xmath1243 , observe that the action of gauge transformations on @xmath13-structures is defined through a map @xmath1244 .
we have @xmath1245 ; differentiating this relation with respect to @xmath306 , and setting @xmath1246 , we obtain the sought equation .
[ kappa iso ] the maps @xmath1247 and @xmath1248 induce isomorphisms @xmath1249\colon \ker d \to \hoch^1(a , a)^{\leq 0}\\ & [ \kappa^1]\colon \coker d \to \hoch^2(a , a)^{\leq 0}\end{aligned}\ ] ] when the base ring @xmath201 is 1 . a field @xmath222 ; or 2 .
the ring of integers @xmath1 ; or more generally 3 .
an integral domain of characteristic zero .
\(i ) we begin with @xmath1250 $ ] .
we claim that @xmath1251=0 $ ] . indeed ,
if @xmath1064 lies in in this kernel , let @xmath1252 be the weierstrass curve @xmath162 over @xmath1253/t^2 $ ] with parameters @xmath1254 .
thus @xmath1252 specializes to @xmath215 at @xmath1255 .
the resulting minimal @xmath13-algebra @xmath1256 is then formal over @xmath1253/t^2 $ ] ; indeed , its class @xmath1257 \in t \hh^2(a , a)$ ] exactly measures non - triviality of @xmath1256 as an @xmath13-deformation of @xmath200 .
since the quasi - isomorphism class of @xmath1258 determines the curve , one finds that @xmath1259/t^2 $ ] .
it follows that @xmath1260 defines a trivial first - order deformation of @xmath215 .
hence @xmath1261 .
we quote from @xcite or section [ hochschild section ] the result that , as graded @xmath222-modules , we have @xmath1262 thus @xmath1263 is abstractly isomorphic to @xmath1264 as a graded vector space , and hence the injection @xmath1250 $ ] is an isomorphism .
now consider @xmath1265 \colon \ker d \to \hh^1(a , a)^{\leq 0}$ ] .
recall that @xmath373 is a @xmath377-module . on the level of cohomology
, @xmath1266 is a @xmath1267-module .
moreover , this module is easily checked to be faithful : that is , the action homomorphism @xmath1268 is injective .
moreover , @xmath1269 is a @xmath1270-module , and the map @xmath1271 $ ] respects the actions on cohomology : @xmath1272 \cdot [ \kappa^1 w ] = [ \kappa^1(\xi\cdot w)].\ ] ] given @xmath1273 , pick a @xmath1274 such that @xmath1275 .
we then have @xmath1276\neq 0 $ ] .
hence @xmath1265 $ ] is injective .
both domain and codomain of @xmath1265 $ ] are isomorphic to @xmath1277 therefore @xmath1250 $ ] is an isomorphism .
\(ii ) it will be helpful to note at the outset that for any ring @xmath201 one has @xmath1278 , that @xmath1279 , and that these canonical isomorphisms are compatible with the construction of the map @xmath1280 .
let @xmath1281 be a residue field of @xmath1 .
form the @xmath1-cochain complex @xmath1282 , a complex of free abelian groups , and note , using ( i ) , that @xmath1283 while @xmath1284 is an isomorphism ( so the latter vector space is 1-dimensional ) . by universal coefficients @xcite
, @xmath1285 has a direct summand @xmath1286 .
consequently @xmath1287 . since @xmath1288 is a finitely - generated abelian group , one deduces @xmath1289 .
one then has @xmath1290 , and so by universal coefficients again , @xmath1291 .
hence @xmath1292 is free of rank 1 . over @xmath1
, one has @xmath1293 , and a look at the exact sequence of the mapping cone then tells us that @xmath1294 is an isomorphism .
the same exact sequence then tells us that @xmath1265 $ ] and @xmath1250 $ ] are isomorphisms .
\(iii ) take the @xmath1-cochain complex @xmath1295 as before .
universal coefficients now tells us that @xmath1296 the ring @xmath201 is torsion - free as an abelian group , so @xmath1297 .
the exact sequence of the mapping cone then tells us that @xmath1250 $ ] is an isomorphism . over @xmath201 ,
one has @xmath1298\cong r$ ] , again because @xmath201 is torsion - free .
part of the exact sequence of the mapping cone reads @xmath1299 } \hoch^1(a_r , a_r)^{\leq 0}\to 0,\ ] ] and since @xmath1300 is non - zero , and @xmath201 an integral domain , the map @xmath1301 must be zero . hence @xmath1265\colon r\to \hoch^1(a_r , a_r)^{\leq 0}$ ] is also an isomorphism . out of caution , work over a field @xmath222 in this remark .
the map @xmath1250 $ ] has a straightforward deformation - theoretic meaning ( a first - order deformation of weierstrass curves gives a first - order deformation of @xmath13-structures on @xmath200 ) .
the map @xmath1265 $ ] may then be characterized as the unique map that makes the following diagram commute : @xmath1302\ar[d]_{[\kappa^0 ] } & \end w_\frakg \ar[d]^{[\kappa^1 ] } \\
\hoch^1(a , a)^{\leq 0 } \ar[r ] & \end \hoch^2(a , a)^{\leq 0}. } \ ] ] the horizontal arrows are the module actions .
this leads to a derived - categorical construction of @xmath1265 $ ] as the composite of the canonical isomorphisms of graded @xmath222-modules @xmath1303 before embarking on the proof of theorem [ ainf comparison ] , we say a word about the methodology .
we are very close here to the framework for deformation theory which uses differential graded lie algebras ( dgla ) .
for instance , the dgla @xmath1304 determines , in characteristic zero , a deformation functor which assigns to a local artinian @xmath222-algebra @xmath200 the solutions in @xmath1305 to the maurer
cartan equation @xmath1306 = 0 $ ] , modulo the group @xmath1307 ( see e.g. @xcite )
. a standard approach would be to show that @xmath1241 is a map of dgla , and conclude , given its effect on cohomology , that the deformation theories controlled by @xmath1308 and @xmath1309 coincide @xcite .
the route we have actually taken is a variant of this standard approach .
everything here works over arbitrary fields ; there is no need for characteristic zero .
a minimal @xmath13-structure on @xmath200 is a hochschild cochain @xmath1310 which satisfies @xmath1311 .
the gerstenhaber square @xmath1312 agrees with @xmath1313 $ ] when @xmath1314 is invertible .
there are no artinian rings in the picture , but the length filtration of hochschild cochains serves as a substitute .
the map @xmath1241 is not quite a lie algebra homomorphism ; it might be possible to promote it to an @xmath1315-homomorphism , but we have chosen to use the functor @xmath313 more directly . we have set up the functor @xmath1221 over regular rings @xmath201 , and must prove that it is an equivalence in the stated sense .
_ essential surjectivity : _ if two minimal @xmath13-structures on @xmath200 , with composition maps @xmath1316 and @xmath1317 , agree for @xmath1318 , then the two structures are gauge - equivalent . indeed , one proves inductively that if @xmath1319 for @xmath1320 then one can find a gauge transformation @xmath841 such that @xmath1321 for @xmath1320 and @xmath1322 . to do so
, one notes that @xmath1323 defines a class in @xmath1324 .
this hochschild module is zero for @xmath1325 : when @xmath201 is a field , this holds by theorem [ hh for cusp curve ] ; when @xmath1326 , it then follows by the universal coefficients ; when @xmath201 is an integral domain of characteristic zero , it then follows by universal coefficients from the @xmath1 case ( cf .
the proof of theorem [ kappa iso ] ) .
one uses a trivialization of @xmath1323 to define the gauge transformation ( cf .
@xcite ) .
now let @xmath1327 be a minimal @xmath13-structure on @xmath200 , over @xmath1 .
our goal is to show that @xmath23 is gauge - equivalent to @xmath1117 for some @xmath1063 .
we shall repeatedly apply gauge transformations to @xmath23 , without notating them . if @xmath23 is formal then @xmath1328 .
if it is not formal then we apply a gauge transformation so as to arrange that @xmath1329 for @xmath1330 but @xmath1331 \neq 0 \in \hh^2(a , a)^{2-d}$ ] . by theorem [ kappa iso ]
, one has @xmath1331= [ \kappa^1(w)]$ ] for a unique @xmath1332 ; we may assume , by applying another gauge transformation , that in fact @xmath1333 .
by lemma [ homogeneous ] , @xmath1334 for @xmath1320 .
thus @xmath1335 for @xmath1336 .
the difference @xmath1337 is a cocycle , as one checks using the @xmath13-relations .
if it is exact , one can adjust @xmath525 by a gauge transformation which leaves @xmath1338 untouched for @xmath1336 such that @xmath1339 equals @xmath1340 , whereupon @xmath1341 is a cocycle , and we can repeat the process .
what we find is that either @xmath1342 is gauge - equivalent to @xmath1343 , or else there is a @xmath1344 such that , after applying a gauge transformation to @xmath1342 , one has @xmath1345 for @xmath1346 and @xmath1347\neq 0 $ ] . in the latter case ,
write @xmath1348 , and consider @xmath1349
. the differences @xmath1350 can be killed by gauge transformations for @xmath1351 .
we continue in the same fashion ; the process stops once has made @xmath1338 agree with @xmath1352 for @xmath1318 . _
faithfulness : _ we shall consider the case of automorphisms : say @xmath1353 , and that @xmath1354 ( @xmath1355 means ` is homotopic to ' ) ; we shall show that then @xmath1246 . from this , the general case will follow : say @xmath1118 and @xmath1356 . let @xmath1357
. then @xmath1358 and @xmath1359 .
recall that the lie algebra @xmath368 is a graded @xmath201-module : @xmath1360 .
there is an induced filtration by lie subalgebras @xmath1361 one easily finds algebraic subgroups @xmath1362 , with @xmath1363 ; for instance , @xmath1364 is the unipotent subgroup @xmath345 .
we also filter the group @xmath1076 , so that @xmath1365 , as follows : take @xmath1366 ( so @xmath1367 , and @xmath1368 with @xmath406 a unit ) . if @xmath1369 , say @xmath1370 ; if in addition , @xmath1371 for @xmath1372 , say @xmath1373 . if @xmath1374 then @xmath1375 . if in addition @xmath526 then we have @xmath1376 \neq 0 \in \hh^1(a , a)^{1-d}$ ] .
hence we have the following observation : if @xmath1353 , if @xmath1377 for some @xmath526 , and if @xmath1378 \neq 0 $ ] , then @xmath1379 .
now let @xmath1380 , where @xmath1381 , with @xmath1353 and @xmath1382 .
since @xmath1354 , @xmath1383 induces the identity on @xmath1384 .
hence @xmath306 lies in the unipotent subgroup @xmath1385 .
hence @xmath1386 , i.e. , @xmath1387 .
thus @xmath1388 , but the fact that @xmath1354 implies that @xmath1389 is a coboundary . by the highlighted observation
, we deduce that @xmath1390 .
we argue similarly that @xmath1389 is a coboundary , hence that @xmath1391 , and finally that @xmath1392 is a coboundary , hence that @xmath1393 . _
fullness_. next consider the assertion that given @xmath1394 and given @xmath1395 such that @xmath1396 , there is some @xmath1397 such that @xmath1127 and @xmath1398 .
we first note a non - emptiness statement : if @xmath1396 where @xmath1394 , then there is some @xmath1399 such that @xmath1400 .
this is true because the formal diffeomorphism - type of the @xmath13-structure @xmath1117 determines the affine coordinate ring , and hence the curve .
that is : @xmath1401 likewise , if @xmath1396 where @xmath1370 , then there is some @xmath1402 such that @xmath1400 . indeed ,
if we know @xmath1117 up to gauge - equivalence then we can reconstruct not only the affine coordinate ring of the curve , but also the weierstrass differential on the curve .
hence , since @xmath1068 and @xmath1071 are groupoids , and @xmath313 functorial , it will be enough to prove fullness when @xmath1403 .
we wish to show that the map @xmath1404 which we already know to be injective , is also surjective .
moreover , by a rescaling argument , we see that it suffices to prove this under the assumption that @xmath1370 .
take some @xmath1405 with @xmath1406 .
if @xmath843 is homotopic to the identity , we are done .
if not then there is a @xmath1407 such that @xmath843 is homotopic to some @xmath1408 , @xmath1409 , where @xmath1410\neq 0 \in \hh^1(a , a)^{1-d}$ ] . by theorem [ kappa iso ] , @xmath1410=[\kappa^0(\xi_1)]$ ] for some @xmath1411 .
we then have @xmath1412/\epsilon^2)$ ] , and @xmath1413 ) = [ w ] \in w \otimes_{\bbk } \bbk[\epsilon]/\epsilon^2 $ ] .
it is possible to lift @xmath1147 to @xmath1414/\epsilon^3)$ ] , with @xmath1415 .
we wish to do so in such a way that @xmath1416 ) = [ w]\in w \otimes_{\bbk } \bbk[\epsilon]/\epsilon^2 $ ] .
the obstruction is the class of @xmath1417 in @xmath1418 .
now , @xmath1250 $ ] maps @xmath373 to @xmath1264 , and carries @xmath1419 to @xmath1420 $ ] . because @xmath1421 extends to a gauge transformation which preserves @xmath1117 , @xmath1422 \in [ w , \hh^1(a , a)^{\leq 0}]$ ] .
hence @xmath1423 .
inductively , we extend @xmath1147 to @xmath1424 ( mod @xmath1425 ) with @xmath1426 . we can then find a homomorphism @xmath1427 which has this series as its 3-jet , and we put @xmath1428 .
one has @xmath1353 .
we next ask whether @xmath1429 is homotopic to @xmath1430 .
the obstruction we encounter now lies in @xmath1431 with @xmath1432 . repeating the argument
, we eventually obtain an element @xmath366 with @xmath1433 .
_ note : this subsection will not be used elsewhere , except in the variant method ( [ variants ] ) of our proof - by - elimination that @xmath156 is mirror to @xmath30 . _ the truncated hochschild cohomology @xmath1434 , in addition to its graded algebra structure , carries the gerstenhaber bracket @xmath1435 $ ] .
these two operations make @xmath1436 a gerstenhaber algebra over @xmath222 : the product is graded commutative , the bracket makes @xmath1437 $ ] a lie superalgebra , and the bracket is a bi - derivation for the product .
moreover , the internal grading @xmath70 is additive under the bracket .
since the brackets can readily be computed from our results , we record it here .
we use the notation of theorem [ hh for cusp curve ] .
[ bracket ] @xmath631 carries a gerstenhaber bracket @xmath1438 $ ] , respecting the internal grading @xmath70 and making the ( canonical ) isomorphism @xmath1439 from theorem [ hh for cusp curve ] a map of gerstenhaber algebras .
the gerstenhaber bracket on @xmath631 is as follows .
the lie algebra @xmath1440 is given by @xmath1441= - \gamma , & & \text{if } 3= 0\\ & \bbk \gamma \oplus \bbk x\gamma \oplus \bbk y \gamma , & & [ \gamma , x\gamma]=0 , \ ; [ \gamma , y\gamma ] = \gamma , \ ; [ x\gamma , y\gamma]=x\gamma , & & \text{if } 2=0.\end{aligned}\ ] ] one has @xmath1442=0 .
$ ] the brackets @xmath1443 $ ] are given by @xmath1444 = -4 x\beta , & [ \gamma , \beta ] = -6 \beta , & & & \text{if } 6\neq 0;\vspace{5 mm } \\ & [ x\gamma , x^2\beta ] = x^2\beta , & [ x\gamma , x\beta ] = -x\beta , & [ x\gamma , \beta ] = 0 , & \\ & [ \gamma , x^2\beta ] = x\beta , & [ \gamma , x\beta ] = -\beta , & [ \gamma , \beta ] = 0 & & \text{if } 3= 0 ; \vspace{5mm}\\ & [ y\gamma , xy\beta ] = xy\beta , & [ y\gamma , y\beta ] = y\beta , & [ y\gamma , x\beta]=0 , & [ y\gamma , \beta ] = 0,\\ & [ x\gamma , xy\beta ] = 0 , & [ x\gamma , y\beta ] = x\beta , & [ x\gamma , x\beta ] = 0 , & [ x\gamma,\beta]=0 , \\ & [ \gamma , xy\beta ] = x\beta , & [ \gamma , y\beta ] = \beta , & [ \gamma , x\beta]= 0 , & [ \gamma,\beta]=0 , & \text{if } 2=0 . \end{array}\ ] ] @xmath631
is generated as a unital @xmath222-algebra by @xmath86 and @xmath1445 .
the remaining brackets are therefore determined by the leibniz rule .
the structure of @xmath86 was already computed in prop .
[ hh1 ] ; the first assertion here is essentially a restatement .
a hochschild 2-cocycle @xmath406 for @xmath200 determines a first - order deformation of @xmath200 as an @xmath13-algebra , and that the gerstenhaber square @xmath1446 \in \hh^3(a , a)$ ] is the obstruction to lifting it to a second - order deformation . by theorem [ kappa iso ] ,
all first - order deformations of @xmath200 come ( via @xmath1248 ) from deformations of @xmath215 as a weierstrass curve .
these weierstrass deformations lift to second order ; hence the algebraic deformation of @xmath200 also lifts .
this shows that @xmath1447 .
the bracket on @xmath1448 is given by @xmath1449 = a\circ b + b\circ a = ( a+b)\circ ( a+b ) - a\circ a - b\circ b$ ] , so @xmath1450=0 $ ] .
the adjoint action of the lie algebra @xmath1451 on @xmath1448 is the natural action of infinitesimal ( @xmath13 ) automorphisms of @xmath200 on first order deformations .
but the lie algebra @xmath1452 acts via lie derivatives on weierstrass deformations of @xmath215 .
namely , take @xmath1453 .
we have a vector field @xmath1454 on @xmath373 .
take @xmath1063 , and regard it as a translation - invariant vector field on @xmath373 ; then the adjoint action of @xmath1455 on @xmath1064 is given by the lie bracket of vector fields : @xmath1456(0)$ ] .
one computes these brackets using the formulae ( [ subst ] ) .
the maps @xmath1271 $ ] from ( [ kappa iso ] ) intertwine this lie derivative with the adjoint action of @xmath1457 on @xmath1448 .
the statement about generation is clear from the definition of @xmath631 .
once the lie algebra @xmath86 has been computed , another approach to obtaining the rest of the brackets is to compute the bv operator @xmath1458 .
along with the product , that determines the brackets , and it is sharply constrained by the bidegrees ( which , for instance , force @xmath1459 ) .
we have used this method as an independent check of the above formulae for @xmath1460 $ ] and @xmath1461$]in particular , as a check on the signs .
[ [ t_0-as - a - liouville - manifold . ] ] @xmath30 as a liouville manifold .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + let @xmath8 be a 2-dimensional torus marked with a point @xmath9 and equipped with a symplectic form @xmath26 . in the complement @xmath28
, @xmath26 is exact ; say @xmath1462 .
a small punctured neighbourhood of @xmath9 is symplectomorphic to the negative end in the symplectization of the contact manifold @xmath1463 , where @xmath1464 is a small loop encircling @xmath9 .
the disc @xmath1465 bounded by @xmath1464 has the property that there is a diffeomorphism @xmath1466 , which is the identity on the boundary , under which @xmath29 pulls back to @xmath1467 .
@xmath30 is a
_ liouville manifold _
( see e.g. @xcite ) : an exact symplectic manifold @xmath1468 whose liouville vector field @xmath1469 is complete , with a given compact hypersurface @xmath1470 enclosing a compact domain @xmath5 in @xmath83 , such that @xmath1469 is nowhere vanishing on @xmath1471 .
the liouville structure of @xmath30 includes the selection of the curve @xmath1464 , but is independent of this choice , as well as that of the form @xmath29 , up to _ liouville isomorphism . _
a liouville isomorphism @xmath1472 is a diffeomorphism such that @xmath1473 for a compactly supported function @xmath1474 .
[ [ grading . ] ] grading .
+ + + + + + + + we specify a _ grading _ of @xmath30 as a symplectic manifold , that is , a trivialization of the square of the canonical line bundle ; in two dimensions , that amounts to an unoriented line field @xmath1475 . anticipating our discussion of @xmath32 , we shall always choose @xmath31 extends over @xmath9 to a line field on @xmath8 .
such line fields form a torsor for @xmath1476 ; one has @xmath1477 . [
[ branes . ] ] branes .
+ + + + + + + the objects of @xmath1478 , the exact fukaya category of the liouville manifold @xmath30 , as defined in @xcite , are ` exact lagrangian branes ' @xmath72 : embedded closed curves @xmath1479 such that @xmath1480 , equipped with spin structures and gradings .
a grading for @xmath86 is a homotopy - class of paths from @xmath1481 to @xmath1482 inside @xmath1483 .
if @xmath86 and @xmath1484 are graded curves then a transverse intersection point @xmath1485 has a degree @xmath1486 , where @xmath54 is the net rotation of the path from @xmath1487 to @xmath1488 to @xmath1489 .
when @xmath31 is _ oriented_pointing along @xmath51 , say the grading for a curve @xmath86 induces an orientation for @xmath86 .
the sign @xmath1490 is then the intersection sign @xmath1491 \cdot [ l]$ ] ( note the order ! ) , regardless of which orientation for @xmath31 was selected ; see figure [ hmscorrespondence ] for relevant examples .
there are four inequivalent spin structures on each curve @xmath86 , defined by the two orientations and the two double coverings .
the morphism - spaces in @xmath1492 are @xmath1-linear cochain complexes ; @xmath1493 is a cochain complex @xmath1494 computing the floer cohomology @xmath1495 ; here @xmath1496 is the image of @xmath86 under the time-1 map @xmath94 of a specified liouville isotopy
. the degrees @xmath1497 define the grading of the complex ; the spin structures determine the sign - contributions of the holomorphic bigons defining the differential .
floer cohomology is invariant under liouville isotopies @xmath1498 ( i.e. , 1-parameter families of liouville automorphisms ) in that @xmath1499 . as a result ,
liouville - isotopic objects are quasi - isomorphic in @xmath1478
. one has @xmath1500
\oplus \z$ ] : this is just a formula for the morse cochain complex for @xmath86 . if @xmath1501 denotes the brane sharing the same oriented , graded curve as @xmath72 but with the other double covering , then one has @xmath1502 \}$ ] , which is the morse complex for @xmath86 with the local system with fiber @xmath1 and holonomy @xmath180 .
the composition map @xmath1503 can be understood purely combinatorially ( * ? ? ?
* ( 13b ) ) , provided that @xmath1504 , @xmath1505 and @xmath1506 are in general position .
the coefficient for @xmath1507 in @xmath1508 is a signed count of immersed triangles bounding @xmath1504 , @xmath1505 and @xmath1506 in cyclic order , with convex corners at @xmath1509 , @xmath1510 and @xmath1507 .
we shall discuss the sign when we discuss the relative fukaya category .
when @xmath1511 are not in general position , one moves @xmath1505 and @xmath1506 by exact isotopies so as to make them so . beyond the differential @xmath1512 and the composition @xmath1207 , the higher @xmath13-structure maps @xmath525 are defined through inhomogeneous pseudo - holomorphic polygons for a complex structure @xmath1513 on @xmath30 compatible with the orientation and the liouville structure at infinity .
we refer to @xcite for the foundations ; since we shall not make any direct calculations with the higher structure maps , we need not say more here . [ [ describing - the - objects . ] ] describing the objects .
+ + + + + + + + + + + + + + + + + + + + + + + oriented simple closed curves in @xmath30 do not realize all free homotopy classes of loops , but just one free homotopy class per non - zero homology class . indeed , suppose that @xmath1514 and @xmath975 are two such curves representing the same class in @xmath1515 .
it follows from the ` bigon criterion ' @xcite that they can be disjoined by isotopies in @xmath30 . assuming that they are disjoint
, they divide @xmath8 into two annuli .
only one of the annuli contains @xmath9 , and hence an isotopy from @xmath1514 to @xmath975 can be realized in @xmath30 . if , moreover , @xmath1514 and @xmath975 are isotopic simple closed curves which are both exact then they cobound an immersed annulus of area zero , and hence are liouville - isotopic .
a non - exact simple closed curve representing a non - trivial homology class can be isotoped to an exact one ( just take an isotopy of an appropriate flux ) .
[ [ generation . ] ] generation .
+ + + + + + + + + + + pick a basis @xmath45 for @xmath1516 with @xmath1517 .
let @xmath50 and @xmath1518 be objects of @xmath1478 of respective slopes @xmath54 and @xmath51 . in figure [ hmscorrespondence ] , the line field @xmath31
is chosen to be parallel to @xmath51 . with that choice , we grade @xmath1504 by the trivial homotopy from @xmath1519 to @xmath1520 , and @xmath1315 by the homotopy depicted ,
so that @xmath1521 where @xmath637 is the generator for @xmath1522 .
we also orient these curves in the respective directions of @xmath54 and @xmath51 ( so , if @xmath31 is oriented in the direction of @xmath51 , the orientations are those obtained from the orientation of @xmath31 and the grading of the curve ) .
it is permissible to choose a different line field @xmath1523 , so long as it is defined on all of @xmath8 , twisting say , @xmath204 times along @xmath1315 and @xmath205 times along @xmath1504 ; but in that case we put the same number of extra twists into the gradings of @xmath1504 and @xmath1315 , so that one still has @xmath1521 .
we take the spin - structures on @xmath1504 and @xmath1524 to be the _ non - trivial _
double covers . a convenient way to keep track of double coverings of curves @xmath86 is to mark a point @xmath1525 , and declare the double cover to be trivial on @xmath1526 and to exchange the sheets over @xmath1527 .
such stars appear in figure [ hmscorrespondence ] .
let @xmath23 be the full @xmath13 sub - category of @xmath1478 with objects @xmath1528 .
every object of @xmath1478 whose
double covering is non - trivial is quasi - isomorphic , in @xmath1529 , to a twisted complex in @xmath23 .
this is because such objects are given , up to possible reversal of orientation and shift in degree , by iterated dehn twists @xmath1530 , @xmath1531 , and the effect of the dehn twist @xmath1532 on the fukaya category is the twist functor along the spherical object @xmath53 , provided that the double covering of @xmath53 it is non - trivial . for a systematic account , including the double - cover condition , see ( * ? ? ?
* theorem 17.16 ) ; a low - tech account of the case at hand is given in @xcite .
curves with trivial spin structures are not quasi - isomorphic in @xmath1533 to object of @xmath1534 ; in fact , we shall see later that these objects do not represent classes in the subgroup of @xmath1535 of @xmath1536 .
what is true , however , is that for any object @xmath6 of @xmath1478 , the direct sum @xmath1537 $ ] is quasi - isomorphic , in @xmath1529 , to a twisted complex in @xmath23 .
one can write down an explicit twisted complex representing @xmath1537 $ ] using ( * ? ? ?
5.8 , theorem 17.16 , and formula 19.4 ) , bearing in mind that there is an exact isotopy @xmath1538 where @xmath1539 is a symplectic dehn twist along @xmath806 .
the isotopy takes place in a compact domain containing the surface bounded by @xmath1464 , and is trivial near the boundary .
the relation arises from the monodromy of an anticanonical lefschetz pencil on @xmath1540 with 8 of the 9 base - points blown up .
consequently , _
@xmath23 split - generates @xmath1478 _ , i.e. , @xmath1541 is a quasi - equivalence .
the _ wrapped _
fukaya @xmath13-category @xmath1542 of a liouville manifold @xmath1468 is set up in @xcite .
it is a @xmath1-linear @xmath13-category containing @xmath1543 as a full subcategory .
its objects are again certain lagrangian branes @xmath72 .
precisely , @xmath86 must be a properly embedded , eventually conical lagrangian submanifold , exact in the strong sense that @xmath1544 for some @xmath1545 .
the brane structure consists of a spin structure and a grading on @xmath86 , and both can be taken to be eventually translation - invariant on the conical end .
the morphism spaces @xmath1546 are cochain complexes @xmath1547 computing wrapped floer cohomology @xmath1548 , which is lagrangian floer cohomology @xmath1549 for a hamiltonian diffeomorphism @xmath94 which ` accerelates ' on the conical end .
that is , @xmath94 acts on the conical end @xmath1550 as @xmath1551 , where @xmath1552 is the time @xmath1553 reeb flow on the contact cross - section @xmath747 and @xmath1554 is a function which increases rapidly in a precise sense . as a consequence @xmath94 ` wraps ' @xmath86 around the end many times , typically producing an infinity of intersections with @xmath1484 .
an object in @xmath1555 is either an object of @xmath1556 , or an eventually - straight oriented arc @xmath200 , with a spin - structure , of which there is only one isomorphism class per orientation , and a grading .
a grading is , again , a way to rotate from @xmath1557 to @xmath1558 inside @xmath1559 .
the oriented arc @xmath200 has an initial segment @xmath1560 and a final segment @xmath1561 , where @xmath1562 and @xmath1563 are points on the circle @xmath1464 .
exact arcs with fixed @xmath1562 and @xmath1563 are liouville - isotopic if and only if they have the same slope , i.e. , represent the same primitive class in @xmath1564 .
exact arcs with the same slope but different endpoints are not liouville - isotopic , but are nonetheless quasi - isomorphic in @xmath1555 . if @xmath72 is an object of @xmath1555 such that @xmath86 is closed , and @xmath1565 any object of @xmath1555 such that @xmath1566 \not \in \ { [ \lambda ] , -[\lambda]\}$ ] in @xmath1567 , one can apply a liouville isotopy to @xmath1504 so as to reduce the number of intersections to the minimal intersection number @xmath1568 .
there are then no bigons , so @xmath1569 .
if @xmath200 is an arc of the same slope as the closed curve @xmath86 , one can apply a liouville isotopy to @xmath200 that disjoins it from @xmath86 , whence @xmath1570 .
it then follows that @xmath1571 : this is an instance of floer - theoretic poincar duality , which applies only when one of the objects is a closed lagrangian .
one can show that @xmath1572 is the non - commutative ( tensor ) algebra @xmath1573 on generators of degree 1 modulo the two - sided ideal @xmath1574 , but we shall not need to use this assertion .
[ [ generation.-1 ] ] generation .
+ + + + + + + + + + + let @xmath200 be any non - compact exact arc , equipped with a brane structure so as to make it an object of @xmath1555 .
let @xmath1504 be a simple closed curve of the same slope as @xmath200 , and @xmath1315 a curve which intersects @xmath1315 transversely at a single point .
equip these three curves with brane structures , with the double - coverings of the two closed curves both non - trivial .
then @xmath1575 split - generates @xmath1555 .
indeed , we have already seen that @xmath1576 split - generates @xmath1478 , so we need only consider the arcs .
any arc can be obtained from @xmath200 by a sequence of dehn twists along @xmath1504 and @xmath1315 , and so by an easy adaptation of ( * ? ? ?
* theorem 17.16 ) , or by a much more elementary argument which applies to the surface case , can be represented as a twisted complex in @xmath1577 , @xmath50 and @xmath53 . the relative fukaya category @xmath32 ( cf .
@xcite ) has the same objects as @xmath1578 ; and @xmath1579 i.e. , @xmath1580 is a free @xmath21-module on the intersections @xmath1581 .
the line field @xmath31 on @xmath8 defines a grading , and hence makes the hom - spaces graded modules .
the @xmath13-structure @xmath1582 is defined through inhomogeneous pseudo - holomorphic polygons , now in @xmath8 .
such polygons @xmath841 count with a weight @xmath1583 , where @xmath1584 is a sign , defined just as in @xmath1578 , and @xmath1585 is the intersection number with @xmath9 . a formula for @xmath1584 is given in @xcite ( see also @xcite ) .
we shall give here only a special case in which all corners of the polygon ( i.e. , @xmath1586 for @xmath1587 and @xmath1588 ) have _ even index _ @xmath1589 .
in that case , @xmath1590 , where where @xmath70 is the number of stars on the boundary ( recall that the stars designate non - trivial monodromies for the double covers ) . the same perturbations that define the @xmath13-structure in @xmath1578 succeed in defining an @xmath13-structure here too .
the proof uses automatic regularity for holomorphic maps to surfaces ( * ? ? ? * ( 13a ) ) , and is otherwise unchanged from the proof in the the exact case .
the resulting @xmath13-structure is an invariant of @xmath1591 .
moreover , up to quasi - equivalence it is independent of @xmath29 ; the proofs of these assertions are straightforward adaptations of their analogues for @xmath30 given in @xcite .
[ [ generation.-2 ] ] generation . + + + + + + + + + + + the objects @xmath50 and @xmath53 split - generate @xmath32 .
the proof is essentially the same as for @xmath1556 .
one can see curves with non - trivial double coverings as explicit twisted complexes in @xmath50 and @xmath53 , cf . @xcite . for arbitrary objects
@xmath6 , the sum @xmath1537 $ ] is again a twisted complex in @xmath50 and @xmath53 ; the proof uses the main results of @xcite just as before , via the relation @xmath1592 $ ] in the graded symplectic mapping class group .
it is significant here that @xmath1593 , and more particularly that @xmath31 extends over @xmath8 , since seidel s argument depends on the presence of absolute gradings ; with that point noted , the argument applies to @xmath24 as it does to @xmath30 . besides the ` open string ' invariants @xmath1543 and @xmath1542
, liouville manifolds @xmath83 have a ` closed string ' invariant , the symplectic cohomology algebra @xmath1594 ( see @xcite for an exposition and foundational references ) .
in the first , place @xmath1594 is a graded - commutative graded ring ; the grading depends on a choice of grading for @xmath83 as a symplectic manifold .
it also comes with a ring map @xmath1595 , which in our grading convention is a map of graded rings ; this pins down our normalization for the grading of @xmath1594 , which in some other accounts ( such as @xcite ) differs from ours by @xmath1596 . as a simple algebraic variant
, we can work with @xmath1597 , an algebra over the commutative ring @xmath222 .
[ symp coh ] for any commutative ring @xmath222 , there is an isomorphism of graded @xmath222-modules @xmath1598/(\gamma_1\gamma_2,\beta(\gamma_1-\gamma_2))\to sh^\bullet(t_0;\bbk),\ ] ] canonical after a choice of basis of @xmath1515 , where @xmath1599 and @xmath1600 .
moreover , @xmath1601 . when @xmath222 is a field , the algebra on the right is isomorphic to @xmath1602 by theorem [ hh nodal ] .
we will see presently that @xmath29 respects products .
we begin with a comment about grading .
the grading of @xmath1603 depends , _ a priori _ , on the choice of line field @xmath31 . if @xmath1604 is another choice , obtained from @xmath31 by a map @xmath1605 representing a class @xmath1606 , where @xmath1607 , the degrees of generators , which are 1-periodic hamiltonian orbits @xmath643 , change according to the formula @xmath1608\rangle$ ] ( we do not bother with the sign ) . in the cochain complex for @xmath1609
we shall describe , all orbits are null - homologous , and hence the grading of @xmath1609 is independent of @xmath31 .
the graded @xmath222-module @xmath1603 is described in ( * ? ? ?
one uses an autonomous hamiltonian which is a perfect morse function @xmath1610 , accelerating appropriately on the cylindrical end .
one has a natural map of algebras @xmath1611 which is an isomorphism onto @xmath1612 ; this is the contribution of the minimum @xmath1613 and two saddle - points @xmath1614 and @xmath1615 of @xmath1610 .
we choose @xmath1610 so that @xmath1616,[s_2])$ ] is the chosen basis for the morse homology @xmath1515 .
define @xmath1617 , @xmath1618 and @xmath1619 . for each @xmath1620
, there is a reeb orbit @xmath1621 which winds @xmath131 times around the puncture ; @xmath1622 contributes classes @xmath1623 and @xmath1624 which span those respective modules .
we define @xmath1625 and @xmath1626 .
we have @xmath1627\cdot [ s_2]=0 $ ] , because @xmath1628 is not contractible .
the exact fukaya category of a liouville manifold @xmath83 is tied to its symplectic cohomology via the ` closed - open string map ' @xcite to its hochschild cohomology @xmath1629 for the details of the construction of @xmath1630 we refer to s. ganatra s doctoral thesis @xcite .
@xmath1631 is a homomorphism of @xmath222-algebras : it intertwines the pair - of - pants product on @xmath1632 with the cup product on @xmath1633 .
this part of the paper pinpoints the weierstrass curve @xmath1634 such that the minimal @xmath13-structure @xmath1635 it induces on the algebra @xmath200 is gauge - equivalent to the @xmath13-structure @xmath218 obtained from the fukaya category @xmath32 .
theorem [ ainf comparison ] assures us that @xmath216 exists , and is unique as an abstract weierstrass curve .
our aim is to show that @xmath1636 .
this will be accomplished in the final section of the paper by an argument involving @xmath29-functions . in this section
we offer two alternative proofs that @xmath1637 .
one is by eliminating all possibilities other than @xmath156 ; the other is by a calculation of ` seidel s mirror map ' .
we also prove our mirror - symmetry theorem for the wrapped category .
lemma [ tate at q=0 ] characterized the weierstrass curve @xmath1638 as having a section which is a node at @xmath155 , for any @xmath1639 .
symplectic topology now enters the picture : [ elimination ] take @xmath1639 , and @xmath1640 a weierstrass curve ( here @xmath1641 ) .
suppose that the exact category @xmath1642 , taken with coefficients in @xmath1643 , is quasi - equivalent to @xmath1644 .
then @xmath162 is nodal .
let @xmath1645 .
one has viterbo s map @xmath1646 , a graded algebra homomorphism @xcite , and restriction homomorphisms @xmath1647 ( everything over @xmath222 ) , one for each lagrangian brane @xmath1648 .
the hochschild cochain complex @xmath1649 for an @xmath13-algebra @xcite has a filtration @xmath1650 by the length of cochains ( so @xmath1651 ) , and so for each @xmath86 one has a quotient map @xmath1652 now , @xmath1653 , ordinary cohomology , by a canonical isomorphism ( * ? ? ? * ( 8c ) ) .
it follows easily from the definitions , plus the gluing theorem for hamiltonian floer theory of which @xcite has a meticulous account that the composite of the maps @xmath1654 is the classical restriction map @xmath1655a _ surjective _ map .
taking the sum of these composite maps for the objects @xmath50 and @xmath53 produces an _ isomorphism _ @xmath1656 we note one more feature of the composite @xmath1657 , which is that it maps @xmath1658 to @xmath1659 , the part spanned by cocycles in @xmath1660 .
indeed , the length - zero component of @xmath1657 returns , for each lagrangian brane @xmath86 , a count of index - zero pseudo - holomorphic discs attached to @xmath86 , with one marked boundary point , of which there are none by exactness ( constant discs have index @xmath180 ) . on the algebro - geometric side ,
suppose that @xmath1661 is a weierstrass curve mirror to @xmath30 over @xmath222 .
by lemma [ perf computes hochschild ] , one has @xmath1662 .
hochschild cohomology is invariant under passing to @xmath1663 , by a form of morita invariance @xcite .
hence , under the hypotheses of the proposition , one has @xmath1664 .
consequently , using ( [ sum of composites ] ) we obtain a map of @xmath222-algebras @xmath1665 such that the composite @xmath1666 is an isomorphism of @xmath222-modules .
we assert that such a homomorphism exists only if @xmath162 is nodal . to prove this
, we must eliminate the cuspidal and smooth cases .
we claim that if @xmath162 were smooth , one would have @xmath1667 $ ] , where @xmath1668 .
additively , this follows the degeneration of the hodge spectral sequence ( [ hodge ] ) .
the spectral sequence is multiplicative , and so the @xmath222-algebra @xmath1669 is isomorphic to the associated graded algebra @xmath1670 .
we have @xmath1671 , since this is even true in the associated graded algebra ; this establishes the claim .
any homomorphism of graded unital @xmath222-algebras @xmath1672\ ] ] obeys , in the notation of lemma [ symp coh ] , @xmath1673 , and therefore @xmath29 fails to surject onto the 1-dimensional part @xmath1674 .
the composite of @xmath29 with the maps to @xmath1675 and @xmath1676 can not then both be surjective .
we can rule out the possibility that @xmath162 is cuspidal , i.e. , that @xmath1677 , by noting that @xmath1678 . this module transforms under the @xmath382-action on @xmath215 , and it has weight @xmath1679 .
the restriction map @xmath1680 respects the weight of the @xmath382-action that is , it is a map of graded vector spaces and hence its restriction to @xmath1681 is zero .
there are other methods for ruling out @xmath215 : * when @xmath639 , the map that @xmath1681 is 1-dimensional , and therefore the composite map @xmath1682 unavoidably has a kernel . * in @xcite , we used abouzaid s model @xcite for the fukaya category of a plumbing to describe the structure maps of @xmath23 and thereby prove non - formality , assuming @xmath639 . *
when @xmath821 , the gerstenhaber bracket on @xmath912 is non - zero by theorem [ bracket ] .
for any liouville domain @xmath83 , the bracket is zero on the image of viterbo s map @xmath1683 .
indeed , @xmath843 is a ring homomorphism , and it is easy to see that @xmath1684 , where @xmath1037 is the bv operator on symplectic cohomology @xcite , and the bracket is , in accordance with the rules of bv algebras , @xmath1685=(-1)^{|x|}\delta(x\cdot y)- x\cdot\delta y - ( -1)^{|x| } ( \delta x ) \cdot y , \ ] ] which implies that @xmath1686 = 0 $ ] .
moreover , _
@xmath1630 preserves gerstenhaber brackets _ :
this was first stated by seidel @xcite , but there is no published proof . in characteristic 2
we have verified it for ourselves using standard gluing methods . over other fields the signs are tricky ,
so we regard it as conjectural .
however , taking this assertion for granted , the composite map @xmath1687 preserves brackets and so can not be injective .
we want to construct an @xmath13-functor @xmath1688 we already have an isomorphism @xmath1689 , valid for any weierstrass curve @xmath162 defined over @xmath1 .
theorem [ ainf comparison ] implies that there is a unique @xmath162 for which @xmath1690 lifts to a quasi - isomorphism @xmath1691 for that particular @xmath162 , @xmath49 extends naturally to a quasi - isomorphism @xmath1692 since the inclusion maps @xmath1693 and @xmath1694 are quasi - equivalences , one obtains a quasi - isomorphism @xmath1695 composing this with a quasi - inverse to the inclusion @xmath1696 , one obtains a quasi - isomorphism @xmath1697 whose restriction to @xmath1556 is the functor we want .
our task , then , is to identify the mirror weierstrass curve @xmath162 . in light of prop .
[ elimination ] , it must be a curve which is nodal over @xmath1698 . _ a priori _ ,
the node is only a @xmath1699-point .
however , the normalization of @xmath162 , defined over @xmath1698 , has two points ( over @xmath1699 ) which map to the node , and since the normalization is a rational curve , these points are actually defined over @xmath1698 . hence the same is true of the node . by clearing denominators , we obtain integer coordinates for the node .
it then defines a section of @xmath157 , which by the proposition must map to a node over @xmath1643 for every prime @xmath155 . by lemma
[ tate at q=0 ] , @xmath162 is therefore equivalent to @xmath25 . we have proved a weakened form of theorem [ mainth ] clause ( iii ) : we have shown that @xmath1700 is a quasi - equivalence , while the theorem claims that @xmath1701 is already an equivalence .
we formulate this step as a separate statement , prop .
[ ex fuk split closed ] , whose proof completes that of the theorem .
[ ex fuk split closed ] @xmath1702 is a quasi - equivalence .
the triangulated @xmath13-category @xmath929 is split - closed .
let @xmath1703 denote the image of @xmath1529 under @xmath1704 .
we must show that the inclusion @xmath1705 is a quasi - equivalence . by thomason s theorem @xcite , it is sufficient to prove equality of grothendieck groups : @xmath1706 . of a triangulated @xmath13-category @xmath223 , we mean @xmath1707 , the grothendieck group of the classical triangulated category @xmath1708in this case , @xmath541 .
] consider the map @xmath1709 which maps an exact lagrangian brane @xmath86 to @xmath179 if and only if the double covering @xmath1710 ( part of the brane structure ) is trivial .
this map descends to a homomorphism @xmath1711 .
this is an adaptation of a result of abouzaid ( * ? ? ?
one views a spin - structure , lifting a given orientation , as a local system for the group @xmath1712 .
the switch from closed , higher - genus surfaces to @xmath30 is irrelevant .
we must show that @xmath1713 is onto .
we have @xmath1714 , and by lemma [ k theory ] , @xmath1715 is an isomorphism .
one also has an isomorphism @xmath1716 , where @xmath1717 is the degree and @xmath380 describes the ` descent data ' under normalization , as at the end of the proof of that lemma .
the image of @xmath1718 contains @xmath52 ( rank 1 , degree 0 ) and @xmath1018 ( rank 1 , degree 1 ) .
hence @xmath1719 is onto .
now take some @xmath1720 , and let @xmath1501 be the same object with the other double covering .
then the class @xmath1721-[l^\star]$ ] is 2-torsion in @xmath1722 , by the last lemma and the fact that the change of covering can be accomplished by an involution of @xmath1529 ( namely , tensoring spin - structures on lagrangians by the restrictions of some real line - bundle @xmath1723 ) .
thus @xmath1724-[\psi_0(l')]$ ] is again 2-torsion in @xmath1725 .
there is a unique 2-torsion class in @xmath1725 , detected by @xmath380 .
therefore @xmath1726 is surjective , and hence @xmath1727 is surjective .
there is an alternative to the argument just given which does not appeal to abouzaid s analysis , but instead observes that the objects @xmath50 and @xmath1728 ( which differ only in their double coverings ) map under @xmath49 to perfect complexes whoses @xmath243-classes differ by the generator of @xmath1729 . for this , we regard @xmath1730 as @xmath50 with a local system with fiber @xmath1 and holonomy @xmath180 .
we have @xmath1731 , and it follows from an easy adaptation of lemma [ where local systems go ] below that @xmath1732 , where @xmath1733 is the 2-torsion section of @xmath1734 and , as usual , @xmath335 is the identity section . from this point , the argument is straightforward .
[ [ the - affine - coordinate - ring . ] ] the affine coordinate ring .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + suppose given an abstract weierstrass curve @xmath1735 over @xmath1736 .
there is then a weierstrass cubic embedding carrying @xmath335 to @xmath161 $ ] ; the affine complement to the closure of @xmath1737 is @xmath1738 , where @xmath1739 , the affine coordinate ring , is the ring of functions on @xmath162 with poles only at @xmath335 : @xmath1740 we want to compute @xmath1741 by determining an equation that holds in @xmath1742 .
( this is almost the method of @xcite , but we are concerned here with the affine rather than homogeneous coordinate ring . ) we shall in fact carry out this computation only for @xmath1743 . since @xmath23 split - generates @xmath32 , the quasi - isomorphism @xmath1744 extends to an @xmath13-functor @xmath1745 .
consider the object @xmath53 .
the dehn twist @xmath1746 , acting as an autoequivalence of @xmath1556 , is homotopic to the twist functor along the spherical object @xmath53 : this is elementary in the present case @xcite , but is an instance of a general result of seidel s @xcite .
now , @xmath1747 .
the twist along the spherical object @xmath1748 is homotopic to the functor @xmath1749 of tensoring with @xmath1018 ( see ( * ? ? ? * ( 3.11 ) ) ; the argument is carried out over fields , but over @xmath21 , seidel
thomas s map @xmath330 must be a unit times the restriction map by base - changing to reduce to the case of fields ) .
thus @xmath49 induces an isomorphism @xmath1750 , where @xmath1751 the ` dehn twist ring ' @xmath1752 , needs explanation neither the direct system , not the ring structure , is obvious .
the maps @xmath1753 which form the direct system are defined via holomorphic sections of a lefschetz fibration over a strip ; this interpretation is part of seidel s analysis of dehn twists @xcite .
the ring structure is easier : the @xmath1613th power of the dehn twist defines a map @xmath1754 which applies the dehn twist to the intersection points between lagrangians .
the product in the ring is given by composing this with the triangle product @xmath1755 , @xmath1756 associativity of this product is easily seen , as is the fact that the unit element @xmath1757 is a 2-sided unit for the multiplication in the dehn twist ring .
however , commutativity is something that we learn from the isomorphism @xmath1758 .
if one knew the maps @xmath1759 explicitly , one would be able to proceed by perfect analogy with the algebro - geometric side of the mirror , as follows .
take the unit @xmath1757 and its images @xmath1760 .
take a basis @xmath1761 for @xmath1762 and a basis @xmath1763 for @xmath1764 .
identify the weierstrass equation as the unique relation satisfied by the seven monomials @xmath1765 .
we can extract nearly complete information about the maps @xmath1766 as follows .
we choose @xmath1504 and @xmath1315 to have just one , transverse intersection point .
then @xmath1767 consists of a single point @xmath1768 , and @xmath1769 .
hence @xmath1770 for some @xmath1771 . from the isomorphism @xmath1772
, we see that @xmath1773 , and hence that @xmath1774 .
moreover , @xmath1775 . in practice ,
then , one obtains a cubic equation by picking @xmath1776 so as to make @xmath1777 a basis for @xmath1762 and @xmath1778 so as to make @xmath1779 a basis for @xmath1764 .
one computes the products @xmath1780 and identifies the unique ( up to scale ) relation that they satisfy .
this relation is necessarily of form @xmath1781 hence it determines @xmath330 .
now let @xmath1782 and @xmath1783 .
then one has @xmath1784
, i.e. , these coordinates satisfy the weierstrass equation .
we learned something interesting _ en route _ here , though we shall not pursue it : the series @xmath330 , which encodes information about sections of a lefschetz fibration , can be computed . to compute with the dehn twist ring , note that one can take for @xmath1002 any compactly supported exact automorphism of @xmath30 that is isotopic to a dehn twist along @xmath1315 . to obtain a convenient model ,
start with the linear symplectomorphism of @xmath1785 given by @xmath1786 = [ x_1 , x_2 - x_1].\ ] ] the fixed - point set of @xmath295 is the line @xmath1787 .
we will take our basepoint to be @xmath1788 where @xmath1789 .
one has @xmath1790 ; let @xmath5 be the @xmath1791-neighborhood of the line segment @xmath1792 $ ] .
take @xmath1793 to be a symplectomorphism such that @xmath1794 ; extend @xmath380 to a symplectomorphism of @xmath8 , still called @xmath380 , which is trivial outside @xmath5 .
let @xmath1795 .
then @xmath1796 ; by adjusting @xmath380 , we may assume that @xmath1002 acts as the identity in some neighborhood of @xmath9 .
thus @xmath1002 restricts to give @xmath1797 .
let @xmath1798 \in t : nx_1 + x_2 = 0\}$ ] . when @xmath1799 , we have that @xmath1800 .
there is a primitive @xmath29 for @xmath1801 making @xmath1504 an exact curve and @xmath1802 an exact symplectomorphism .
we must exhibit a primitive @xmath29 for @xmath26 such that @xmath1803=0 \in h^1_c(t_0;\r)$ ] and @xmath1804 .
for the first requirement , it suffices to show that @xmath1805 for curves @xmath806 forming a basis for @xmath1806 .
such a basis is given by @xmath1807 , with chosen orientations for these two curves .
for any primitive @xmath29 for @xmath26 , one has @xmath1808 .
it suffices to choose @xmath29 , therefore , in such a way that @xmath1504 and @xmath1809 are both exact lagrangians .
it is easy to find a 1-form @xmath1810 on @xmath30 such that @xmath1811 on a small regular neighborhood @xmath747 of @xmath1812 and such that @xmath1813 .
now , @xmath1814 is a circle , isotopic to a loop encircling @xmath9 .
there is therefore no obstruction to finding a 1-form @xmath1815 , supported outside @xmath1812 , such that @xmath1816 ; then @xmath1817 is the required primitive .
[ primitive ] for any natural number @xmath747 , one can choose a primitive @xmath29 for @xmath1801 such that the curves @xmath1818 are exact for @xmath1819 .
choose @xmath1820 , and take @xmath29 as in the lemma .
since @xmath1504 is exact , so too is @xmath1821 for any @xmath1531 .
but for @xmath1822 we have @xmath1823 , use a @xmath29 as in the proposition , taking @xmath747 at least @xmath1824 .
we have @xmath1825 ; \\ l_0 \cap l_{(1,-2 ) } & = \ { \zeta_0 , \zeta_1\ } , & & \zeta_k = [ k/2,0 ] ; \\ l_0 \cap l_{(1,-3 ) } & = \
{ \eta_0 , \eta_1,\eta_2 \ } , & & \eta_k = [ k/3,0 ] .
\end{aligned}\ ] ] in calculating products in the dehn twist ring @xmath1752 , immersed triangles in @xmath30 count with sign @xmath1679 . to avoid repetition
, we do not give the argument here but defer it to section [ homog subsection ] . with this understood ,
it is straightforward to calculate that in the dehn twist ring of @xmath1478 , one has @xmath1826 we put @xmath1827 ; then @xmath1828 is a @xmath1-basis for @xmath1829 .
next , we compute @xmath1830 we deduce that @xmath1831 we put @xmath1832 , and note that @xmath1833 is a basis for @xmath1834 .
further computations yield @xmath1835 where @xmath1836 are the intersection points @xmath1837\in l_0\cap l_{-6}$ ] .
these relations imply that @xmath1838 putting @xmath1839 , @xmath183 and @xmath1840 , we obtain the weierstrass relation in the desired form @xmath1841
we restate clause ( iv ) of theorem [ mainth ] , in slightly refined form .
the category @xmath929 is a dg enhancement for @xmath1842 .
enlarge it to any dg enhancement @xmath1843 of the bounded derived category @xmath1844 ( over @xmath1 ) .
that is , @xmath1845 is a dg category containing @xmath929 as a full subcategory , with an equivalence of triangulated categories @xmath1846 extending the canonical equivalence @xmath1847 .
a standard method to construct such an enhancement would be to use injective resolutions for coherent sheaves , and then to exhibit equivalence of that approach to the approach via ech complexes in the case of perfect complexes by the method of ( * ? ? ?
* lemma 5.1 ) . however , there is no requirement for the enlargement to be of geometric origin .
[ psi wrap ] there is a @xmath1-linear @xmath13-functor @xmath1848 which extends to a quasi - equivalence @xmath1849 and which restricts to @xmath1850
. it will be helpful to make the quasi - equivalence @xmath1718 more explicitly by describing its effect on lagrangians with finite - rank local systems ; these may be regarded as twisted complexes in @xmath1492 . as usual , let @xmath1851 be a lagrangian brane of slope @xmath1852 with non - trivial double covering . for any @xmath1853 , let @xmath1854 denote the brane @xmath1315 equipped with a rank 1 @xmath222-local system of holonomy @xmath1855 .
this is an object in a larger fukaya category @xmath1856 whose objects are exact lagrangian branes with local systems of finite rank free @xmath1-modules . for a @xmath222-linear @xmath13-category @xmath223 ,
let @xmath1857 denote the category of finitely generated projective @xmath223-modules , assigning to each object a finite cochain complex of finitely generated projective @xmath222-modules . with @xmath1858 , the object @xmath1854 defines a left yoneda - module @xmath1859 .
let @xmath1860 be an @xmath13 functor quasi - inverse to @xmath1718 . over @xmath1 , the existence of such
a functor is not quite trivial .
however , our earlier analysis of ech complexes implies that one can define a quasi - inverse ( or even strict inverse ) @xmath1861 to @xmath1862 .
we then define @xmath1863 by extending the latter functor to twisted complexes .
module categories are contravariant , and so @xmath1863 induces a functor @xmath1864 [ where local systems go ] work over a base ring @xmath201 which is a commutative , unital , normal , noetherian domain .
identify the normalization of @xmath156 with @xmath1865 by sending the preimages of the nodal section of @xmath156 to @xmath1866 , [ 1:0]\}$ ] and @xmath335 to @xmath193 $ ] .
two such identifications exist , of which one has the property that for each @xmath1867 , the module @xmath1868 is represented by a locally free resolution of the skyscraper sheaf @xmath1869 at the section @xmath1870 \colon \spec r\to \mathbb{p}^1(r)$ ] .
there is , for each @xmath1867 , a rank 1 @xmath222-local system @xmath1871 over @xmath30 for which @xmath1872 and @xmath1873 .
moreover , @xmath1874 . the local system
@xmath1871 induces a strict autoequivalence @xmath1875 of @xmath1856 : on objects : leave the lagrangian brane unchanged but tensor the local system by the restriction of @xmath1871 . on morphism - spaces , for each intersection point @xmath1876 , map @xmath637 to @xmath1877 , where @xmath1878 is the isomorphism that sends @xmath357 to @xmath1879 .
define @xmath1880 , the full subcategory of @xmath1856 on the two objects @xmath1504 with its trivial local system and @xmath1881 = @xmath1315 with its local system of holonomy @xmath1610 .
then @xmath1882 is a minimal @xmath13-structure .
moreover , @xmath1875 induces a trace - preserving isomorphism @xmath1883 . by theorem [ ainf comparison ] , the @xmath13-structure @xmath1882 is gauge - equivalent to @xmath196 for a unique weierstrass curve @xmath162 . as an abstract curve , we have @xmath1884 , but the weierstrass data ( basepoint , differential ) might not be standard .
we can think of @xmath162 as @xmath156 with standard differential , but different basepoint @xmath1885 . by theorem [ ainf comparison ] ,
the isomorphism @xmath1886 arises from a weierstrass isomorphism @xmath1887 .
automorphisms of @xmath1888 are the same thing as automorphisms of @xmath538 that map @xmath1889 to @xmath1890 ; thus they form a group @xmath1891 , where @xmath1892 acts as the antipodal involution .
our construction gives rise to a homomorphism @xmath1893 mapping @xmath1610 to the automorphism @xmath1894 such that @xmath1895 is represented by @xmath1896 .
the homomorphisms @xmath1897 are by construction compatible with base change @xmath1898 .
we claim that they must map @xmath550 to @xmath550 . indeed ,
since @xmath550 is normal in @xmath1899 , there is a quotient map @xmath1900 , also natural in @xmath201 .
one must have @xmath1901 when @xmath438 is an algebraically closed field , since then every @xmath1902 is a square . by naturality , @xmath1901 for arbitrary fields @xmath438 ( embed @xmath438 into an algebraic closure ) , and so for arbitrary domains @xmath201 of the sort specified in the statement ( embed @xmath201 into its field of fractions ) . in view of their compatibility with base change , we view the @xmath1897 collectively as a natural transformation @xmath1903 where @xmath1904 is the category of rings satisfying the conditions listed in the statement , and @xmath1905 is the multiplicative group functor : @xmath1906 .
now , @xmath528 is co - represented by @xmath1907 $ ] , meaning that @xmath1908the identification makes the set @xmath1909 into a group .
the isomorphism sends @xmath1910 to the homomorphism @xmath1911 , @xmath1912 .
we note that @xmath260 is indeed a noetherian normal domain ! by yoneda s lemma , @xmath51 must arise from some ring endomorphism @xmath1913 .
moreover , @xmath1914 .
these endomorphisms give rise to the set - theoretic natural transformations @xmath1915 , of which @xmath1916 is a group homomorphism but @xmath1917 is not .
we note also that @xmath1897 must be injective for each @xmath201 , which leaves us only with the two possibilities @xmath1918 .
one of these possibilities is the right one , the other not ; this is the ambiguity left in the statement of the lemma .
the imprecision in the previous lemma is easily resolved . to fix an identification of the normalization of @xmath156 with @xmath538 of the sort described
, it suffices to describe its effect on the tangent space to @xmath156 at @xmath335 .
now , @xmath1919 canonically , via the 1-form @xmath26 ; and @xmath1920}\mathbb{p}^1 = t_{1}\mathbb{a}^1 = r$ ] . the correct identification is the one given in these terms by @xmath1921 . that is because @xmath1922 can be understood as @xmath1923 , which is identified by @xmath49 with @xmath1924 .
the latter identification is the one determined by the trace - maps . the material in this subsection is not used elsewhere in the paper .
in lemma [ where local systems go ] , we established that over a base ring @xmath201 , the functor @xmath1925 maps @xmath1881 , that is , @xmath53 with a rank 1 local system of holonomy @xmath1926 , to the skyscraper sheaf located at the section @xmath1610 of the smooth locus @xmath1927 = \mathbb{g}_{\mathsf{m}}(r)$ ] . over @xmath1 , rank 1 local systems are not very interesting , higher rank local systems more so . working over @xmath1 , take a local system on @xmath53 with fiber @xmath1928 and holonomy @xmath1929 ; denote this object by @xmath1930 .
it maps under @xmath1931 to a module for @xmath1932 co - represented by some quasi - coherent complex @xmath1933 .
that is , @xmath1934 .
let @xmath435 be the smooth locus in @xmath156 ; thus @xmath1935 \cong \mathbb{g}_{\mathsf{m}}(\z)$ ] .
for @xmath1929 , let @xmath1936 denote the @xmath1937$]-module @xmath1928 , on which @xmath1553 acts as @xmath94 .
let @xmath1938 be the associated quasi - coherent sheaf on @xmath435 , and let @xmath1939 be the push - forward of @xmath1938 to @xmath156 .
then @xmath1933 is quasi - isomorphic to the sheaf @xmath1939 .
notice that @xmath1940 is quasi - isomorphic in @xmath1941 to an object of @xmath1942 . as such ,
it is a _ compact _ object of the dg category of @xmath23-modules ( see for instance @xcite ) .
quasi - equivalences preserve compact objects ; consequently @xmath1943 is compact as an object of @xmath1944 .
therefore @xmath1943 is quasi - isomorphic to a perfect complex @xcite . since only its quasi - isomorphism class matters
, we may assume that @xmath1943 is a strictly perfect complex a finite complex of locally free sheaves .
it will be helpful to work over base _
fields _ @xmath222 .
we then take the object @xmath1945 associated with @xmath1946 . _ step 1_. work over an algebraically closed field @xmath222 .
we claim that @xmath1943 is then quasi - isomorphic to a sheaf supported whose support is contained in the eigenvalue spectrum @xmath1947 in @xmath1948= v(\bbk)= v\times_{\spec \z } \spec \bbk$ ] .
if @xmath1949 , we have a hyperext spectral sequence @xmath1950 if @xmath1951 then ( only ) @xmath1952 and @xmath1953 are non - zero .
if @xmath1954 then @xmath1955 .
one has @xmath1956 , where @xmath1957 is the inclusion .
since @xmath435 is the spectrum of a regular local ring of dimension 1 , the @xmath1958-modules vanish except for @xmath1959 . since it is supported in two adjacent columns , the spectral sequence degenerates at @xmath677 .
therefore @xmath1960 except when @xmath1961 and @xmath1962 .
hence @xmath1963 for @xmath1964 . next ,
consider the hyperext spectral sequence @xmath1965 the cohomology sheaf @xmath1966 is supported on the singular section together with @xmath1967 .
one has @xmath1968 which is zero for @xmath1969 because @xmath1966 has affine support .
thus @xmath1970 except when @xmath1971 .
so the spectral sequence degenerates , and we see that @xmath1972 for @xmath1973
. we may therefore truncate the complex @xmath1943 , replacing it by @xmath179th cohomology @xmath1974 , to which it is quasi - isomorphic .
further , @xmath1974 is a torsion sheaf since its stalks are generically zero .
notice that if we have a short exact sequence of modules @xmath1936 , the corresponding sheaves @xmath1974 form a long exact triangle . from that , and the fact that over the algebraically closed field @xmath222 , matrices are conjugate to upper triangular matrices , we see that @xmath1974 is supported in @xmath1975 , and therefore in @xmath1976 . _
_ over an arbitrary field @xmath222 , @xmath1943 is quasi - isomorphic to a torsion sheaf @xmath1977 , supported in @xmath1975 , such that @xmath1978 canonically . indeed , by ( flat )
base change from @xmath222 to its algebraic closure , we see that the @xmath70th cohomology sheaf of @xmath1943 vanishes for each @xmath1973 . by truncation
we may replace the complex by its zeroth cohomology @xmath1979 .
moreover , @xmath313 vanishes at the generic point , and its stalk at the singular section is zero , so it torsion and supported in @xmath1980 . moreover , @xmath1981 via the spectral sequence @xmath1982 above .
_ over @xmath1 , @xmath1943 is quasi - isomorphic to a torsion sheaf supported in @xmath435 .
moreover , @xmath1983 canonically . since @xmath1943 is a perfect complex , its cohomology sheaves are coherent .
hence @xmath1984 is a finitely generated @xmath1-module for each @xmath70 . by step 2 and the compatibility of the construction with base change
, one has @xmath1985 for every field @xmath222 and every @xmath1973 .
hence @xmath1986 for @xmath1973 . on the other hand
, @xmath1987 is a coherent sheaf on @xmath3 , for each @xmath70 ; or in other words , it is a finitely generated abelian group @xmath1988 . by the projection formula , and step 2 , @xmath1989 for any field @xmath222
. therefore @xmath1990 .
consequently @xmath1991 is supported in @xmath435 , and so is the module associated with its sections @xmath1984 , which is @xmath179 if @xmath1973 .
hence @xmath1943 is quasi - isomorphic to its @xmath179th cohomology sheaf .
one has @xmath1992 , again via the spectral sequence @xmath1993 . _
step 4_. completion of the proof .
since it is torsion and supported in @xmath435 , the sheaf @xmath1974 is the push - forward of a coherent sheaf on @xmath435 .
we think of this as the sheaf associated with a module @xmath1994 . we stress that @xmath1994 is canonically identified with @xmath1995 as an @xmath1-module , so @xmath1996 for a well - defined matrix @xmath1997 .
moroever , the map @xmath1998 is compatible with conjugation of matrices ( i.e. @xmath1999 ) .
these points apply over any @xmath1-algebra @xmath222 , and the construction is compatible with base change .
hence the map @xmath2000 arises from a map of @xmath1-schemes @xmath2001 .
we claim that @xmath2002
. it will suffice to show that @xmath2003 when we base - change to @xmath1699 , an algebraic closure of the rationals .
the induced map @xmath2004 is the identity on the diagonal matrices , and therefore , by compatibility with conjugation , also on the zariski - open set of diagonalizable matrices .
therefore it is the identity map .
let @xmath2005 be an arc of slope @xmath1852 , graded so that @xmath2006 lies in degree @xmath179 , and oriented so that it runs into @xmath9 .
our functor @xmath2007 will carry @xmath1565 to @xmath2008 , the skyscraper sheaf along the nodal section .
@xmath1555 is generated by @xmath1556 and @xmath1565 .
first , if orient @xmath2009 in the opposite direction , we obtain an isomorphic object of @xmath1555 .
this reflects the fact that a spin - structure on a lagrangian ( which trivializes @xmath2010 and @xmath2011 ) is more data than is needed ; a pin - structure , trivializing @xmath2011 , is sufficient @xcite .
changing the orientation corresponds to an automorphism @xmath2012 of the object .
the brane structure also involves a double covering of @xmath2009 , but that is necessarily trivial .
with these points noted , we find that any object of @xmath1555 whose lagrangian is non - compact is quasi - isomorphic to a shift of an iterated dehn twist of @xmath1565 along closed , exact curves equipped with non - trivial double coverings .
these dehn twists act on @xmath2013 by spherical twists @xcite ; hence the arcs are represented by twisted complexes in @xmath1556 and @xmath1565 .
@xmath2014 is generated by @xmath2015 and the skyscraper sheaf @xmath2008 along the nodal section @xmath70 .
it suffices to show that @xmath2016 is generated , as an abelian group , by the image of @xmath537 and the class @xmath2017 $ ] . here
@xmath2018 denotes the grothendieck group of coherent sheaves , while @xmath243 is the grothendieck group of vector bundles .
the quotient @xmath2019 is certainly generated by coherent sheaves supported on @xmath165 , the closure of @xmath2020 .
thus it will suffice to show that @xmath2021 , the grothendieck group of the abelian category of coherent sheaves supported along @xmath165 , is generated by the class of @xmath2008 .
we now proceed as in ( * ? ? ?
ii 6.3.4 ) : @xmath2022 is the abelian category of finitely - generated modules @xmath83 for @xmath562/(y^2+xy - x^3)$ ] such that @xmath2023 for some @xmath258 , where @xmath2024 .
such a module has a filtration @xmath2025 and therefore @xmath2021 is generated by the factors of such filtrations , i.e. , by modules @xmath747 with @xmath2026 . as sheaves ,
those are precisely the push - forwards of sheaves on @xmath165 , or equivalently those of the form @xmath2027 for some coherent sheaf @xmath313 on @xmath3 .
since @xmath2028 , the result follows .
we have a sequence of @xmath13-functors @xmath2029 here @xmath2030 is the ( covariant ) left yoneda functor , @xmath2031 .
the functor @xmath2032 is restriction , which is a quasi - equivalence because @xmath196 split - generates @xmath929 .
we denote by @xmath1944 a dg enhancement of the unbounded derived category of quasi - coherent complexes , as in @xcite ; the last map , @xmath2033 is again a left yoneda functor .
it follows from ( * ? ? ?
* theorem 8.9 ) or @xcite that @xmath2033 is a quasi - isomorphism .
indeed , the dg category of modules over @xmath1932 is , rather trivially , equivalent to the dg category of dg functors @xmath2034 $ ] .
there are further equivalences @xmath2034\to [ \qc(\eut_0 ) , \qc(\spec \z)]_c$ ] where the @xmath406 denotes that the functors respect filtered colimits ( think of quasi - coherent complexes as filtered colimits of perfect complexes ) ; and @xmath2035_c$ ] ( mapping an integral kernel to its push - pull functor ) .
the conclusion of this discussion is as follows : there is a homotopy - commutative diagram of @xmath13-functors @xmath2036 & \modules \eub_{\eut_0 } & \qc(\eut_0)\ar[l ] \\
\euf(t_0)^\exact\ar[u ] \ar[rr]^{\simeq } & & \tw \vb(\eut_0)\ar[u ] } \ ] ] we now define @xmath2037 to be an object of @xmath1944 corresponding under mirror symmetry to the arc @xmath1565 . precisely : define @xmath2037 to be a choice of quasi - coherent complex whose associated @xmath2038-module @xmath2039 is quasi - isomorphic to @xmath2040 .
we have the full subcategory @xmath2041 , and , by restricting the arrow @xmath2042 in the previous diagram , a functor - sequence @xmath2043 the arrow @xmath2044 is an embedding , and the composite @xmath2045 is well - defined .
similarly , we have quasi - embeddings @xmath2046 and a functor - sequence @xmath2047 whose effect on cohomology lifts to a functor @xmath2048 mapping @xmath2037 to @xmath1565 . the composite @xmath2049 is the identity functor on @xmath2050 ( this composite does not involve the potentially information - losing functor @xmath2051 ) . hence @xmath2052 is full
. a key point in the proof of mirror symmetry for the wrapped category will be the following assertion : [ where the arc goes ] one has @xmath2053 in @xmath1944 , where as before , @xmath2008 is the push - forward of @xmath2054 by the singular section @xmath2055 . taking the proposition for granted for the moment
, we explain how to complete the proof of theorem [ psi wrap ] .
first , it implies that @xmath2056 , where @xmath1843 is the dg category of bounded complexes with coherent cohomology .
we now have a functor sequence @xmath2057 now , one can set up the full subcategory @xmath2058 so as to be a minimal @xmath13-category .
( in particular , in the case of endomorphisms of @xmath1565 , one can easily draw a perturbation @xmath2059 of @xmath2009 so that they do not jointly bound any immersed bigons . )
hence there is a map @xmath2060 inducing the identity map on cohomology .
the full subcategory @xmath2061 has projective hom - spaces , and thus there is a map @xmath2062 inducing the identity on cohomology . using homological perturbation theory , we obtain a functor @xmath2063 such that @xmath2052 is the functor previously so - denoted .
we must show that @xmath2052 is faithful .
to do so , it will suffice to prove it on the hom - spaces @xmath2064 and @xmath2065 , where @xmath6 runs through a split - generating set : @xmath2066 or @xmath53 or @xmath2009 .
the graded @xmath1-modules @xmath2064 are in each case isomorphic to their mirrors @xmath2067 ; likewise @xmath2068 .
they are free abelian graded groups , of finite rank in each degree .
the map @xmath2069 is surjective and is therefore an isomorphism and the same goes for @xmath2065 .
this completes the proof of theorem [ psi wrap ] , modulo that of proposition [ where the arc goes ] . if we work over a base _ field _
@xmath222 , then the quasi - coherent complex @xmath2070 mirror to the arc is quasi - isomorphic to @xmath2008 , the skyscraper at the nodal point @xmath70 of the @xmath222-variety @xmath2071 .
let @xmath2072 be the @xmath438th cohomology sheaf of @xmath2070 .
in general , if @xmath2073 and @xmath2074 are complexes of quasi - coherent sheaves over a scheme @xmath6 , at least one of them bounded , one has a right half - plane spectral sequence arising from a filtration on a complex computing @xmath2075 , with @xmath2076 abutting to @xmath2077 .
in particular , we have a spectral sequence @xmath2078 where @xmath334 $ ] . since @xmath2052 is full and @xmath2079 , we have @xmath2080 .
this spectral sequence degenerates at @xmath677 . indeed , let @xmath2081 denote the module of global sections of @xmath2082 , where @xmath1513 is the open inclusion of the smooth locus @xmath2083 $ ] .
let @xmath2084 $ ] , a maximal ideal of @xmath2085 $ ] , and let @xmath200 denote the localization of @xmath2085 $ ] at @xmath2086 . then @xmath2087 . since @xmath200 is a regular ring of dimension 1 ,
these ext - modules vanish except for @xmath1971 or @xmath357 .
therefore the spectral sequence is concentrated in two adjacent columns , and so degenerates .
thus @xmath2088 , and so @xmath2089 .
consequently , @xmath2072 is for each @xmath438 supported in the open set @xmath2090/(y^2+xy - x^3)$ ] .
next consider the spectral sequence @xmath2091 since @xmath2092 , one has @xmath2093 for @xmath2094 , while the space @xmath2095 is at most 1-dimensional .
one has @xmath2096 , but this cohomology module vanishes for each @xmath2097 .
this spectral sequence therefore also degenerates , and so @xmath2098 vanishes for all @xmath1973 .
since @xmath2099 is quasi - coherent with affine support in @xmath2100 , it follows that @xmath2101 .
because of the vanishing of the non - zero cohomology sheaves , we have a diagram of quasi - isomorphisms @xmath2102 , where @xmath2103 denotes the truncation @xmath2104 .
therefore @xmath2070 is quasi - isomorphic to the sheaf @xmath2105 , with which we may replace it .
moreover , @xmath2105 likewise has affine support in @xmath2100 , and so it is either zero , or is the push - forward from @xmath2100 of the sheaf associated with the @xmath412/(y^2+xy - x^3)$]-module @xmath222 ( with some action of @xmath637 and @xmath643 ) .
thus @xmath2105 is the skyscraper sheaf @xmath2106 at a closed point @xmath637 of @xmath2071 .
now , if @xmath637 is a regular point , then we have by lemma [ where local systems go ] @xmath2107 for some @xmath1855 , and thus @xmath2052 maps @xmath1565 and @xmath1881 to isomorphic objects of @xmath2108 .
it is easy to check that the yoneda functor @xmath2109 reflects isomorphism , and sends no object to the zero object . since it is a composite of that yoneda functor and an embedding
, @xmath2052 again reflects isomorphism .
thus it can not map @xmath1881 and @xmath1565 to isomorphic objects ; nor can it map @xmath1565 to the zero - object .
so @xmath637 is the unique singular point @xmath70 .
again , let @xmath2110 be the restriction of @xmath2111 to the smooth locus @xmath1957 .
one has @xmath2112 where @xmath2113 $ ] , an identification under which the section @xmath335 is defined by the homomorphism @xmath2114 , @xmath2115 .
let @xmath2116 be the @xmath201-module of sections @xmath2117 . since @xmath2111 is quasi - coherent
, @xmath2110 is the sheaf associated with @xmath2116 .
let @xmath2118 , and let @xmath2119 be the localization at @xmath2086 .
our task is to show that the @xmath2119-module @xmath2120 is zero .
we observe that @xmath2121 .
for this , consider the wrapped fukaya category with @xmath1698-coefficients ; the arc now maps to @xmath2122 .
one has @xmath2123 .
also , @xmath2124 .
thus , @xmath2125 is the sheaf associated with its module of sections @xmath2126 over @xmath2127 .
the previous lemma then implies that @xmath2128 .
now , @xmath2129 , where @xmath2130 ; in particular , @xmath260 contains every prime of @xmath1 , and hence @xmath2119 is a @xmath1698-algebra .
thus @xmath2131 is a @xmath1698-algebra , yet is torsion as an abelian group .
hence @xmath2131 .
we now proceed on the same lines as the argument for the previous lemma .
we now know that @xmath2111 is supported in @xmath2132/(y^2+xy - x^3)$ ] .
we deduce , just as before , that @xmath2133 for @xmath2094 , and hence that @xmath2037 is quasi - isomorphic to the cohomology sheaf @xmath2134 ; and that @xmath2134 corresponds to the @xmath2135-module @xmath1 , with some action of @xmath637 and @xmath643 ; but @xmath637 and @xmath643 necessarily act as zero , because otherwise @xmath2134 would be the image under @xmath2052 of a skyscraper sheaf at a non - singular section , contradicting the fact that @xmath2052 reflects isomorphism
. therefore @xmath2136 .
in this section , we offer a third proof that @xmath1743 has the equation @xmath2 .
more significantly , this proof extends to show that @xmath216 is the tate curve .
the tate curve was constructed by raynaud using formal schemes ; we have followed the expositions by deligne
rapoport @xcite and conrad @xcite , and gross s reinterpretation of the construction in toric language @xcite .
we review the construction .
the convex hull of @xmath2140 and @xmath2141 is a cone @xmath2142 ; the collection of cones @xmath2143 , their boundary faces @xmath2144 , and _ their _ common endpoint @xmath2145 , form a rational fan @xmath2146 in @xmath2147 . each cone @xmath406 of @xmath2146 has a dual cone @xmath2148 ( the use of inner products rather than dual spaces is an aid to visualization ) . to be explicit , @xmath2149 is the half - plane @xmath2150 , and @xmath2151 .
the dual cones @xmath2152 have semigroup @xmath201-algebras @xmath2153 $ ] spanned by monomials @xmath2154 where @xmath2155 .
we glue together the affine schemes @xmath2156 $ ] so as to form a toric scheme @xmath2157 with an affine open cover @xmath2158 . for each @xmath114
, there is a map of cones @xmath2159 , @xmath2160 .
the dual to this map is the map of cones @xmath2161 given by @xmath2162 , which induces a map of semigroup @xmath201-algebras @xmath2163\to r[\sigma^\vee_{i+1/2}\cap \z^2]$ ] and hence a function @xmath2164 .
the maps @xmath2159 assemble to form a map of fans @xmath2165 , whence the functions @xmath2166 consistently define a morphism @xmath2167 $ ] .
_ this morphism makes @xmath2137 an @xmath2168$]-scheme .
_ we can regard @xmath1553 as a regular function on @xmath2137 ; its restriction to @xmath2169 is the monomial @xmath2170 .
we now describe the gluing construction of @xmath2137 in scheme - theoretic terms , without the toric language .
the exposition follows deligne
rapoport s ( _ op .
_ ) ; we recall it for convenience .
the lattice points @xmath2171 are just the @xmath2172-linear combinations of @xmath2173 and @xmath2174 . in @xmath2175
$ ] , let @xmath2176 and @xmath2177 . then @xmath2178 [ x_{i } , y_{i+1}]}{(x_i y_{i+1 } - t)}.\ ] ] now let @xmath2179 as subsets of @xmath2137 . in abstract terms , we have @xmath2180 \stackrel{x_i\mapsto x_i}{\cong } \spec r[t ] [ x_i , x_i^{-1 } ] \\ & v_i \subset u_{i-1/2 } , & & v_i = u_{i-1/2}[y_i^{-1 } ] \stackrel{y_i\mapsto y_i}{\cong } \spec r[t ] [ y_i^{-1 } , y_i ] , \end{aligned}\ ] ] and these two descriptions of @xmath2181 are matched up by putting @xmath2182 .
the scheme obtained from the union of the open sets @xmath2169 by gluing @xmath2183 to @xmath2169 is @xmath2184 .
no additional gluing is required because when @xmath2185 one has @xmath2186 .
in the scheme @xmath2187\to \spec r[t , t^{-1}]$ ] , all the open sets @xmath2188 $ ] are identified with one another .
thus one has isomorphisms @xmath2187 \cong u_{1/2}[t^{-1 } ] \cong \mathbb{g}_{\mathsf{m}}(r[t , t^{-1}])$ ] .
the fiber @xmath2189 } r$ ] is the union of the subsets @xmath2190/(x_iy_{i+1}).\ ] ] the sets @xmath2191 are disjoint when the indices @xmath114 and @xmath1513 are not adjacent ( @xmath2192 ) .
moreover , @xmath2193 is glued to @xmath2191 along @xmath2194=r[y_i , y_i^{-1}]$ ] .
thus @xmath2195 is an infinite chain of @xmath538 s , say @xmath2196 .
the group @xmath1 acts on @xmath2137 covering the trivial action on @xmath2197 $ ] .
the action is induced by a @xmath1-action on @xmath2147 preserving the fan @xmath2146 , given by @xmath2198 .
one has @xmath2199 ; the action identifies @xmath2200 with @xmath2201 and @xmath2202 with @xmath2203 .
we wish to form quotients of @xmath2137 by the group @xmath2204 , but find ourselves unable to do so in the category of @xmath2168$]-schemes .
one can , however , construct quotients by @xmath2205 of thickened neighborhoods of the zero - fiber , i.e. of @xmath2206 , by virtue of the following fact : we write down a concrete model for the quotient @xmath2210 , when @xmath1407 .
we take the affine schemes @xmath2207 for @xmath2211 , and we identify the open subset @xmath2212 of @xmath2213 with the open subset @xmath2214 of @xmath2215 in just the same way as we identify @xmath2212 with an open subset of @xmath2216 to form @xmath2137 .
the result is a proper scheme @xmath2217/(t^k)$ ] whose specialization to @xmath1255 is a cycle of @xmath383 @xmath538 s .
we find it convenient to rename @xmath1553 as @xmath2218 ; so we have @xmath2219/(q^{k / d})$ ] .
these schemes form an inverse system in @xmath438 ; passing to the inverse limit as @xmath2220 , we obtain a proper formal scheme @xmath2221 .
there are tale quotient maps @xmath2222 , where by @xmath2223 we mean the formal base - change from @xmath2224 to @xmath2225 given by @xmath2226 .
the construction of @xmath2227 does not work in quite the same way when @xmath2228 , because one is then gluing @xmath2214 to itself , and the zariski - open cover by the @xmath2169 becomes merely an tale cover .
nonetheless , we can define @xmath2229 as the tale quotient @xmath2230 .
this quotient is the locally - ringed space whose functions are the @xmath1892-invariant functions of @xmath2231 . to see that it is a formal scheme , we need only note that @xmath2232 can be covered by two @xmath1892-invariant formal - affine open sets , which is straightforward to check .
notice also that @xmath2233 $ ] has sections @xmath2234\to u_{i+1/2}$ ] , defined by @xmath2235 and @xmath2236 .
the @xmath1-action intertwines these sections , and consequently , @xmath2237 has a distinguished section @xmath335 .
one now wants to polarize the formal scheme @xmath2232 , that is , to identify an ample line - bundle @xmath2238 . a very ample power of @xmath2239
will then define a projective embedding of @xmath2232 , and in doing so , will ` algebraize ' @xmath2232 , refining it to a true scheme over @xmath2240 .
is that a projective embedding of a complex manifold makes this analytic object algebraic . ]
that ample line bundles algebraize formal schemes is a general setting is a theorem of grothendieck , but here , as noted by gross , it is quite concrete .
we will find a line bundle @xmath2241 , by which we mean a sequence of line - bundles @xmath2242 and isomorphisms @xmath2243 .
we will find a basis @xmath2244 for @xmath2245by this we mean bases @xmath2246 for @xmath2247 , carried one to another by the maps in the inverse system defining a plane embedding of @xmath2232 . in this way we will see that @xmath2248 is cut out from @xmath2249/(t^k))$ ] by an equation which reduces modulo @xmath2250 to that cutting out @xmath2251 . passing to the limit
, we see that @xmath2232 is cut out by an equation from @xmath2252 , and hence arises from a projective scheme @xmath22 .
as gross explains , one obtains a line bundle @xmath2253 by observing that @xmath2146 is the fan dual to an unbounded convex polygon @xmath2254 , namely , the convex hull in @xmath2147 of a sequence of points @xmath2255 , @xmath2139 , given by @xmath2256 and @xmath2257 ( figure [ polytope ] ) .
the cone @xmath2264 is the tangent wedge @xmath2265 at @xmath2266 . for each @xmath2267
, we define @xmath2268 to be the free @xmath2269-module @xmath2270 on a generator @xmath2271 . here
@xmath9 is a formal symbol .
we assemble the @xmath2268 into an invertible sheaf @xmath2272 by declaring that the transition function from @xmath2273 to @xmath2274 is multiplication by @xmath2275 .
each lattice point @xmath2276 defines a section @xmath2277 : over @xmath2169 , we view @xmath1469 as a point in @xmath2278 , and so assign to it a section @xmath2279 .
these local sections agree on overlaps and so define a global section .
the @xmath1-action on @xmath2137 lifts to a @xmath1-action on @xmath379 ( we continue to follow gross ) .
define @xmath2280 by @xmath2281 we have @xmath2282 , from which it is easy to see that @xmath2283 .
thus @xmath1002 generates an action of @xmath1 on @xmath1037 .
we lift the action of @xmath2284 to @xmath379 as follows .
take the map @xmath2285 which defines the @xmath1-action ( namely , @xmath2286 and @xmath2287 ) and lift it to a map @xmath2288 ( we have specified in particular the effect of @xmath2289 on @xmath2271 ) . as a result
, @xmath379 descends to a line bundle @xmath2239 over each scheme @xmath2290 , and indeed over the formal scheme @xmath2227 .
the projective embeddings @xmath2291 for @xmath2292 cut out @xmath2227 as a projective scheme @xmath2293 , lifting the formal scheme structure .
this is even true for @xmath2228 .
the scheme @xmath2294 is the tate curve .
it has its distinguished section @xmath335 .
the line bundle @xmath2295 has a global section @xmath2296 .
this formula is to be interpreted initially on @xmath2207 , where it is a finite sum .
it descends to a section of @xmath379 over @xmath2210 , and thereby a section over @xmath2227 for each @xmath383 .
it is instructive to write @xmath29 in the open set @xmath2214 in the following forms : @xmath2297 where @xmath2298 .
up to a factor of @xmath2299 , @xmath2300 is exactly the fourier expansion for the classical theta - function @xmath2301 , written in terms of @xmath2302 and @xmath2303 , where @xmath637 is the coordinate on @xmath2304 . observe ( i ) that @xmath2305 vanishes precisely where @xmath2306 , whence @xmath379 has degree 1 on the geometric fibers of @xmath2307 ; ( ii ) that @xmath2308 , so @xmath29 vanishes where @xmath2309 ; and hence ( iii ) that @xmath29 vanishes precisely where @xmath2309 , i.e. along @xmath335 .
thus @xmath2310 . using riemann
roch , we see that @xmath2311 is very ample relative to the morphism @xmath2307 , and hence embeds @xmath22 as a weierstrass cubic in @xmath2252 . thus @xmath22 becomes a weierstrass curve , with a canonical differential @xmath26 .
moroever , this perspective makes clear that @xmath22 coincides , as a weierstrass curve , with the curve described in the introduction to this paper .
( however , we shall not review the derivation of the fourier expansions of @xmath2312 and @xmath2312 ) .
we have a canonical @xmath1-basis @xmath2313 for @xmath2314 .
the action of @xmath131 is given by @xmath2315 .
to get a basis over @xmath21 , it suffices to consider @xmath2316 , since we then obtain the remaining lattice points by multiplying by powers of @xmath131 .
there is also a canonical @xmath21-basis @xmath2317 for the @xmath1-invariant part @xmath2318 , @xmath2319 where @xmath155 runs over the cyclic group @xmath2320 . to define the ` theta - functions ' @xmath2321 ,
let @xmath2322 be the piecewise - linear function whose graph is the boundary of @xmath1037 .
define automorphisms @xmath2323 of the dilated polygon @xmath2324 , sending vertex @xmath2325 to the adjacent vertex @xmath2326 , by @xmath2327 then @xmath2328 lies on the boundary of @xmath2324 , and hence so does @xmath2329 for each @xmath2330 .
put @xmath2331 again , the formula for @xmath2321 makes sense , as a finite formal sum , on @xmath2207 , hence on each @xmath2227 , and so finally on @xmath22 .
it is clear that @xmath2317 is a basis for @xmath2332 .
one has @xmath2333 .
explicit multiplication rules for classical theta - functions are standard .
whilst these theta - functions are not precisely identical to the classical theta functions which give canonical bases for @xmath2334 , they do obey a very similar multiplication rule @xcite : @xmath2335 here @xmath2336 is the weighted average with respect to a distribution determined by @xmath2337 and @xmath2338 , @xmath2339 and @xmath2340 three points about @xmath94 are noteworthy : * for any convex function @xmath330 , the quantity @xmath2341 is non - negative , by jensen s inequality .
one has @xmath2342 . *
@xmath94 is a piecewise - linear approximation to the quadratic function @xmath2343 indeed , @xmath2344 for @xmath1531 , and @xmath94 is affine - linear on intervals @xmath2345 $ ] . *
the expression @xmath2346 defined analogously to @xmath1469 , is the area of the triangle @xmath2347 with vertices @xmath2348 ( see figure [ triangles ] ) .
the last fact is key to gross s derivation of ` classical ' hms @xcite .
is the relevant exponent . ] for a fixed @xmath258 , the theta - functions @xmath2352 are _ almost _ mirror to the intersection points of @xmath1504 and @xmath1818 .
we say `` almost '' because correction factors are required , since it is @xmath2353 and not @xmath2354 which appears in the product rule for theta - functions .
it turns out that @xmath1469 has a similar geometric interpretation , one which will be equally central in our derivation of arithmetic hms .
we point out that whilst subsection [ tate section ] was a review , and the material of subsection [ lattice points and mult ] has so far also been standard , this interpretation is to our knowledge original : [ lattice points ] fix @xmath2355 , and say a point in @xmath2356 is a _
perturbed lattice point _ if it is congruent to @xmath2357 modulo @xmath575 .
then , if @xmath2358 , the number of perturbed lattice points inside @xmath2347 is equal to @xmath2353 .
let @xmath2359 be the right triangle with vertices @xmath2349 , @xmath2360 and @xmath2361 .
thus there is a horizontal edge , a vertical edge , and an edge which is a segment of the line @xmath2362 .
write @xmath2363 and @xmath2364 .
let @xmath2365 be the right triangle with vertices @xmath2366 , @xmath2360 and @xmath2367 ; thus @xmath2365 is of the same form as @xmath2359 , substituting @xmath2368 for @xmath2337 and @xmath2369 for @xmath2370 .
the number @xmath2009 of perturbed lattice points in @xmath2347 is the difference @xmath2371 where @xmath2372 is the number of perturbed lattice points in @xmath2373 ( see figure [ triangles2 ] ( left ) ) . in this case ,
@xmath2384 and @xmath2385 .
we calculate @xmath2386 hence , writing @xmath2387 with @xmath2388 , we have @xmath2389 we now compute the difference @xmath2009 by inputting the relation @xmath2390 : @xmath2391 we drop the assumption that @xmath2392 is an integer . in this case
, we can shave off the right - hand edge of the triangle @xmath2359 , so that its vertical edge is at @xmath2393 instead of @xmath2394 ( figure [ triangles2 ] ) .
our previous formula applies to this shaved triangle @xmath2374 .
the number of perturbed lattice points in the trapezium which we shaved off @xmath2359 is computed as follows : the trapezium is a rectangular strip together with a triangle at the bottom .
the triangle contains no lattice points , while the strip contains @xmath2395 lattice points .
similarly , we shave off the right - hand edge of @xmath2365 ; the trapezium which we shave off contains @xmath2396 perturbed lattice points .
note that the difference between the counts of points in these trapezia is @xmath2397 .
we earlier discussed the affine coordinate ring @xmath2400 . to work with this ring
, one must understand the direct system as well as the multiplication of sections of powers of @xmath2401 . on any weierstrass curve @xmath211 ,
the divisor @xmath2402 is very ample .
it defines a homogeneous coordinate ring @xmath2403 , a graded @xmath201-algebra whose isomorphism determines the curve .
the homogeneous coordinate ring has the advantage over the affine one that it is directly determined by the composition maps in @xmath197 .
for a weierstrass curve @xmath353 over @xmath1736 , let @xmath2404 denote the extended homogeneous coordinate ring @xmath2405 .
this bigraded ring differs from @xmath2406 only by the presence of the summand @xmath885 .
this summand has its trace map @xmath2407 .
the ring @xmath2404 determines @xmath162 , and the trace map then determines the differential @xmath26 .
the truncated ring @xmath2408 , where @xmath2409 , with its bigrading and trace @xmath2410 , already determines @xmath2411 , because from it a defining cubic equation ( in hesse form ) can be read off in @xmath2412 . however
, the regular section @xmath335 might not be fully determined by @xmath2413 , or even @xmath2414 , inasmuch as one could replace @xmath335 by a different regular section @xmath1733 such that the divisor @xmath2415 is linearly equivalent to @xmath2416 .
let @xmath216 be an abstract weierstrass curve over @xmath21 whose category @xmath2417 admits a quasi - isomorphism with @xmath23 such that the induced isomorphism @xmath2418 is the standard one .
we must show that @xmath1636 as weierstrass curves .
we shall initially prove that they are isomorphic as curves with differential .
for this it is sufficient to show that @xmath2419 by a trace - respecting bigraded ring isomorphism .
we recall that @xmath2420 denotes an oriented exact lagrangian in @xmath30 of slope @xmath2421 , equipped with its non - trivial double covering , and graded in such a way that @xmath2422 for @xmath2423 ( this is not quite a complete specification of the grading ) .
we have @xmath2424 we use prop .
[ primitive ] to set up the basepoint @xmath9 and the 1-form @xmath29 on @xmath30 in such a way that the exact curve @xmath1818 is the image in @xmath19 of a straight line through the origin in @xmath2356this for @xmath2425 .
we realize the non - trivial double covering of @xmath1818 by selecting the point @xmath2426 , where @xmath2427 , and declaring the double covering @xmath2428 to be trivial away from @xmath230 and to exchange the sheets at @xmath230 .
take @xmath2429 .
the differential on @xmath2430 is zero , since there are no immersed bigons bounding @xmath1504 and @xmath1818 .
thus we have a basis @xmath2431 for @xmath2432 , where @xmath2433 , and @xmath2434\in \r^2/\z^2 = t$ ] .
we have @xmath2435 for @xmath2436 , where @xmath1002 is the nearly - linear dehn twist along @xmath1315 set up at ( [ model twist ] ) ; hence @xmath2437 has basis @xmath2438 when @xmath2439 .
we claim that the floer product @xmath2440 } hf(l_0^\#,l_{(1,-n_1)}^\ # ) \to hf(l_0,l_{(1,-n_1-n_2)})$ ] is given by @xmath2441 an expression formally identical to the multiplication rule for theta - functions ( [ structure constants ] ) ; @xmath2442 and @xmath1469 are as defined there .
the contributions to @xmath2443 are immersed triangles , the images of embedded triangles in @xmath2356 .
the first vertex is a lift of @xmath2444 to @xmath2356 , which we assume is the point @xmath2445 .
the second vertex is a lift of @xmath2446 $ ] which lies on the line of slope @xmath2447 through @xmath2349 ; thus it is of form @xmath2448 where @xmath2449 .
the third vertex is then at @xmath2450 .
the contribution of the triangle just described is @xmath2451 , where the exponent is the number of perturbed lattice points in the triangle , as computed in proposition [ lattice points ] , and the sign @xmath2452 depends on the orientations , double coverings and the intersection signs of the corners determined by the orientations .
a formula for @xmath2453 is given in @xcite ; it is simplest when , as here , all the corners have intersection number @xmath180 ( and therefore even index for floer _
_ co__homology ) . in that case ,
@xmath1590 , where where @xmath70 is the number of stars on the boundary .
there are @xmath2454 stars on the @xmath1504 boundary , @xmath2455 on the @xmath2456 boundary , and @xmath2457 on the @xmath2458 boundary , so @xmath2459 and @xmath2460 .
this justifies our claim .
define a @xmath21-linear map @xmath2461 by linearly extending the assignment @xmath2462 for @xmath2463 together with the canonical ring - isomorphism @xmath2464 defined by the weierstrass differential @xmath26 . in view of ( [ structure constants ] ) and prop .
[ lattice points ] , and the fact that the unit of @xmath2465 is also the unit of @xmath2466 , the map @xmath2467 is a bigraded ring isomorphism preserving the trace . therefore , there is an isomorphism @xmath2468 , where @xmath26 is the standard weierstrass differential on @xmath22 and @xmath1733 is some section of @xmath2469 , the regular locus in @xmath22 . since @xmath22 is a
_ generalized elliptic curve _ ( see @xcite or ( * ? ? ? * def .
2.1.4 ) ) , one has a homomorphism @xmath2470 , from the group @xmath982 of sections of @xmath2471 , to the automorphism group of @xmath22 fixing the base : sections act on @xmath22 by fiberwise translations .
consequently , all regular sections are related by automorphisms of @xmath22 .
hence @xmath2472 is isomorphic to @xmath2473 . at this point
we have proved clauses ( i ) and ( ii ) of theorem [ mainth ] apart from the uniqueness statement in ( i ) . that is easily taken care of : @xmath23 split - generates @xmath2474 , and the functor @xmath49
is determined , up to natural quasi - equivalence , by its effect on a full , split - generating subcategory . in the argument above
, we identified the mirror essentially by describing a canonical isomorphism of graded rings @xmath2475 given , in degrees @xmath2476 where @xmath2477 is exact , by matching up the canonical bases : @xmath2478 .
the argument did not depend on this assignment being the one arising from the functor @xmath49 , but it is natural to expect that this is so .
the assignment @xmath2479 , defined when @xmath2480 , respects products .
hence , taking @xmath2481 , we see that @xmath2482 extends to a ring isomorphism , because @xmath260 is generated in degree @xmath2483 and subject only to a relation in degree 9 .
the functor @xmath49 defines another such isomorphism .
thus @xmath2484 extends to an automorphism of @xmath260 .
this must come from a weierstrass automorphism @xmath54 of @xmath22 .
the only possibilities for @xmath54 are the identity and the hyperelliptic involution , but we do not identify @xmath54 here .
we begin by clarifying how to construct @xmath116 .
we now consider @xmath50 and @xmath53 as objects of the ` absolute ' fukaya category @xmath71 over @xmath76so their exactness plays no role .
they form a full subcategory @xmath2485 , which we identify with @xmath2486 for some mirror weierstrasss curve @xmath2487 .
thus we get a functor @xmath2488 inducing a quasi - equivalence @xmath2489 .
we identify @xmath2442 via the homogeneous coordinate ring for its cubic embedding , just as we did for @xmath216 .
thus we must count triangles according to their areas , not the number of lattice points they contain . by the argument of @xcite
( see also @xcite ) , we identify the homogeneous coordinate ring of @xmath2442 as that of @xmath2490 , with the bases of intersection points @xmath2491 mapping to standard bases of @xmath29-functions .
thus @xmath2492 as weierstrass curves . since @xmath2442 is smooth , every bounded coherent complex is quasi - isomorphic to a perfect complex , and so @xmath2493 .
thus we can think of @xmath116 as a functor @xmath2494 we must now compare the two functors @xmath2495 : first @xmath118 , and second , @xmath49 followed a base - change functor call this composite @xmath2496 .
both extend to functors defined on @xmath2474 , and it will suffice to show that these are homotopic . the inclusion @xmath2497 induces a quasi - equivalence @xmath2498 , and the definitions of @xmath116 and @xmath2496 both depend on a choice of quasi - inverse @xmath2499 .
we take this to be the same in both cases , so that the functors have the same effect on objects otherwise the assertion of homotopy makes no sense . the construction of the functor @xmath89 depends on choices of functions @xmath2500 for each exact lagrangian @xmath86 such that @xmath2501 .
pick such a function @xmath2502 , and choose @xmath2503 so that the unique intersection point @xmath2504 has symplectic action 0 .
the endomorphism space @xmath2505 is defined via a hamiltonian image @xmath2506 ; endow this with the function @xmath2507 ; similarly for @xmath1315 .
since @xmath89 acts as the identity on objects , it defines a quasi - isomorphism @xmath2508 .
it induces the canonical isomorphism @xmath2509 , because the relevant intersection points have action zero .
we have quasi - isomorphisms @xmath2510 and @xmath2511 , both inducing the canonical isomorphisms on cohomology .
that is , we have two minimal @xmath13-structures on the graded algebra @xmath200 , and identifications of both of them with @xmath2512 . by theorem [ ainf comparison ] ,
these two identifications must be induced by an automorphism of @xmath2513 respecting the weierstrass data @xmath2514 .
this can only be the identity map .
the result follows .
a. bondal , m. kapranov , _ representable functors , serre functors , and reconstructions _
( russian ) izv .
nauk sssr ser . mat .
53 ( 1989 ) , no . 6 , 11831205 , 1337
; translation in math .
ussr - izv .
35 ( 1990 ) , no .
3 , 519541 .
p. deligne , _ courbes elliptiques : formulaire daprs j. tate _ , modular functions of one variable , iv ( proc
. internat .
summer school , univ .
antwerp , antwerp , 1972 ) , pp .
lecture notes in math .
476 , springer , berlin , 1975 .
deligne , p. ; rapoport , m. _ les schmas de modules de courbes elliptiques .
_ modular functions of one variable , ii ( proc .
summer school , univ .
antwerp , antwerp , 1972 ) , pp .
lecture notes in math .
349 , springer , berlin , 1973 .
d. quillen , _ on the ( co- ) homology of commutative rings _ , in _ applications of categorical algebra _ ( proc .
pure math .
xvii , new york , 1968 ) pp .
soc . , providence , r.i . , 1970 . | this paper explores a refinement of homological mirror symmetry which relates exact symplectic topology to arithmetic algebraic geometry . we establish a derived equivalence of the fukaya category of the 2-torus , relative to a basepoint , with the category of perfect complexes of coherent sheaves on the tate curve over the formal disc @xmath0 .
it specializes to a derived equivalence , over @xmath1 , of the fukaya category of the punctured torus with perfect complexes on the curve @xmath2 over @xmath3 , the central fibre of the tate curve ; and , over the ` punctured disc ' @xmath4 , to an integral refinement of the known statement of homological mirror symmetry for the 2-torus .
we also prove that the wrapped fukaya category of the punctured torus is derived - equivalent over @xmath1 to coherent sheaves on the central fiber of the tate curve . |
the study of kerr optical frequency comb generation has been the focus of extensive research efforts in recent years .
these spectral grids are obtained through pumping a ultra - high @xmath0 whispering - gallery mode ( wgm ) resonator with a narrow - linewidth continuous ( cw ) laser .
almost all previous research have been devoted to the investigation of kerr comb generation in the anomalous group velocity dispersion ( gvd ) regime .
in fact , it was thought for a long time that normal gvd kerr comb generation was impossible , and later on , it has been shown to occur only under fairly exceptional circumstances ( see for example refs .
this explain why the quasi - totality of the scientific literature on kerr combs assumes a laser pump frequency in the anomalous gvd regime for the bulk material of the resonator .
the theoretical explanation of the difficulty to generate kerr combs in the normal gvd regime was the purpose of the first part of this research work , presented in ref .
@xcite .
when the dispersion is anomalous , earlier studies on kerr comb generation have shown that above a given threshold , the long - lifetime photons originating from the pump interact nonlinearly with the medium and populate the neighboring cavity modes through four - wave mixing ( fwm ) .
the resulting permanent state features an all - to - all coupling amongst the excited modes , which can enable various dynamical outputs such as phase - locked ( through turing patterns or solitons ) , pulsating and even chaotic states .
in particular , the phase - locked states are expected to be useful for a wide spectrum of applications @xcite .
synthetic studies where all these behaviors are associated to well identified regions of the parameter space are scarce .
most research articles so far have focused on specific phenomenologies ( modulational instability , solitons , chaos , breathers , etc . ) and our objective in this paper is to provide a larger viewpoint for the understanding of kerr comb generation with anomalous gvd .
the models used so far to investigate kerr comb generation are either based on a modal expansion approach @xcite or on a spatiotemporal formalism @xcite which is a variant of the lugiato - lefever equation ( lle ) @xcite .
both approaches can be shown to be equivalent under certain conditions @xcite , but the latter appears to be more powerful to investigate the collective dynamics of a large number of modes .
( color online ) ( a ) schematical representation of a wgm resonator pumped with cw laser .
( b ) eigenmodes of the resonator .
the real location of the modes with anomalous dispersion is represented in solid lines , while the dashed lines represent the location of the modes if the dispersion was null ( perfect equidistance ) .
note that the anomalous dispersion pulls the modes rightwards ( blueshift ) .
, width=302 ] in this article which is the second part of ref .
@xcite , we study the bifurcation structure related to the different steady state solutions when the nonlinear wgm resonator is pumped in the anomalous gvd regime
. our parameter space will be two - dimensional , and will involve the pump power and the laser detuning relatively to the cavity resonance .
these parameters have been chosen because of their experimental relevance .
the plan of the article is the following . in sec .
[ model_eq_stab ] , we briefly present the model , as well as the various equilibria and the related spatial bifurcations . then , a synthetic bifurcation map is obtained and discussed in sec .
[ bifurcationmap ] .
turing patterns ( rolls ) arising from modulational instability ( mi ) are shown to emerge in the system , and they are investigated in sec .
[ turingpatterns ] .
in particular , their super- or sub - critical nature is analyzed in the time domain with respect to the detuning frequency of the pump laser .
the mi gain is also determined analytically in our case and enables to understand the main properties of these patterns , such that the number of rolls , or their growth rate as the pump power of the laser is increased .
we study the emergence of bright cavity solitons and soliton molecules in secs .
[ brightcavitysolitons ] and sec . [ brightsolitonsmolecules ] respectively .
the connexion between these solitons and sub - critical turing patterns is evidenced and enables to understand their main dynamical properties .
we also show that soliton molecules can coexist in the resonator , and generate very complex kerr comb spectra .
breather solitons are studied in sec .
[ breathingsolitons ] , and we show that the breathing period is necessarily of the order of magnitude of the photon lifetime .
chaotic behavior is investigated in sec .
[ chaos ] , and we identify two main routes to chaos , either through the destabilization turing patterns through higher order bifurcations , of from the destabilization of solitonic structures .
section [ influencedispersion ] focuses on the study of the effects of the magnitude of the dispersion parameter in the dissipative structures that are formed in the cavity .
the last section of this article resumes our results .
+ + denomination & eigenvalues @xmath1 & pictogram & bifurcation & location in fig .
[ figurebifnotatscale ] + & & & & + type 1 & @xmath2 & ( -2,1)(1,-0.8 ) ( 0,0)(1,0)3 ( 0,0)(0,1)3 ( 0,0)(-1,0)3 ( 0,0)(0,-1)3 ( -1.2,0.0 ) ( -3,0.0 ) ( 3,0.0 ) ( 1.2,0.0 ) & & + & & & & + & & & & + type 2 & @xmath3 & ( -2,1)(1,-0.8 ) ( 0,0)(1,0)3 ( 0,0)(0,1)3 ( 0,0)(-1,0)3 ( 0,0)(0,-1)3 ( -2.3,-1.5)@xmath4(0.0,0.0 ) & @xmath5 & @xmath6 + & & & & + & & & & + type 3 & @xmath7 & ( -2,1)(1,-0.7 ) ( 0,0)(1,0)3 ( 0,0)(0,1)3 ( 0,0)(-1,0)3 ( 0,0)(0,-1)3 ( 0.0,-1.2 ) ( 0.0,-3 ) ( 0.0,3 ) ( 0.0,1.2 ) & & + & & & & + & & & & + type 4 & @xmath8 & ( -2,1)(1,-0.7 ) ( 0,0)(1,0)3 ( 0,0)(0,1)3 ( 0,0)(-1,0)3 ( 0,0)(0,-1)3 ( -1.65,-1)@xmath4 ( -2,0.0 ) ( 2,0.0 ) & @xmath9 & @xmath10 + & & & & + & & & & + type 5 & @xmath11 & ( -2,1)(1,-0.7 ) ( 0,0)(1,0)3 ( 0,0)(0,1)3 ( 0,0)(-1,0)3 ( 0,0)(0,-1)3 ( -1.65,-1)@xmath4 ( 0.0,-2 ) ( 0.0,2 ) & @xmath12 & @xmath13 , @xmath14 , @xmath15 , @xmath16 , @xmath17 + & & & & + & & & & + type 6 & @xmath18 & ( -2,1)(1,-0.7 ) ( 0,0)(1,0)3 ( 0,0)(0,1)3 ( 0,0)(-1,0)3 ( 0,0)(0,-1)3 ( 1.5,0.0 ) ( -1.5,0.0 ) ( 0.0,1.5 ) ( 0.0,-1.5 ) & & + & & & & + & & & & + type 7 & @xmath19 & ( -2,1)(1,-0.7 ) ( 0,0)(1,0)3 ( 0,0)(0,1)3 ( 0,0)(-1,0)3 ( 0,0)(0,-1)3 ( -1.65,0.5)@xmath4 ( -1.65,-2.5)@xmath4 & @xmath20 & @xmath21 , @xmath22 + & & & & + & & & & + type 8 & @xmath23 & ( -2,1)(1,-0.7 ) ( 0,0)(1,0)3 ( 0,0)(0,1)3 ( 0,0)(-1,0)3 ( 0,0)(0,-1)3 ( -3,-1)@xmath4 ( 0,-1)@xmath4 & & + & & & & + & & & & + type 9 & @xmath24 & ( -2,1)(1,-0.7 ) ( 0,0)(1,0)3 ( 0,0)(0,1)3 ( 0,0)(-1,0)3 ( 0,0)(0,-1)3 ( -1.5,1.5 ) ( 1.5,1.5 ) ( -1.5,-1.5 ) ( 1.5,-1.5 ) & & + & & & & +
as in the first part of this study , the system is a wgm disk pumped by a cw pump laser . as displayed in fig .
[ figure1 ] . for the sake of the self - consistency of this article ,
we very briefly recall the main features of the model in the next subsection .
the eigenvalues of the wgm cavity are taylor - expanded as @xmath25 , where @xmath26 is the integer eigennumber that unambigously labels the modes belonging to the fundamental ( or toric ) family of wgms , while @xmath27 and @xmath28 are respectively the free - spectral range ( fsr ) of the wgm resonator and the second - order dispersion coefficient .
the fsr obeys @xmath29 , with @xmath30 being the velocity of light in vacuum , @xmath31 the main radius of the disk , @xmath32 is the refraction index of the disk for a frequency equal to the resonance @xmath33 , and @xmath34 being the round - trip time . in its normalized form
, the total intracavity field obeys the following partial differential equation when the modal linewidths are quasi - degenerated @xcite @xmath35 where @xmath36 is the complex envelope of the total intra - cavity field , @xmath37 $ ] is the azimuthal angle along the circumference , and @xmath38 is the dimensionless time , with @xmath39 being the loaded ( total ) modal linewidth .
this lle has periodic boundary conditions , and @xmath40 represents the intra - cavity fields dynamics in the moving frame @xcite .
the parameters of this equation are : the frequency detuning @xmath41 where @xmath42 and @xmath43 are respectively the angular frequencies of the pumping laser and the cold - cavity resonance ; the overall dispersion parameter @xmath44 ( positive for normal gvd and negative for anomalous gvd ) ; and finally the pump field intensity @xmath45^{1/2 } \ , [ { p}/{\hbar \omega_0}]^{1/2}$ ] where @xmath46 is the intensity ( in w ) of the laser pump at the input of the resonator , @xmath47 is the nonlinear gain , @xmath32 and @xmath48 are respectively the linear and nonlinear refraction indices of the bulk material , @xmath49 is the effective volume of the pumped mode , and @xmath50 being the extrinsic ( coupling ) modal linewidth @xcite .
it is important to recall that @xmath40 represents the intracavity field in a moving frame that is circumferentially rotating at frequency @xmath27 .
the order of magnitude for these various parameters are given and commented in part i @xcite .
as demonstrated in ref .
@xcite , the intracavity field can be expanded as @xmath51 } \
, , \label{psi_expansion}\end{aligned}\ ] ] where @xmath52 corresponds to the normalized modal fields introduced in refs .
+ the equilibria @xmath53 of eq . ( [ final_eq_dimensionless ] ) obey @xmath54 \rho^2 \equiv g ( \alpha,\rho ) \ , .
\label{def_f2}\end{aligned}\ ] ] with @xmath55 .
this cubic equation in @xmath56 only has one solution when @xmath57 ( always stable ) .
when @xmath58 , there are three equilibria @xmath59 when @xmath60 $ ] , with @xmath61 $ ] and @xmath62/{3}$ ] .
the intermediate solutions is always temporally unstable while the extremal solutions are always stable ( hysteresis ) .
the spatial bifurcation study in the anomalous gvd case is more complex than in the normal gvd regime because several solutions of interest depend on the boundary condition .
disregarding the finiteness of the @xmath63-domain is therefore not a correct assumption most of the time .
however , since both the modal expansion and the spatiotemporal models already gave a solid theoretical understanding of the effects induced by the periodic boundary conditions , we can combine these previous knowledge with the fixed - points eigenvalue analysis in order to gain detailed understanding of the various solutions of the system .
as for the normal gvd case , we set @xmath64 and introduce the intermediate variables @xmath65 in order to rewrite the lle under the form @xmath66 where @xmath67 , with @xmath68 and @xmath69 being the real and imaginary parts of @xmath40 , respectively .
( color online ) supercritical turing patterns ( so - called _ soft _ excitation ) .
the parameters are @xmath70 , @xmath71 , and @xmath72 [ the pumping power @xmath73 can be directly calculated using eq .
( [ def_f2 ] ) ] .
( a ) transient dynamics .
( b ) final pattern in the azimuthal direction .
( c ) 3d representation .
( d ) corresponding kerr comb .
, title="fig:",width=173 ] ( color online ) supercritical turing patterns ( so - called _ soft _ excitation ) .
the parameters are @xmath70 , @xmath71 , and @xmath72 [ the pumping power @xmath73 can be directly calculated using eq .
( [ def_f2 ] ) ] .
( a ) transient dynamics .
( b ) final pattern in the azimuthal direction .
( c ) 3d representation .
( d ) corresponding kerr comb .
, title="fig:",width=128 ] ( color online ) supercritical turing patterns ( so - called _ soft _ excitation ) .
the parameters are @xmath70 , @xmath71 , and @xmath72 [ the pumping power @xmath73 can be directly calculated using eq .
( [ def_f2 ] ) ] .
( a ) transient dynamics .
( b ) final pattern in the azimuthal direction .
( c ) 3d representation .
( d ) corresponding kerr comb .
, title="fig:",width=128 ] ( color online ) supercritical turing patterns ( so - called _ soft _ excitation ) .
the parameters are @xmath70 , @xmath71 , and @xmath72 [ the pumping power @xmath73 can be directly calculated using eq .
( [ def_f2 ] ) ] .
( a ) transient dynamics .
( b ) final pattern in the azimuthal direction .
( c ) 3d representation .
( d ) corresponding kerr comb .
, title="fig:",width=173 ] the jacobian matrix around an equilibrium @xmath74 is @xmath75 and the corresponding eigenvalues obey @xmath76 note that since @xmath77 is negative here ( anomalous gvd ) , it has been rewritten as @xmath78 in order to facilitate the comparison with the normal dispersion case studied in part i @xcite . for being of fourth polynomial order , eq .
( [ polynom ] ) will always yield a quadruplet of eigenvalues for each solution .
hence , in @xmath79@xmath73 plane , there will be one quadruplet in the single - equilibrium area , three between the lines @xmath80 , and two quadruplets onto these boundary lines .
we analyze below the nature ( real , pure imaginary or complex ) of the eigenvalues as a function of the sign of the discriminant @xmath81 of eq .
( [ polynom ] ) .
the solutions of the paired solutions of eq .
( [ polynom ] ) are @xmath83 \ , .
\label{sol_discr_pos}\end{aligned}\ ] ] the product of these paired solutions is @xmath84 , which corresponds to @xmath85 ( see eq .
( [ def_f2 ] ) ) .
three sub - cases have to be considered depending on the sign of @xmath85 : * if @xmath86 , the eigenvalues can be written as @xmath87 if @xmath88 ( eigenvalues of type 1 ) ; as @xmath89 if @xmath90 ( type 2 ) , and as @xmath91 if @xmath92 ( type 3 ) . * if @xmath93 , the eigenvalues can be written as @xmath94 if @xmath88 ( type
4 ) ; and as @xmath95 if @xmath92
( type 5 ) .
the eigenvalues degenerate here to type 2 when @xmath90 . * if @xmath96 , the eigenvalues have the form @xmath97 ( eigenvalue of type 6 ) . in this case , the solutions simply obey ( double - root ) @xmath99 \ , .
\label{sol_discr_null}\end{aligned}\ ] ] hence , the eigenvalues can be written as @xmath100 when @xmath101 ( type 7 ) ; as @xmath102 when @xmath103 ( type 2 ) ; and as @xmath104 when @xmath105 ( type 8) . here
, the eigenvalues are complex : @xmath107 \ , .
\label{sol_discr_neg}\end{aligned}\ ] ] and they have the explicit form @xmath108 , here corresponding to type 9 .
( color online ) subcritical turing patterns ( so - called _ hard _ excitation ) .
the parameters are @xmath109 , @xmath71 , and @xmath72 [ the pumping power @xmath73 can be directly calculated using eq . ( [ def_f2 ] ) ] .
( a ) transient dynamics .
( b ) final pattern in the azimuthal direction .
( c ) 3d representation .
( d ) corresponding kerr comb .
, title="fig:",width=173 ] ( color online ) subcritical turing patterns ( so - called _ hard _ excitation ) .
the parameters are @xmath109 , @xmath71 , and @xmath72 [ the pumping power @xmath73 can be directly calculated using eq .
( [ def_f2 ] ) ] .
( a ) transient dynamics .
( b ) final pattern in the azimuthal direction .
( c ) 3d representation .
( d ) corresponding kerr comb . ,
title="fig:",width=128 ] ( color online ) subcritical turing patterns ( so - called _ hard _ excitation ) .
the parameters are @xmath109 , @xmath71 , and @xmath72 [ the pumping power @xmath73 can be directly calculated using eq .
( [ def_f2 ] ) ] .
( a ) transient dynamics .
( b ) final pattern in the azimuthal direction .
( c ) 3d representation .
( d ) corresponding kerr comb .
, title="fig:",width=128 ] ( color online ) subcritical turing patterns ( so - called _ hard _ excitation ) .
the parameters are @xmath109 , @xmath71 , and @xmath72 [ the pumping power @xmath73 can be directly calculated using eq .
( [ def_f2 ] ) ] .
( a ) transient dynamics .
( b ) final pattern in the azimuthal direction .
( c ) 3d representation .
( d ) corresponding kerr comb .
, title="fig:",width=173 ]
the classification of the eigenvalues of the various fixed points in the @xmath79@xmath73 plane performed in the preceding section enables to obtain a stability map which is displayed in fig . [ figurebifnotatscale ] .
the various areas of this map are similar to the ones we have obtained in the normal gvd regime @xcite , where we had also explained that the @xmath110 symmetry in our system imposes that the relevant bifurcations should necessarily be _ reversible _
( see ref .
@xcite , chapter 4 ) .
the following four bifurcations can emerge in fig .
[ figurebifnotatscale ] : * @xmath9 bifurcation : also referred to as takens - bogdanov bifurcation , it arises when a quadruplet of eigenvalues is of type 4 .
this bifurcation occurs here along the line @xmath10 . *
@xmath12 bifurcation : it corresponds to a quadruplet of type 5 , and it is present in our system in the lines @xmath13 , @xmath15 and @xmath17 . *
@xmath20 bifurcation : also known as the `` @xmath111 resonance '' or the `` hamiltonian hopf '' bifurcation , it emerges when a quadruplet of eigenvalues is of type 7 . in fig .
[ figurebifnotatscale ] , it corresponds to the lines @xmath21 and @xmath22 . *
@xmath5 bifurcation : this co - dimension 2 bifurcation arises when a quadruplet of eigenvalues degenerates to the origin ( type 2 ) .
this situation is only witnessed at the point @xmath112 .
( color online ) diagram showing the difference between supercritical ( @xmath113 ) and subcritical ( @xmath114 ) pitchfork bifurcation towards turing patterns .
the continuous line denotes a stable amplitude while the dashed line stands for a unstable steady state .
( a ) in the supercritical case , no kerr comb is possible below the threshold pump power @xmath115 . above this threshold ,
the sidemodes of the kerr comb can continuously grow from infinitesimally small to significantly large as the pump is increased .
the stable equilibrium and the turing pattern are never simultaneously stable .
this excitation mode is sometimes referred to as _ soft_. ( b ) in the subcritical case , kerr comb is possible in a small interval _ below _ the threshold pump power @xmath115 .
moreover , in the small interval below @xmath115 where the comb can be excited , a stable turing pattern does coexist with the equilibrium solution ( bistability ) . as a consequence ,
the transition from the flat state to the turing pattern is abrupt , and the sidemodes of the comb can not be infinitesimally small .
this excitation mode is sometimes referred to as _
hard_. , title="fig:",width=132 ] ( color online ) diagram showing the difference between supercritical ( @xmath113 ) and subcritical ( @xmath114 ) pitchfork bifurcation towards turing patterns .
the continuous line denotes a stable amplitude while the dashed line stands for a unstable steady state .
( a ) in the supercritical case , no kerr comb is possible below the threshold pump power @xmath115 . above this threshold
, the sidemodes of the kerr comb can continuously grow from infinitesimally small to significantly large as the pump is increased .
the stable equilibrium and the turing pattern are never simultaneously stable .
this excitation mode is sometimes referred to as _ soft_. ( b ) in the subcritical case , kerr comb is possible in a small interval _ below _
the threshold pump power @xmath115 .
moreover , in the small interval below @xmath115 where the comb can be excited , a stable turing pattern does coexist with the equilibrium solution ( bistability ) . as a consequence ,
the transition from the flat state to the turing pattern is abrupt , and the sidemodes of the comb can not be infinitesimally small .
this excitation mode is sometimes referred to as _
hard_. , title="fig:",width=132 ] ( color online ) numerical simulations showing the super - critical and sub - critical nature of the turing patterns as the detuning parameter @xmath79 is varied across the critical value @xmath116 .
the other parameters are @xmath71 and @xmath72 .
( a ) growth of the pattern for the super - critical case , where @xmath117 .
it can be seen that after a critical value given by @xmath118 , the amplitude of the pattern grows smoothly .
( b ) growth of the pattern in the sub - critical case , with @xmath119 . here
, the hysteresis area can clearly be identified when @xmath79 is first smoothly swept upwards , and then downwards .
note that the hysteresis area increases with the detuning @xmath120 .
, title="fig:",width=283 ] ( color online ) numerical simulations showing the super - critical and sub - critical nature of the turing patterns as the detuning parameter @xmath79 is varied across the critical value @xmath116 .
the other parameters are @xmath71 and @xmath72 .
( a ) growth of the pattern for the super - critical case , where @xmath117 . it can be seen that after a critical value given by @xmath118 , the amplitude of the pattern grows smoothly .
( b ) growth of the pattern in the sub - critical case , with @xmath119 . here
, the hysteresis area can clearly be identified when @xmath79 is first smoothly swept upwards , and then downwards .
note that the hysteresis area increases with the detuning @xmath120 .
, title="fig:",width=264 ] note that in the pictograms of type 3 and 6 , some eigenvalues are located on the imaginary axis , leading to the so - called @xmath121 and @xmath122 bifurcations respectively .
however , because of the @xmath110 reversibility of the system , these eigenvalues stay on the imaginary axis and impede the transversality condition to be fulfilled .
these bifurcations therefore dynamically irrelevant in our system and have not been highlighted in table [ nomenclaturebifurcations ] @xcite .
these four bifurcations are exactly those that were obtained in the normal gvd case .
however , the eigenvalue structure in the anomalous gvd case is totally different : all the eigenvalues are rotated by @xmath123 , because the eigenvalues of the normal and anomalous gvd regimes only differ by a multiplicative factor @xmath124 .
this _ essential _ difference is the one that explains the intrinsically different dynamics that can be witnessed in both dispersion regimes : very limited in the normal dispersion case , and very rich when the dispersion is anomalous .
for example , it was shown in part i @xcite that no kerr comb generation was possible for @xmath57 because the line @xmath21 had eigenvalues of type 8 , which does not correspond to a bifurcation . however , in the anomalous gvd regime , this same line corresponds to eigenvalues of type 7 , that is , to a @xmath20 bifurcation .
this bifurcation at @xmath125 has been studied in much detail in ref .
@xcite using a modal expansion approach , and it was shown that it corresponded to the generation of the so called primary comb .
further analysis shows that this bifurcation corresponds in fact to modulational instability and leads to azimuthal turing patterns @xcite . from a more general perspective
, it appears that at the opposite of what was observed in the normal gvd regime , kerr combs can emerge in the anomalous gvd regime for any value of the cavity detuning @xmath79 .
the variety of solutions that can be obtained depending on the parameters and on the initial conditions is analyzed in detail in the following sections .
( color online ) determination of the parametric ( or mi ) gain in the system , with the parameters @xmath126 and @xmath127 .
( a ) positive gain is experienced in the system when the figurative point is in the shaded area of the plane @xmath128@xmath56 , and negative gain is experienced in the white area .
the densely dotted line within the shaded area indicate the location of the maximum gain .
for a fixed value of @xmath56 ( that is , of the pump power @xmath73 ) , the corresponding horizontal line intersects the gain area when @xmath129 , and thereby delimit the modes @xmath130 which can grow through mi .
( b ) mi gain corresponding to the three pump levels of ( a ) .
note that as the pump power is increased , the maximum gain mode is shifted away from the pump , and the mi gain bandwidth is both shifted outwards and increased as well . , title="fig:",width=264 ] ( color online ) determination of the parametric ( or mi ) gain in the system , with the parameters @xmath126 and @xmath127 .
( a ) positive gain is experienced in the system when the figurative point is in the shaded area of the plane @xmath128@xmath56 , and negative gain is experienced in the white area .
the densely dotted line within the shaded area indicate the location of the maximum gain . for a fixed value of @xmath56 ( that is , of the pump power @xmath73 )
, the corresponding horizontal line intersects the gain area when @xmath129 , and thereby delimit the modes @xmath130 which can grow through mi .
( b ) mi gain corresponding to the three pump levels of ( a ) .
note that as the pump power is increased , the maximum gain mode is shifted away from the pump , and the mi gain bandwidth is both shifted outwards and increased as well .
, title="fig:",width=264 ]
pattern formation in systems dynamically described by partial differential equations was investigated for the first time by alan turing in his seminal work of morphogenesis @xcite . in our system ,
the so - called _ turing patterns _ originate from the @xmath20 bifurcation arising at @xmath98 for @xmath101 ( lines @xmath21 and @xmath22 ) . in this section
, we will mainly focus for the sake of simplification on the case @xmath57 where this bifurcation is the only one that can occur in the system ( portion of line @xmath21 ) .
the emergence of turing patterns has been studied in ref .
@xcite with a modal expansion approach ( where they induced the so - called _ primary combs _ ) , and compared to experimental measurements in ref .
@xcite . in the general case of pattern formation in lle equations ,
an abundant literature is indeed available and include for example refs .
@xcite ( and references therein ) , and refs .
@xcite for example focus on the special case of spatial solitons where the laplacian term stands for diffraction instead of dispersion . in our case
, it can be shown that turing patterns emerge following two different scenarios , either following a supercritical bifurcation ( soft excitation ) , or a sub - critical bifurcation ( hard excitation ) , which are analyzed hereafter .
( color online ) formation of bright solitons .
the parameters are @xmath103 , @xmath131 , and @xmath132 .
( a ) transient dynamics .
( b ) final pattern in the azimuthal direction .
( c ) 3d representation .
( d ) corresponding kerr comb . ,
title="fig:",width=173 ] ( color online ) formation of bright solitons .
the parameters are @xmath103 , @xmath131 , and @xmath132 .
( a ) transient dynamics .
( b ) final pattern in the azimuthal direction .
( c ) 3d representation .
( d ) corresponding kerr comb . ,
title="fig:",width=128 ] ( color online ) formation of bright solitons .
the parameters are @xmath103 , @xmath131 , and @xmath132 .
( a ) transient dynamics .
( b ) final pattern in the azimuthal direction .
( c ) 3d representation .
( d ) corresponding kerr comb . , title="fig:",width=128 ] ( color online ) formation of bright solitons .
the parameters are @xmath103 , @xmath131 , and @xmath132 .
( a ) transient dynamics .
( b ) final pattern in the azimuthal direction .
( c ) 3d representation .
( d ) corresponding kerr comb .
, title="fig:",width=173 ] the super- of subcritical nature if the turing patterns originating form the lle was already foreshadowed in the original work of lugiato and lefever @xcite .
it has later on been studied extensively by several research groups investigating dissipative structures in nonlinear optical cavities , and two recent noteworthy works on this topic are refs .
the essential difference between a super- and a subcritical pitchfork in our context is explained in fig .
[ super_sub ] , and it depends on how the comb emerges around the threshold pump power @xmath133 which is obtained from eq .
( [ def_f2 ] ) by setting @xmath134 . the numerical simulation of the lle shows in fig . [ num_sub_super ] that effectively , as the critical value @xmath116 is crossed , the growth of the turing rolls undergoes a structural change which
is mathematically explained by the paradigm of super- and sub - critical pitchfork bifurcations .
( color online ) formation of a soliton molecule .
the parameters are exactly those of fig .
[ single - soliton ] , and only the initial condition has changed ( more powerful pulse ) . it can be inferred that a soliton is in fact a pulse isolated from a sub - critical turing pattern , and that a soliton molecule corresponds to several of such pulses .
( a ) transient dynamics .
( b ) final pattern in the azimuthal direction .
( c ) 3d representation .
( d ) corresponding kerr comb .
( e ) soliton molecule formed with 5 solitons obtained with a different initial condition .
( f ) soliton molecule formed with 7 solitons obtained with a different initial condition . , title="fig:",width=173 ] ( color online ) formation of a soliton molecule .
the parameters are exactly those of fig .
[ single - soliton ] , and only the initial condition has changed ( more powerful pulse ) .
it can be inferred that a soliton is in fact a pulse isolated from a sub - critical turing pattern , and that a soliton molecule corresponds to several of such pulses .
( a ) transient dynamics .
( b ) final pattern in the azimuthal direction .
( c ) 3d representation .
( d ) corresponding kerr comb .
( e ) soliton molecule formed with 5 solitons obtained with a different initial condition .
( f ) soliton molecule formed with 7 solitons obtained with a different initial condition . , title="fig:",width=128 ] ( color online ) formation of a soliton molecule .
the parameters are exactly those of fig .
[ single - soliton ] , and only the initial condition has changed ( more powerful pulse ) .
it can be inferred that a soliton is in fact a pulse isolated from a sub - critical turing pattern , and that a soliton molecule corresponds to several of such pulses .
( a ) transient dynamics .
( b ) final pattern in the azimuthal direction .
( c ) 3d representation .
( d ) corresponding kerr comb .
( e ) soliton molecule formed with 5 solitons obtained with a different initial condition .
( f ) soliton molecule formed with 7 solitons obtained with a different initial condition . , title="fig:",width=151 ] ( color online ) formation of a soliton molecule .
the parameters are exactly those of fig .
[ single - soliton ] , and only the initial condition has changed ( more powerful pulse ) .
it can be inferred that a soliton is in fact a pulse isolated from a sub - critical turing pattern , and that a soliton molecule corresponds to several of such pulses .
( a ) transient dynamics .
( b ) final pattern in the azimuthal direction .
( c ) 3d representation .
( d ) corresponding kerr comb .
( e ) soliton molecule formed with 5 solitons obtained with a different initial condition .
( f ) soliton molecule formed with 7 solitons obtained with a different initial condition . , title="fig:",width=151 ] ( color online ) formation of a soliton molecule .
the parameters are exactly those of fig .
[ single - soliton ] , and only the initial condition has changed ( more powerful pulse ) .
it can be inferred that a soliton is in fact a pulse isolated from a sub - critical turing pattern , and that a soliton molecule corresponds to several of such pulses .
( a ) transient dynamics .
( b ) final pattern in the azimuthal direction .
( c ) 3d representation .
( d ) corresponding kerr comb .
( e ) soliton molecule formed with 5 solitons obtained with a different initial condition .
( f ) soliton molecule formed with 7 solitons obtained with a different initial condition . , title="fig:",width=151 ] ( color online ) formation of a soliton molecule .
the parameters are exactly those of fig .
[ single - soliton ] , and only the initial condition has changed ( more powerful pulse ) .
it can be inferred that a soliton is in fact a pulse isolated from a sub - critical turing pattern , and that a soliton molecule corresponds to several of such pulses .
( a ) transient dynamics .
( b ) final pattern in the azimuthal direction .
( c ) 3d representation .
( d ) corresponding kerr comb .
( e ) soliton molecule formed with 5 solitons obtained with a different initial condition .
( f ) soliton molecule formed with 7 solitons obtained with a different initial condition .
, title="fig:",width=151 ] on the one hand , a supercritical bifurcation to turing patterns occurs when @xmath113 . in this case
, the unique equilibrium @xmath53 is stable when below the threshold pump power @xmath135 and unstable when @xmath136 , leading to the emergence of the turing pattern ( primary kerr comb in the spectral domain ) . above the pump threshold @xmath137 ,
the sidemodes of the kerr comb can continuously grow from infinitesimally small to significantly large as the pump is increased
. however , the equilibrium and the turing pattern are never simultaneously stable .
the formation of these supercritical patterns is displayed in fig . [ supercritical_turing ] .
it is interesting to note that the rolls are smooth , and yield a kerr comb characterized by isolated spectral lines with multiple fsr separation , the multiplicity being equal to the number of rolls . on the other hand ,
a sub - critical bifurcation to turing patterns arises when @xmath114 . here , as for the supercritical case , the equilibrium @xmath53 is stable for @xmath135 and unstable above .
however , kerr comb is possible in a small range _ below _ the threshold pump power @xmath115 .
hence , in the small interval below @xmath115 where the comb can be excited , a stable turing pattern does coexist with the equilibrium solution .
this situation creates a bistability and also induces hysteresis , as the dynamical state of the system will not be the same if the pump is adiabatically increased comparatively to when it is decreased .
the consequence of this bistability is that the transition from the equilibrium to the turing pattern is abrupt , and the sidemodes of the kerr comb can not be infinitesimally small as it was the case for the supercritical case .
figure [ subcritical_turing ] shows the formation of these sub - critical patterns . at the opposite of the super - critical rolls , small pedestals can be observed in this case , and the rolls also appear to be sharper
. in the context of kerr comb generation , the super- and sub - critical bifurcation have sometimes been referred to as _ soft _ and _ hard _ excitation modes , respectively @xcite .
it is also noteworthy that the turing patterns beyond @xmath138 are still sub - critical : however , the eigenvalue structure becomes more complex because this area in the parameter space @xmath79@xmath73 can encompass multiple - equilibria and the four types of bifurcations listed in sec .
[ bifurcationmap ] .
this is the area where bright cavity solitons and related structures can emerge .
we will study these complex sub - critical patterns the next sections .
( color online ) coexistence of two soliton molecules in the resonator .
the parameters are those of fig .
[ single - soliton ] . ( a ) initial conditions and final state .
( b ) corresponding kerr comb . , title="fig:",width=264 ] ( color online ) coexistence of two soliton molecules in the resonator .
the parameters are those of fig .
[ single - soliton ] .
( a ) initial conditions and final state .
( b ) corresponding kerr comb .
, title="fig:",width=264 ] the number of rolls in the turing pattern arising from the @xmath20 bifurcation at @xmath98 necessarily requires to account for the boundary conditions .
the reason is that in this case , the patterns fill the whole @xmath63-domain and the number of rolls along the azimuthal direction has to be an integer .
hence , the acknowledging for the modal structure of the patterns is there particularly relevant to understand this phenomenology .
( color online ) soliton breathers .
the parameters are @xmath139 , @xmath71 , and @xmath132 ( @xmath140 ) .
( a ) time - domain dynamics .
( b ) maximal and minimal pulse - shapes .
( c ) color - coded visualization of the time - domain dynamics of a soliton breather .
( d ) color - coded visualization of the time - domain dynamics of a complex structure corresponding to a higher - order soliton breather .
, title="fig:",width=173 ] ( color online ) soliton breathers .
the parameters are @xmath139 , @xmath71 , and @xmath132 ( @xmath140 ) .
( a ) time - domain dynamics .
( b ) maximal and minimal pulse - shapes .
( c ) color - coded visualization of the time - domain dynamics of a soliton breather .
( d ) color - coded visualization of the time - domain dynamics of a complex structure corresponding to a higher - order soliton breather . ,
title="fig:",width=128 ] ( color online ) soliton breathers .
the parameters are @xmath139 , @xmath71 , and @xmath132 ( @xmath140 ) . ( a )
time - domain dynamics .
( b ) maximal and minimal pulse - shapes .
( c ) color - coded visualization of the time - domain dynamics of a soliton breather .
( d ) color - coded visualization of the time - domain dynamics of a complex structure corresponding to a higher - order soliton breather .
, title="fig:",width=151 ] ( color online ) soliton breathers .
the parameters are @xmath139 , @xmath71 , and @xmath132 ( @xmath140 ) .
( a ) time - domain dynamics .
( b ) maximal and minimal pulse - shapes .
( c ) color - coded visualization of the time - domain dynamics of a soliton breather .
( d ) color - coded visualization of the time - domain dynamics of a complex structure corresponding to a higher - order soliton breather .
, title="fig:",width=151 ] according to the lle , a perturbation @xmath141 of the equilibrium ( flat solution ) @xmath53 obeys the linearized equation @xmath142 & = & - ( 1 + i\alpha ) \delta \psi + 2 i |\psi_{\rm e}|^2 \delta \psi + i \psi_{\rm
e}^2 \delta \psi^ * \nonumber \\ & & -i \frac{\beta}{2 } \frac{\partial^2}{\partial \theta^2}[\delta \psi ] \ , .
\label{perturb_lle}\end{aligned}\ ] ] following eq .
( [ psi_expansion ] ) , we can expand this perturbation according to the ansatz @xmath143 where @xmath144 corresponds to the eigennumber of the wgms with respect to the pumped mode @xmath145 .
the interacting eigenmodes can therefore be synthetically labelled as @xmath146 , @xmath147 , @xmath148 , the mode @xmath149 being the central mode . after inserting the ansatz of eq .
( [ ansatz_perturbation ] ) into eq .
( [ perturb_lle ] ) , we obtain an equation which can be used to perform an hermitian projection in order to track the individual dynamics of the modal perturbations @xmath150 . a projection onto a given mode @xmath151
consists in multiplying the equation by @xmath152 , and integrating the product from @xmath153 to @xmath154 with respect to @xmath63 .
the result of this projection yields two equations for the modal perturbations , which appear to be pairwise coupled according to @xmath155 = \left [ \begin{array}{ll } { \cal m } & { \cal n } \\ { \cal n}^ * & { \cal m}^ * \end{array } \right ] \left [ \begin{array}{l } a_{l } \\
a_{-l}^ * \end{array } \right ] \ , , \label{eq_matrix_perturb_a}\end{aligned}\ ] ] where the overdot stands for the derivative with respect to the dimensionless time @xmath156 , while @xmath157 the eigenvalues of the matrix in eq .
( [ eq_matrix_perturb_a ] ) define if a small signal perturbation ( noise ) in the modes @xmath158 increases or decreases with time . in particular , the real part of the leading eigenvalue ( the one with the largest real part ) can be viewed as the a _ gain _
parameter , which can be explicitly written as @xmath159 ^ 2 } \bigg\ } \ , , \label{eq_gain}\end{aligned}\ ] ] where @xmath55 .
( color online ) chaos .
( a ) 3d snapshot of a chaotic state for @xmath160 , @xmath71 , and @xmath161 .
( b ) corresponding kerr comb .
( c ) 3d snapshot of a chaotic state for @xmath160 , @xmath71 , and @xmath162 .
( d ) corresponding kerr comb .
( e ) 3d snapshot of a chaotic state for @xmath103 , @xmath71 , and @xmath163 .
( f ) corresponding kerr comb . ,
title="fig:",width=128 ] ( color online ) chaos .
( a ) 3d snapshot of a chaotic state for @xmath160 , @xmath71 , and @xmath161 .
( b ) corresponding kerr comb .
( c ) 3d snapshot of a chaotic state for @xmath160 , @xmath71 , and @xmath162 .
( d ) corresponding kerr comb .
( e ) 3d snapshot of a chaotic state for @xmath103 , @xmath71 , and @xmath163 .
( f ) corresponding kerr comb . , title="fig:",width=173 ] ( color online ) chaos .
( a ) 3d snapshot of a chaotic state for @xmath160 , @xmath71 , and @xmath161 .
( b ) corresponding kerr comb .
( c ) 3d snapshot of a chaotic state for @xmath160 , @xmath71 , and @xmath162 .
( d ) corresponding kerr comb .
( e ) 3d snapshot of a chaotic state for @xmath103 , @xmath71 , and @xmath163 .
( f ) corresponding kerr comb .
, title="fig:",width=128 ] ( color online ) chaos .
( a ) 3d snapshot of a chaotic state for @xmath160 , @xmath71 , and @xmath161 .
( b ) corresponding kerr comb .
( c ) 3d snapshot of a chaotic state for @xmath160 , @xmath71 , and @xmath162 .
( d ) corresponding kerr comb .
( e ) 3d snapshot of a chaotic state for @xmath103 , @xmath71 , and @xmath163 .
( f ) corresponding kerr comb . , title="fig:",width=173 ] ( color online ) chaos .
( a ) 3d snapshot of a chaotic state for @xmath160 , @xmath71 , and @xmath161 .
( b ) corresponding kerr comb .
( c ) 3d snapshot of a chaotic state for @xmath160 , @xmath71 , and @xmath162 .
( d ) corresponding kerr comb .
( e ) 3d snapshot of a chaotic state for @xmath103 , @xmath71 , and @xmath163 .
( f ) corresponding kerr comb . , title="fig:",width=128 ] ( color online ) chaos .
( a ) 3d snapshot of a chaotic state for @xmath160 , @xmath71 , and @xmath161 .
( b ) corresponding kerr comb .
( c ) 3d snapshot of a chaotic state for @xmath160 , @xmath71 , and @xmath162 .
( d ) corresponding kerr comb .
( e ) 3d snapshot of a chaotic state for @xmath103 , @xmath71 , and @xmath163 .
( f ) corresponding kerr comb .
, title="fig:",width=173 ] at threshold , we have already demonstrated that @xmath98 . on the other hand
, a mode @xmath128 is excited through mi when it experiences positive gain : the threshold gain can therefore be defined by @xmath164 .
hence , we deduce from the two preceding relations that at threshold , the two modes @xmath165 with @xmath166 are excited through mi .
this number also corresponds to the number of rolls that will be observed in the temporal domain .
this analysis corresponds to the one that has been performed in ref .
@xcite to explain the emergence of the so - called primary comb . from a more general perspective , the modes @xmath128 that can be directly excited by the pump are such that @xmath167 .
the modes @xmath168 for which the gain is maximal are referred to as the maximum gain modes ( mgm ) , and they are found through the condition @xmath169 , which yields @xmath170 figure [ mi_gain ] graphically displays how the mi gain leads to the emergence of a turing patterns .
when the system is pumped above threshold ( @xmath82 ) , two symmetric spectral bands are created around the pump .
the modes that are the most likely to arise from noise are those who have the largest gain , namely @xmath168 . near threshold , only two sidemodes ( around @xmath171 ) are generated and in the temporal domain : the flat background becomes unstable and leads to the emergence of the rolls , which correspond here to a sinusoidal modulation of the flat solution ( from this phenomenology was coined the term modulational instability ) . however , as the pump is increased , @xmath172 increases as well as it can be seen in fig .
[ mi_gain](b ) . but more importantly
, higher - order sidemodes ( harmonics ) are generated at eigennumbers @xmath173 , where @xmath174 is an integer number . as a consequence ,
the modulation in the time domain is not sinusoidal anymore , but gradually morphs into a train of sharply peaked pulses .
the existence of bright cavity solitons in nonlinear optical cavities is a well documented topic .
they arise as a balance between nonlinearity and anomalous dispersion ( which defines their shape ) , and a balance between gain and dissipation ( which defines their amplitude ) .
figure [ single - soliton ] displays the transient dynamics towards a cavity soliton . the initial condition here is a very narrow and small pulse , which grows and converges towards a soliton characterized by a narrow pulse - width and small pedestal oscillations .
this soliton is sub - critical as it emerged for a pump power for which the steady state is such that @xmath175 : hence , it does not emerge for arbitrarily small ( noisy ) perturbations of the intra - cavity background field .
in fact , it can be inferred that the soliton is a pulse that has been `` carved out '' of a sub - critical turing pattern .
this explains at the same time the sub - criticality and the pedestal oscillations , which are indeed observable in sub - critical turing patterns when the detuning @xmath79 is not too close to the critical value @xmath176 .
however , at the opposite of turing patterns , the bright soliton is a _
localized structure _ in the sense that it does not feel the boundaries when they are at a distance that is significantly larger than its pulse - width .
hence , the soliton of fig .
[ single - soliton](c ) dynamically behaves as if its background had an infinite extension .
the spectrum of this soliton as presented in fig .
[ single - soliton](d ) has a single - fsr spacing and displays hundreds of mode - locked wgms .
such solitons have also been observed experimentally in recent experiments @xcite .
it is also known that the spectral extension of this comb becomes larger as the pulses are narrower : this situation is observed when @xmath177 as discussed in sec .
[ influencedispersion ] .
when the parameters lead to the formation of super - critical turing patterns , the final steady - state is invariably the same patterns regardless of the initial conditions ( provided that there is no zero - energy mode at @xmath178 ) . for sub - critical structures ,
the situations is indeed very different , as the final output critically depends on the initial conditions .
as far as solitons are concerned , the initial conditions can lead to single - peaked pulses as displayed in fig .
[ single - soliton ] .
however , more energetic initial conditions can lead to the formation of multi - peaked solutions that which are here referred to as _
soliton molecules_. these molecules can be considered has a limited number of pulsed carved out of a sub - critical turing pattern .
figure [ soliton_molecule ] shows how the three - peaked soliton molecule is formed .
this dissipative structure is subcritical and localized , exactly as the single soliton .
the corresponding kerr comb is also characterized by a single - fsr spacing , but in this case , at the opposite of what is observed in the single - peaked soliton case , the spectrum displays a slow modulation . for a pulse - like
excitation , the number of solitons in a molecule can be controlled by the energy @xmath179 and it can be shown that the number of solitons in the molecule is governed by a snaking bifurcation .
figures [ soliton_molecule](e ) and ( f ) display soliton molecules with @xmath180 and @xmath181 elements , respectively , and larger numbers can be achieved as long as the molecule is blind to the finiteness of the @xmath63domain .
it is also noteworthy that different soliton molecules can coexist inside the disk .
such composite structures can be obtained for example using initial conditions as displayed in fig .
[ coexistence_molecules](a ) .
the corresponding kerr combs look noisy , and might even wrongfully be considered as `` chaotic '' : however , in the time domain , the pattern is perfectly periodic and deterministic .
genuinely chaotic spectra will be studied in sec .
[ chaos ] .
note that if @xmath182 individual and non - interacting bright solitons are exactly separated by an angle of @xmath183 in the @xmath63-domain , the resulting kerr comb will feature a multiple - fsr structure ( exactly like the spectra of turing rolls ) .
these kind of structures were presented in ref .
@xcite with dark solitons .
an interesting solution that can be obtained in the lle is the breather soliton @xcite .
it consists in a soliton whose amplitude varies periodically in time .
however , this period of the breathing is very low , and is of the order of the photon lifetime . in the spectral domain
, the comb corresponding to a single breathing soliton looks like the one of a normal ( steady ) soliton , except that there are modulation side - bands _ inside _ the modal resonance linewidths . as displayed in fig .
[ breathers](a ) and ( b ) , the breather soliton oscillates in time and it approximately keeps the same pulse - width . figure .
[ breathers](c ) shows the oscillating behavior at a larger timescale , and it can be observed that this breather soliton is a localized structure that is boundary - blind . on the other hand , fig .
[ breathers](d ) presents a higher order breather soliton with a complex structure consisting of multiple peaks that are not oscillating in synchrony .
actually , these soliton breathers can have a very wide variety of shapes and oscillation behaviors depending on the initial conditions .
( color online ) influence of the magnitude of the dispersion parameter @xmath77 .
( a ) case of turing patterns .
the influence of @xmath77 is essentially to change the number or rolls .
( b ) case of solitons .
the effect of @xmath77 here is to change the pulse - width .
, title="fig:",width=264 ] ( color online ) influence of the magnitude of the dispersion parameter @xmath77 .
( a ) case of turing patterns . the influence of @xmath77 is essentially to change the number or rolls .
( b ) case of solitons .
the effect of @xmath77 here is to change the pulse - width .
, title="fig:",width=264 ]
it is well known that chaos can potentially arise in any nonlinear system with at least three degrees of freedom .
the lle is indeed a highly nonlinear and infinite - dimensional system , and in fact , the phenomenology of interest ( kerr comb generation ) for us essentially relies on the nonlinearity . from a practical viewpoint , almost all high - dimensional and nonlinear systems display chaos when they are strongly excited . chaos in kerr combs
has been unambiguously spotted both theoretically and experimentally in ref .
@xcite , where the lyapunov exponent had been computed and shown to be positive under certain circumstances .
another study on this topic is ref .
@xcite . from the preceding sections , at least two routes to chaos can be identified in this system .
the first route corresponds to unstable turing patterns . in figs .
[ chaos](a ) and ( b ) shows that in that case , the kerr comb is made of very strong spectral lines corresponding to the primary comb , and apparent spectral lines standing in between , and which are the signature to what was referred to as the secondary comb in refs .
hence , in this case , the route to chaos as the pump power @xmath73 is increased is a sequence of bifurcations starting with the primary comb which becomes unstable and leads to the emergence of a secondary comb ; later on , higher - order combs are sequentially generated until a fully developed chaotic state is reached , as it can be seen in figs .
[ chaos](c ) and ( d ) . the second route to chaos corresponds to unstable solitons . here , as the pump power
is increased , the solitons become unstable and the system enters into a `` turbulent '' regime characterized by the pseudo - random emergence of sharp and powerful peaks , as it can be seen in figs . [ chaos](e ) .
this kind of chaos is very likely give birth in the wgm resonator to the extreme events referred to as _ rogue waves_.
the bifurcation map displayed in fig .
[ figurebifnotatscale ] in the @xmath79@xmath184 plane disregards the effect of the magnitude of the dispersion .
indeed , the effect induced by the value of @xmath185 depends on the localized or non - localized nature of solution under study . more precisely , if we consider both super- and sub - critical turing patterns ( non - localized structures ) the effect of decreasing @xmath185 is straightforward as it increases the number of rolls according to eq .
( [ def_l_mgm ] ) just above the pump
. however , as the pump is increased beyond the bifurcation , such a decrease of @xmath185 also reduces the pulse - width of the individual turing rolls in the sub - critical case . as far as solitons are concerned ( localized structures ) , the effect of reducing @xmath185 is essentially to decrease the pulse width of the solitons .
sub - critical turing patterns and solitons have the same behavior in this regards , and this is a direct consequence of the fact that they are intimately connected from a topological point of view .
it should be noted once again that reducing the magnitude of second order gvd to arbitrarily small values can increases the relevance of higher - order dispersion terms in the lugiato - lefever model @xcite .
a general consequence of a decrease in @xmath185 is that as the patterns have a narrower pulse - width , and as we commented in part i @xcite , this will generally allow for the excitation of a large number of solitons in the cavity .
accordingly , the corresponding kerr comb spectra will also display a higher complexity .
in this article , we have investigated the bifurcation structure of the lugiato - lefever equation which is used to model kerr comb generation using wgm resonators .
we have focused on the case of anomalous dispersion and our analysis has evidenced a plethora of possible steady states .
turing patterns arise in the system through a @xmath20 bifurcation , and they can be characterized by their super- or sub - critical nature .
the threshold for these patterns can be analytically determined , as well as the number of rolls .
we have also been able to determine analytically the parametric gain allowing modulational instability .
we have also shown how the super- or sub - critical nature of the patterns affect their temporal dynamics and the way they can be excited .
our investigations have also enabled to analyze the formation of solitons .
we have shown that more complex structures , referred to a soliton molecules , can be generated as well in the system , and can eventually coexist along the azimuthal direction of the resonator .
breather solitons have been analyzed as well , and we have also investigated the emergence of chaos in the system . the complexity of the related kerr combs has also been studied and discussed
. future work will be devoted to the investigation of the effect of higher - order nonlinearity and dispersion @xcite , and to the tailoring of the spectral characteristics of the combs for various technological applications @xcite .
the authors acknowledge financial support from the european research council through the project nextphase ( erc stg 278616 ) .
they would also like to thank mariana haragus for very insightful comments and suggestions .
authors would also like to acknowledge for the support of the _ msocentre de calcul de franche - comt_. 99 c. godey , i. balakireva , a. coillet and y. k. chembo , _ phys .
e _ , submitted ( 2011 ) .
a. b. matsko , a. a. savchenkov , and l. maleki , _ opt .
lett . _ * 37 * , 43 ( 2012 ) .
a. coillet , i. balakireva , r. henriet , k. saleh , l. larger , j. m. dudley , c. r. menyuk , and y. k. chembo , to be published in _ ieee photonics journal _ , ( 2013 ) . t. j. kippenberg , r. holzwarth , and s. a. diddams , _ science _ * 322 * , 555 ( 2011 ) .
f. ferdous , h . miao , d. e. leaird , k. srinivasan , j. wang , l. chen , l. t. varghese and a. m. weiner , _ nature photonics _ * 5 * , 770 ( 2011 ) .
a. a. savchenkov , a. b. matsko , w. liang , v. s. ilchenko , d. seidel and l. maleki , _ nature photonics _ * 5 * , 293 ( 2011 ) .
s. b. papp and s. a. diddams , _ phys .
a _ * 84 * , 053833 ( 2011 ) .
p. delhaye , s. b. papp , and s. a. diddams , _ phys .
lett . _ * 109 * , 263901 ( 2012 ) .
t. herr , k. hartinger , j. riemensberger , c. y. wang , e. gavartin , r. holzwarth , m. l. gorodetsky , and t. j. kippenberg , _ nature photonics _ * 6 * , 480 ( 2012 ) . c. y. wang , t. herr , p. delhaye , a. schliesser , j. hofer , r. holzwarth , t. w. haensch , n. picqu , and t. j. kippenberg , _ nature communications _ * 4 * , 134 ( 2013 ) .
a. coillet , r. henriet , k. p. huy , m. jacquot , l. furfaro , i. balakireva , l. larger , and y. k. chembo , to be published in _ j. vis .
_ , ( 2013 ) .
a. a. savchenkov , a. b. matsko , d. strekalov , m. mohageg , v. s. ilchenko , and l. maleki , _ phys .
lett . _ * 93 * , 243905 ( 2004 ) .
y. k. chembo , d. v. strekalov , and n. yu , _ phys .
* 104 * , 103902 ( 2010 ) . y. k. chembo and n. yu , _ phys .
a _ * 82 * , 033801 ( 2010 ) . i. h. agha , y. okawachi , and a. l. gaeta , _ opt . express _
* 17 * , 16209 ( 2009 ) .
a. b. matsko , a. a. savchenkov , w. liang , v. s. ilchenko , d. seidel , and l. maleki , _ opt .
lett . _ * 36 * , 2845 ( 2011 ) . y. k. chembo and c. r. menyuk , _ phys . rev
. a _ * 87 * , 053852 ( 2013 ) .
s. coen , h. g. randle , t. sylvestre , and m. erkintalo , _ opt .
lett . _ * 38 * , 37 ( 2013 ) . l. a. lugiato and r. lefever , _ phys
lett . _ * 58 * , 2209 ( 1987 ) .
a. coillet , r. henriet , p. salzenstein , k. phan - huy , l. larger , and y. k. chembo , _ ieee sel .
quantum electron .
_ * 19 * , 6000112 ( 2013 ) .
m. haragus and g. iooss , `` local bifurcations , center manifolds , and normal forms in infinite - dimensional dynamical systems '' , springer ( 2010 ) .
a. m. turing , _ phil .
trans . of the r. soc .
b _ * 237 * , 37 ( 1952 ) . l. a. lugiato , guest editor , `` special issue on cavity solitons '' , _ ieee j. quantum electron .
_ * 39 * , feb
n. akhmediev and a. ankiewicz , editors , `` dissipative solitons : from optics to biology and medicine '' , lect .
notes phys .
* 751 * ( 2008 ) .
a. j. scroggie , w. j. firth , g. s. mcdonald , m. tlidi , r. lefever , and l. a. lugiato , _ chaos , solitons , and fractals _ * 4 * , 1323 ( 1994 ) . m. santagiustina , p. colet , m. san miguel , and d. walgraef , _ phys .
_ * 79 * , 3633 ( 1997 ) .
w. j. firth , g. k. harkness , a. lord , j. m. mcsloy , d. gomila , and p. colet , _ j. opt
. soc . am .
b _ * 19 * , no . 4 , pp . 747752 , apr 2002 .
d. gomila , m. a. matias , and p. colet , _ phys .
_ * 94 * , 063905 ( 2005 )
. t. miyaji , i. ohnishi , and y. tsutsumi , _ physica d _ * 239 * , 2066 ( 2010 ) .
g. kozyreff , _ physica d _ * 241 * , 936 ( 2012 ) .
a. b. matsko , a. a. savchenkov , v. s. ilchenko , d. seidel , and l. maleki , _ phys
a _ * 85 * , 023830 ( 2012 ) . t. herr , v. brasch , j. d. jost , c. y. wang , n. m. kondratiev , m. l. gorodetsky and t. j. kippenberg , _ arxiv:1211.0733v3 _ ( 2013 ) .
a. b. matsko , a. a. savchenkov , and l. maleki , _ opt .
_ * 37 * , 4856 ( 2012 ) .
f. leo , l. gelens , p. emplit , m. haelterman , and s. coen , _ optics express _ _ 21 _ , 9180 ( 2013 ) .
a. b. matsko , w. liang , a. a. savchenkov , and l. maleki , _ opt .
_ * 38 * , 525 ( 2013 ) .
p. colet , m. a. matias , l. gelens , d. gomila , _ arxiv:1305.6801v1 _ ( 2013 ) .
l. gelens , m. a. matias , d. gomila , t. dorissen , p. colet , _ arxiv:1305.6804v1 _ ( 2013 )
. y. k. chembo and n. yu , _ opt . lett . _ * 35 * , 2696 ( 2010 ) . i. s. grudinin , l. baumgartel , and n. yu , _ opt .
express _ * 20 * , 6604 ( 2012 ) .
m. r. e. lamont , y. okawachi , a. l. gaeta , _ arxiv:1305.4921 _ ( 2013 ) . | we present a stability analysis of the lugiato - lefever model for kerr optical frequency combs in whispering gallery mode resonators pumped in the anomalous dispersion regime .
this article is the second part of a research work whose first part was devoted to the regime of normal dispersion , and was presented in ref .
@xcite .
the case of anomalous dispersion is indeed the most interesting from the theoretical point of view , because of the considerable variety of dynamical behaviors that can be observed . from a technological point of view
, it is also the most relevant because it corresponds to the regime where kerr combs are predominantly generated , studied , and used for different applications . in this article
, we analyze the connection between the spatial patterns and the bifurcation structure of the eigenvalues associated to the various equilibria of the system .
the bifurcation map evidences a considerable richness from a dynamical standpoint .
we study in detail the emergence of super- and sub - critical turing patterns in the system .
we determine the areas were bright isolated cavity solitons emerge , and we show that soliton molecules can emerge as well .
very complex temporal patterns can actually be observed in the system , where solitons ( or soliton complexes ) co - exist with or without mutual interactions .
our investigations also unveil the mechanism leading to the phenomenon of breathing solitons .
two routes to chaos in the system are identified , namely a route via the so called secondary combs , and another via soliton breathers .
the kerr combs corresponding to all these temporal patterns are analyzed in detail , and a discussion is led about the possibility to gain synthetic comprehension of the observed spectra out of the dynamical complexity of the system . |
in these supplementary notes , we describe effects of collisions of unidirectional charged vortex rings ( cvrs ) due to the small variations , @xmath110 , of their velocities around the mean value @xmath111 .
any such collision of primary cvrs generally results in secondary cvrs of quite different size hence depleting the number of the primary cvrs .
we consider purely one - dimensional trajectories of cvrs , @xmath91 , between the injector at @xmath6 and collector at @xmath9 @xcite .
we assume a uniform electric field @xmath112 , a negligible initial radius of cvrs , @xmath113 , and a constant logarithmic parameter @xmath114 ( where @xmath115 is the core radius ) . within the time of flight from the injector to collector , @xmath57 , the lengthening of the beam , caused by the range of velocities @xmath116 , does not exceed the duration of injection @xmath55s and is , hence , neglected .
the velocity and energy of a quantized vortex ring of radius @xmath0 carrying one electron of charge @xmath49 : @xmath117 @xmath118 where @xmath119g@xmath16 is the density of superfluid helium and @xmath120@xmath121s@xmath122 is the quantum of circulation .
the radius grows linearly with @xmath5 , @xmath123 where the radius of cvrs at collector , @xmath9 , is @xmath124 the time of flight between @xmath6 and @xmath36 is @xmath125 i.e. the trajectory of primary cvrs is @xmath126 where the time of flight from injector to collector is @xmath127 the number density of primary rings in the beam @xmath90 , along the trajectory @xmath91 , evolves according to : @xmath131 where @xmath93 is the rate of losses due to ring - ring collisions . for small @xmath4 ,
at which @xmath132 , the number of cvrs is conserved , and the solution is @xmath133 , i.e. @xmath134 .
experimentally , the value of constant @xmath135 can be determined from the magnitude of the peak of the collector current due to primary cvrs , @xmath136 where @xmath137 and @xmath138 are the radius and transparency of the collector grid , @xmath139 and @xmath140 .
we arrive at @xmath141 the density at collector is @xmath142 thus , in this regime of conserved primary cvrs , their density can be varied by changing either the intensity of injection ( as measured by @xmath96 ) or drive voltage @xmath56 . for each cvr ,
the probability of a collision with another cvr per unit time is @xmath143 , where @xmath144 ( as @xmath145 ) , and the cross - section is @xmath146 , where @xmath147 .
naive geometric considerations suggest @xmath148 , although further research into this problem is necessary @xcite .
if each collision effectively removes two primary cvrs from the coherent beam , the total number of removed cvrs per unit time and unit volume : @xmath149 after using eqs.[nablav]&[tau1 ] , and @xmath150 from the above expression , eq.[rateequation ] becomes : @xmath151 its asymptotic solution is @xmath152 at @xmath82 , @xmath153 the rhs is independent of both the injection intensity and @xmath56 . for @xmath154 and @xmath155@xmath12 m , it is equal to @xmath156@xmath105 , which is a factor of 8 greater than the experimental value of @xmath157@xmath105 .
there might be several reasons for such a discrepancy .
firstly , the cross - section for the effective removal of cvrs from the coherent beam of similar orientation and velocities might be several times greater than the geometric guess @xmath158 .
secondly , in this model we restricted all cvrs to motion along the @xmath5-axis with only a spread of radii ; in reality they also have a small random spread of directions of motion caused by the conserved transverse component of the impulse of the initial cvrs our estimates show that its contribution to the frequency of collisions is comparable with that calculated here .
another process that should further limit the density of primary cvrs is the removal of all cvrs that bump into the secondary vortex rings formed upon any binary collision discussed above .
large secondary cvrs appear with each primary collisions at a rate , per unit volume and time , @xmath159 as these larger rings are considerably slower than the primary ones , they block all the primary cvrs from behind within the effective cross - section @xmath102 , where @xmath160 , and the upper limit on its geometrical value is @xmath161 @xcite . the typical number of primary cvrs lost in such a capture ( the `` multiplication factor '' ) is @xmath162 .
the rate of losses per unit volume and time is hence @xmath163 with @xmath164 , eq.[rateequation ] becomes @xmath165 its asymptotic solution is : @xmath166 for @xmath82 this gives @xmath167 independent of the injection intensity and @xmath56 . for @xmath100 , @xmath103 , @xmath155@xmath12 m , @xmath55s and @xmath46 cm
, this produces @xmath168@xmath105 , which is only factor of 2 greater than the experimental value .
furthermore , as we only considered the effect of longitudinal variations of the velocities @xmath169 of primary cvrs but disregarded the comparable effect from their transverse velocities , the agreement is quite good . to conclude
, the solutions , @xmath170 , found for either of the simple models of removal of primary vortex rings upon collisions , eq.[nr1 ] and eq.[nr2 ] , qualitatively agree with our experimental observations .
moreover , the latter is actually in a reasonable quantitative agreement . thus multiple pile - ups explain the observed universal value of the product @xmath107 when cvrs collide frequently due to their large numbers and large radii .
further experiments with varyable flight path @xmath171 , duration of injection @xmath54 and distribution of cvrs radii @xmath172 , as well as numerical simulations of the interaction between cvrs , will help improve our understanding of the microscopic processes that occur within beams of unidirectional vortex rings .
m. leadbeater , t. winiecki , d. c. samuels , c. f. barenghi , and c. s. adams , phys .
* 86 * , 1410 ( 2001 ) .
m. leadbeater , d. c. samuels , c. f. barenghi , and c. s. adams , phys .
a * 67 * , 015601 ( 2003 ) . in our experiments , cvrs motion very near the injector is not quite one - dimensional , but the number of reconnections is small there , hence the end result should not be strongly affected by this approximation .
our preliminary calculations of the interaction between two parallel vortex rings of slightly different radii show that reconnections occur with the cross - section @xmath174 , while a noticible perturbation of the directions of motion and velocities ( even without reconnections ) occurs with the cross - section even larger than @xmath175 . | collisions in a beam of unidirectional quantized vortex rings of nearly identical radii @xmath0 in superfluid @xmath1he in the limit of zero temperature ( 0.05k ) were studied using time - of - flight spectroscopy .
reconnections between two primary rings result in secondary vortex loops of both smaller and larger radii .
discrete steps in the distribution of flight times , due to the limits on the earliest possible arrival times of secondary loops created after either one or two consecutive reconnections , are observed .
the density of primary rings was found to be capped at the value @xmath2 independent of the injected density .
this is due to collisions between rings causing piling - up of many other vortex rings .
both observations are in quantitative agreement with our theory .
turbulence appears in various systems : fluids , plasmas , interstellar matter with common properties such as the existence of long - lived regions of concentrated vorticity , whose reconnections facilitate the evolution of the flow field and redistribution of the kinetic energy between length scales .
a paradigm of an isolated vortical structure is a vortex ring @xcite , and their pair interactions are a testbed of the physics of vortex reconnections .
numerical simulations of collisions of two vortex rings predict various outcomes : either a single ring or several rings , depending on the initial conditions @xcite .
there were experimental attempts to visualize these processes in classical fluids @xcite ; however , they are often hard to interpret because of the inevitable decay , core instabilities and poor characterization of vortex rings in viscous fluids . quantized vortex rings in superfluids have an advantage because they are slender and stable , and can hence be well - characterized quantitatively @xcite .
recently , there were many theoretical investigations into reconnections of quantized vortex lines @xcite .
in particular , for acute angles between two antiparallel reconnecting vortex lines the generation of a cascade of small vortex rings was predicted @xcite .
the outcome is reminiscent of that of the crow instability @xcite of antiparallel vortices observed in air .
experimentally , reconnections of vortex lines in superfluid @xmath1he have been visualized @xcite but only at high temperatures when vortex motion is damped .
reconnections of vortex loops comprising a vortex tangle , i.e. _ quantum turbulence _ ( qt ) @xcite , especially those leading to the emission of vortex rings @xcite , are an important mechanism of redistributing energy towards smaller length scales in qt @xcite . in superfluid
@xmath3he - b @xcite and @xmath1he @xcite , collisions and subsequent reconnections , in a dense beam of vortex rings generate qt . longer and more intensive beams of rings result in tangles that show large - scale velocity fluctuations @xcite and the late - time decay @xcite , both characteristic of classical turbulence .
this implies the existence of the _ inverse cascade _ of energy from the small length scales ( of order ring radii ) into which the initial energy is injected up to the size of the resulting tangle .
it was speculated @xcite that the inverse cascade might be maintained by the merger of pairs of rings into larger loops . yet , no direct quantitative observations of ring - ring reconnections have been reported so far .
this letter reports the first quantitative observations of either one or two consecutive reconnections , and the discovery of the ensuing universal state of depleted density within a beam of unidirectional quantized vortex rings all of similar radii , with their number density @xmath4 and radius @xmath0 under our control .
the resulting mechanism of seeding the large - scale velocity fluctuations out of a seemingly random beam of vortex rings is suggested . in our experiments , to create vortex rings of a required size and to detect their arrival , each was tagged by an excess electron trapped on the vortex core .
applying an electric field along the @xmath5-axis allowed small seed charged vortex rings ( cvrs ) , injected at @xmath6 , to grow to the desired radius @xmath7 , and also to trace the location of the reconnection process that resulted in small secondary charged vortex rings .
the radius of a quantized vortex ring is directly related to its self - induced velocity , @xmath8 which determines its arrival time at the collector at @xmath9 . the numbers of primary and secondary vortex rings as a function of their radii could be extracted from the time - dependence of the collector current @xmath10 through time of flight spectroscopy .
the radius of primary rings at the collector , @xmath11 , was varied within 16@xmath12 m , with number density @xmath13 between @xmath14 and @xmath15@xmath16 ; while the mean radius of the seed cvrs was estimated as @xmath17@xmath12 m .
the energy of a cvr , subject to a potential @xmath18 , is @xmath19 in the absence of dissipation at @xmath20k .
the velocity and energy depend on @xmath0 @xcite , @xmath21 where @xmath22 is the circulation quantum , @xmath23 is the density of superfluid , @xmath24 and @xmath25@xcite .
a deeper insight can be gained within the approximation for constant @xmath26 and uniform field @xmath27 .
the radius of a cvr then grows linearly with @xmath5 , @xmath28 and the time for a cvr to travel from @xmath6 to @xmath9 , @xmath29 increases with energy @xmath30 because cvrs slow down as they expand ( see eq.[ve ] ) . in what follows , unless specified , we will be using the approximation @xmath31 . with increasing density of cvrs ,
collisions become more frequent .
these collisions are caused by small fluctuations in the direction and magnitude of the rings velocities mainly due to the variations in initial radii @xmath32 and direction of the seed cvrs when injected at @xmath6 .
along with reconnections upon a direct collision , hydrodynamic dipole - dipole interactions between neighboring cvrs ( that grows in strength with increasing @xmath4 and @xmath0 and are hence the strongest near the collector at @xmath33 ) affect the cvr s velocities . the coulomb repulsion between neighboring cvrs of @xmath341@xmath12 m is much weaker than their hydrodynamic interaction . a reconnection of two cvrs results in secondary vortex loops , which are generally non - circular .
the two trapped electrons are now carried by either one or by two ( if any ) of the secondary rings .
one special case allows an exact analysis of the consecutive trajectory of one of the electrons when a secondary cvr has a small initial radius ( @xmath35 ) .
then its initial deformation and direction of motion can be disregarded , because , under the pull of the electric field , it quickly gains sufficient energy and impulse along the @xmath5-direction , so to a good accuracy can be treated as a circular vortex ring @xcite .
if such singly - charged loops are created after collisions at some @xmath36 , their arrival at the collector ( @xmath9 ) at time @xmath37 $ ] will be earlier than of any other cvrs with either larger initial size ( slower ) or double charge ( more energetic , hence , slower ) .
the earliest arrival time , @xmath38 will be for collisions at @xmath39 . furthermore , if a secondary small ring grows and then reconnects with another vortex loop at some point @xmath40 ( @xmath41 ) , and this creates a new small singly - charged ring , the latter will arrive at the collector at time @xmath42 $ ] . the earliest arrival , at time @xmath43 of these second - generation secondary cvrs will correspond to two reconnections at @xmath44 and @xmath45 . the experimental cell @xcite , a cube - shaped volume of side @xmath46 cm , was filled with isotopically - pure liquid @xmath1he @xcite at pressure 0.1bar and temperature 0.05k ( see inset in fig.1 ) .
seed cvrs were injected through a gridded opening in the center of one plate .
they then traveled along the axis of the container ( @xmath5-axis ) towards the center of the opposite plate to the collector electrode , placed behind a frisch grid of radius @xmath47 mm and geometric transparency @xmath48 .
all currents and potentials are quoted with the opposite sign as if electrons had _ positive _ charge @xmath49 .
cvrs were subject to the propelling field set by potentials of plates @xmath50 and @xmath51 , thus gaining energy @xmath30 while travelling between the injector and collector grids @xcite .
the dependence @xmath18 was close to the linear @xmath52 ( see inset in fig.2 ) .
the seed cvrs resulted from reconnections within the dense vortex tangle , generated by the current of electrons emitted from a tungsten tip @xcite behind the injector grid through the voltage @xmath53 .
these seed cvrs are injected in a broad range of angles ; however , the impulse gained from the strong driving field quickly forces them to travel in nearly the same @xmath5-direction with a relatively narrow distribution of radii .
the intensity and duration of the injected pulse were controlled by adjusting @xmath53 and its duration @xmath54 ( all data presented here are for @xmath55s ) .
for the same @xmath53 and @xmath54 , the total charge injected through the grid was increasing with increasing drive voltage @xmath56 nearly linearly for all studied voltages . to quantify the time of flight @xmath57 of cvrs
, we take the time interval between the middle of the tip voltage pulse and the position of the maximum of @xmath10 ( and subtract the electronics response time of 0.03s ) .
v , but different tip voltages @xmath53 .
the theoretical arrival time for primary cvrs , @xmath57 ( eq.[tau1 ] ) , and of the earliest arriving secondary cvrs of first generation , @xmath58 ( eq.[tau2 ] ) , and of second generation , @xmath59 ( eq.[tau3 ] ) , are shown by arrows .
inset : experimental cell with electric field pattern.,width=264 ] typical records of the collector current , @xmath10 , following the injection of a pulse of cvrs are shown in fig.[figivstvarutip2 ] .
these are all for the same drive voltage @xmath60v but several different injection currents .
there is a well - defined peak at time @xmath61s corresponding to the arrival of primary cvrs . with increasing density of cvrs
, this peak initially grows in magnitude while maintaining its position , @xmath57 , and width , @xmath62 . at higher numbers of injected cvrs
, however , a broad pedestal begins to grow , coexisting with the original peak ( whose magnitude is now saturated ) .
this broad peak - pedestal is due to the secondary cvrs that result from collisions between primary cvrs .
the current at @xmath63 reflects the arrivals of larger secondary vortex loops , while that at the earlier arrival times @xmath64 are from smaller secondary cvrs .
a sharp step builds up at @xmath65 , coinciding with the earliest possible arrival of the first generation of secondary cvrs ( eq.[tau2 ] ) . at the highest intensity of injection ,
another sharp step begins to form at @xmath66 , corresponding to the earliest possible arrival of the second generation of secondary cvrs ( eq.[tau3 ] ) .
v , but different values of @xmath56 .
the right , middle and left arrows of the corresponding color point at the arrival time of primary cvrs , @xmath57 , and the theoretical earliest arrival times for secondary cvrs of first ( eq.[tau2 ] ) and second ( eq.[tau3 ] ) generation , respectively .
the values of @xmath57 are defined as the positions of sharp peaks , except for @xmath67v where the peak due to primary cvrs is swamped by the broad pedestal due to secondary cvrs , hence the theoretical value of @xmath57 ( see fig.3 ) is used . for @xmath68v and 67.5v ,
the signal is too faint for the step at @xmath59 to be observable .
inset : the electrostatic potential @xmath18 along the cell s axis.,width=264 ] in fig.[figivstvaru ] , we show @xmath10 , similar to those in fig.[figivstvarutip2 ] , but now for the same tip voltage @xmath69v and four different drive voltages @xmath56 . with increasing @xmath56 , the position of the peak @xmath57 increases as expected for isolated cvrs , eq.[tau1 ] .
the peak s magnitude @xmath70 initially grows with @xmath56 but then , above @xmath71v , decreases even though the total collected charge @xmath72 keeps increasing .
simultaneously , the broad pedestal due to secondary cvrs progressively overgrows the primary peak until completely swamping it at @xmath67v .
the sharp steps due to the earliest possible arrivals of secondary cvrs of first generation at @xmath73 ( eq.[tau2 ] ) and second generation at @xmath74 ( eq.[tau3 ] ) are labeled by arrows .
we thus obtained quantitative evidences of either single or two consecutive reconnections of cvrs during their motion from the injector to collector .
furthermore , the substantial contribution to the collector current right after the cut - offs indicates that very small cvrs are created with high probability .
this might contradict the expectations that reconnections result in vortex loops of size _
comparable _ to the radius of curvature of the initial vortex lines @xcite , but would support the picture of a cascade of small vortex rings created by large - amplitude kelvin waves generated after a reconnection of nearly antiparallel vortex lines when dissipation is small @xcite . vs. @xmath56 .
closed symbols correspond to resolvable sharp peaks due to primary cvrs ; open symbols are for the broadened peaks due to secondary cvrs where sharp peaks due to primary cvrs are no longer visible .
vertical bars indicate the peak width at half - maximum . the solid line shows the theoretical arrival times
@xmath57 ( eqs.12 ) for @xmath75@xmath12 m . the dashed line shows the approximate solution for @xmath76 and effective @xmath77@xmath12 m ( eq.[tau1 ] ) .
cartoons illustrate the expansion and progression of : ( left ) isolated primary ( red ) cvrs at low density , and ( right ) a reconnection resulting in secondary ( blue ) cvrs.,width=264 ] in fig.[figtvsu ] , the experimental arrival times @xmath78 for several intensities of injection are plotted . for small drive voltages @xmath56 ( i.e. when the radii of cvrs @xmath11 and density of cvrs @xmath13 are small ) the experimental points agree with the theory for isolated cvrs ( from eqs.[ve][r_appr ] ) .
to characterize the range of the distribution of times of flight , the vertical bars show the width of @xmath10 at the @xmath79 level .
one can see that at small @xmath56 the width is constant , being equal to the injection duration @xmath55s . above a certain value of @xmath80v ( that decreases with increasing injection intensity , @xmath53 )
, the collector pulse broadens and the position of the maximum of @xmath10 no longer agrees with the theoretical prediction for isolated cvrs ( this coincides with the complete disappearance of the sharper peak due to primary cvrs , as on the trace for @xmath67v in fig.[figivstvaru ] ) ; secondary cvrs dominate @xmath10 in these conditions of high @xmath4 and @xmath0 . vs. their radius @xmath0 at @xmath9 measured for several different drive voltages @xmath56 ( top axis ) and injector tip voltages , @xmath53 ( see legend ) .
, width=264 ] the density of primary rings reaching the collector , @xmath81 , can be found from the value of the collector current at its maximum ( but only for the records @xmath10 that have a sharp peak at @xmath82 , dominant over the pedestal due to the secondary cvrs ) , @xmath83 where the relation @xmath84 ( from eqs.[ve][r_appr ] ) has been used . in fig.[fignvsr ] , we plot @xmath85 ( calculated from the experimental values of @xmath70 using eq.[im ] ) vs. @xmath86 , the radii of primary cvrs near the collector , ( calculated from eq.[r_appr ] with @xmath87 ) .
again , there are two regimes : at low @xmath88 , @xmath85 increases with an increase of either @xmath53 or @xmath56 , but at large @xmath88 , @xmath85 becomes a _ decreasing _ function of @xmath56 , independent of @xmath53 .
yet , the total injected charge ( as measured by the integral of the collector current ) increases monotonically at all @xmath56 .
i.e. , only at small @xmath88 and @xmath85 , the cvr s density can be controlled by varying the injection current . at high @xmath88 , only a small fraction of charge arrives with primary cvrs of now _ universal _
density @xmath89 ; the rest ( secondary cvrs ) contribute to the broad pedestal in @xmath10 ( fig.[figivstvarutip2][figivstvaru ] ) .
the density of primary rings in the beam @xmath90 , along the trajectory @xmath91 , evolves according to eq.11 in @xcite : @xmath92 where @xmath93 is the rate of losses , per unit volume and time , due to ring - ring collisions . for
small injected density and radii , the collisions can be neglected , @xmath94 , and the solution for the density near collector is @xmath95 i.e. it can be varied by changing either the injected density of cvrs ( characterized by @xmath96 ) or drive voltage @xmath56 as observed in the experiment . when collisions become rife at higher densities and radii , accounting for the removal of primary cvrs due to their binary collisions results in @xmath97 where we used @xmath98@xmath12 m and the geometric cross - section for collisions @xmath99 with @xmath100 .
furthermore , the subsequent removal of primary cvrs that bump into the slower loops @xcite left after the described collisions of primary cvrs will result in the solution , @xmath101 where the geometric cross - section for these pile - ups is estimated as @xmath102 with @xmath103 . both eq.[nr1 ] and eq.[nr2 ] reproduce the experimental universal dependence @xmath104@xmath105 qualitatively , while eq.[nr2 ] is actually quite close _ quantitatively _ ( our 1-dimensional model underestimates @xmath106 by disregarding the transverse component of the relative motion of primary cvrs and hence overestimates the value of @xmath107 by a factor of 2 or so @xcite ) .
the fact of occasional piling - up of many primary cvrs , in turn , helps to explain the appearance of large - scale velocity fluctuations in the ensuing vortex tangle . in the initial random beam of primary cvrs ,
the fluctuations of the coarse - grained velocity on the length scales greater than @xmath108 ( `` quasi - classical '' flow ) are small ; the energy spectrum is concentrated around the small length scale of order @xmath0 .
the small secondary vortex rings , observed in this work , and kelvin waves excited by reconnections are evidence of the direct cascade of energy towards smaller length scales @xcite .
however , following any of the pile - ups of many vortex rings , strong fluctuations of the coarse - grained velocity field on the `` quasi - classical '' length scales @xmath109 are being created .
this is the inverse cascade of energy in this strongly anisotropic system @xcite .
to summarize , we obtained quantitative evidence for collisions and reconnections of pairs of unidirectional vortex rings of similar radii that result in the creation of vortex loops of unequal size , including many small ones .
we observed discrete steps at the time dependence of the collector current , that correspond to the earliest arrivals of the first and second generations of secondary cvrs .
as each collision can cause a removal of many primary vortex rings , increasing the density of injected cvrs results in a new state in which the density of primary vortex rings is maintained at the critically depleted level independent of their initial density .
the larger loops produced in the collisions become the seeds of quasi - classical qt with large - scale flow structures , which appear out of a seemingly random beam of small quantized vortex rings .
this work was supported by the engineering and physical sciences research council ( grants no .
gr / r94855 , ep / h04762x/1 and ep / i003738/1 ) . |
quantum teleportation is one of the fundamental applications of entanglement in quantum information @xcite . in the teleportation process , a maximally
entangled state of two systems , @xmath2 and @xmath1 , is distributed to the sender and the receiver .
the sender then combines the input system @xmath0 with system @xmath2 of the entangled pair to perform a bell measurement , which is described by a projection on a maximally entangled state of @xmath0 and @xmath2 . as a result of this projection , the unknown quantum state of @xmath0
is transferred to the remote system @xmath1 , which now carries quantum coherences that correspond precisely to the quantum coherences of @xmath0 before the bell measurement .
although the hilbert space formalism of the teleportation process is mathematically simple and consistent , it describes non - local effects of measurements on an entangled state .
the origin of these effects is not obvious , since the quantum measurement process combines the physical disturbance of properties by interaction with changes to the statistics caused by additional information gained in the process . it may therefore be possible to explain the non - local effects of teleportation entirely in terms of subjective information gain , as indicated by the description of continuous variable quantum teleportation using wigner functions @xcite . in that case
, the analogy between wigner functions and classical phase space distributions can be used to construct a classical model of teleportation , where the bell measurement merely uncovers the correct value of the phase space distance that describes the relation between the physical properties of input system @xmath0 and output system @xmath1 before the bell measurement .
the projection of the state vector thus appears as a process that modifies the information about system @xmath1 without altering any of its physical properties .
this viewpoint may also apply to the general quantum teleportation of @xmath3-level systems , since there is no experimentally accessible evidence of a change of the physical properties in @xmath1 caused by the performance of the bell measurement in @xmath0 @xcite .
the standard formalism of quantum mechanics does not provide any direct answer to the question of how and when the physical properties are transferred from system @xmath0 to system @xmath1 in a quantum teleportation .
it has even been suggested that teleportation corresponds to time travel , where the system @xmath2 is the channel that transfers the state back in time , so that it is available in @xmath1 at the moment of entanglement generation @xcite .
although this seems to be a rather extreme interpretation of the formalism , it is based on the fundamental observation that the teleportation process relies on a time - symmetric combination of entangled state preparation and bell state measurement .
the conventional formulation of quantum mechanics tends to hide this symmetry by overemphasizing the role of the initial state , while neglecting the implications of measurement results for the situation before the measurement . to correct this inadvertent bias
, it may be useful to apply a time - symmetric description to the situation between state preparation and bell measurement .
such a time - symmetric description is provided by the post - selected measurement statistics obtained in weak measurement , which give equal importance to the initial and the final state of a quantum process @xcite .
it has already been shown that the statistics obtained by weak measurements can be used to identify the disturbance of a physical system in a general measurement by separating the actual changes of a physical property from the effects of new information about these properties @xcite . in the same spirit
, weak measurements could be applied to identify the physical properties of the systems involved in the teleportation before the bell measurement is performed . in the following
, we will analyze the complete statistics of the systems @xmath0 , @xmath2 , and @xmath1 between state preparation and bell measurement , taking into account the post - selection of a specific result of the bell measurement . in this analysis ,
the effects of the bell measurement correspond to a bayesian update of the initial state that selects the appropriate sub - ensemble from the total statistics without any physical changes to the properties of the systems .
the weak measurement analysis of quantum teleportation thus shows that the correlations between the physical properties of input @xmath0 and output @xmath1 that enable teleportation can already be found in the quantum fluctuations of the initial state before the bell measurement is actually performed .
the rest of the paper is organized as follows . in section [ sec : tele ] , a general formulation of the teleportation process is given and the observable effects of the bell measurement are considered . in section [ sec : weak ] , the statistics of weak measurements is introduced . in section [ sec : local ] , it is shown how the input state determines the statistics of local weak measurements on any one of the three systems in the interval between entanglement preparation and bell measurement . in section [ sec :
precorr ] , the correlations between input and output are analyzed .
the results show that all physical properties in @xmath0 and in @xmath1 are perfectly correlated , as indicated by a weak value of zero for projections onto mutually orthogonal states in @xmath0 and in @xmath1 . in section [ sec : partial ] , it is shown that partial teleportation by a quantum controlled bell measurement leaves the correlations between @xmath0 , @xmath2 , and @xmath1 visible in the final output , where the ideal correlations of teleportation are replaced by the correlations of optimal cloning . in section [ sec : discuss ] , the implications of the results for the teleportation process are discussed and a statistical interpretation is considered .
section [ sec : conclude ] summarizes the results and concludes the paper .
we consider the teleportation of the arbitrary quantum state @xmath4 of a @xmath3-level system , as illustrated in fig .
initially , system @xmath0 carries the input state , and systems @xmath2 and @xmath1 are in a maximally entangled state , @xmath5 the quantum information of the initial state @xmath4 can then be transfered to the remote system @xmath1 by a joint measurement of systems @xmath0 and @xmath2 .
this measurement is described by a projection onto an orthogonal basis of maximally entangled states , also known as a bell basis . since maximally entangled states can be transformed into each other by local unitary transformations , it is possible to express the @xmath6 states @xmath7 of the bell basis in terms of the initial entangled state @xmath8 by @xmath9 where the @xmath6 unitary operations @xmath10 are orthogonal in terms of their product traces , @xmath11 the bell measurement results in the projection of the initial state in @xmath0 , @xmath2 , and @xmath1 described by the partial inner product in @xmath12 , @xmath13 after the bell measurement , the state of system @xmath1 is given by a well - defined unitary transformation of the initial state @xmath4 .
the effect of this unitary can be compensated by applying the unitary @xmath10 that describes the transformation of the entangled state @xmath8 into the bell state @xmath7 .
that relates the output state in @xmath1 to the unknown input in @xmath0.,width=472 ] although the measurement formalism seems to describe an instantaneous change of the state , the bell measurement performed at @xmath0 does not change the total statistics observable in system @xmath1 . for random outcomes of the bell measurement ,
the average quantum state in @xmath1 is @xmath14 which is equal to the result obtained by tracing out @xmath0 and @xmath2 before the measurement .
it is therefore possible to describe the local state in @xmath1 before the bell measurement as a mixture of the states conditioned by the different outcomes of the bell measurement @xcite .
however , it is important to note that the decomposition in eq.([eq : decomp ] ) works for any state @xmath4 , so it is still necessary to explain how the bell measurement can fix the choice of a particular state @xmath4 . for this purpose ,
it is useful to consider the correlations between @xmath0 and @xmath1 in more detail .
a direct observation of correlations between the input @xmath0 and the output @xmath1 before the bell measurement is impossible because a measurement of @xmath0 will alter the state and therefore change the results of the bell measurement .
the teleported state would simply be the output state of the measurement performed in @xmath0 , and no insights into the actual teleportation process would be gained . however , it is possible to perform a weak measurement , where the measurement interaction is so small that the changes to the quantum state can be neglected .
the results of these measurements are equally determined by initial state @xmath15 and post - selected final measurement projection @xmath16 and can be expressed in terms of weak values @xcite .
for an observable described by the operator @xmath17 , the weak value is @xmath18 the right hand side of this equation shows the general expression of weak values using the density operator @xmath19 as initial state and the measurement operator @xmath20 for the final measurement . in this formulation , it is easy to see that the weak values can be interpreted as expectation values of a transient state @xcite .
for complex weak values , this transient state is given by the ordered product of density operator and measurement operator , normalized to a trace of one , @xmath21 note that the operator ordering for this non - hermitian operator product is based on the convention used for the mathematical definition of complex weak values in eq.([eq : wv ] ) . in principle , it is also possible to represent only the real part of the weak values by using the self - adjoint part of the transient state , as defined by the average of the above operator ordering with the opposite operator ordering @xcite . in the following ,
our main reason for using the operator representation for complex weak values is that its mathematical form is more simple and allows us to present the results in a more compact form .
however , the same results could also be obtained with the self - adjoint transient states originally introduced in @xcite . in the present context , it is sufficient that the transient states correctly describe the real parts of the weak values , and that the set of operators @xmath22 obtained for the different measurement outcomes @xmath23 can be recombined to obtain the original quantum state , @xmath24 where @xmath25 is the probability of @xmath23 given by the product trace of @xmath19 and @xmath20 .
the final measurement of @xmath23 thus decomposes the initial probability density @xmath19 into sub - ensembles @xmath22 that represent the weak value statistics for each measurement outcome @xmath23 .
we can now apply the transient state analysis of weak measurement statistics to the case of quantum teleportation .
the initial density matrix is given by the product of input state @xmath26 and entangled state @xmath27 , and the bell measurement is described by the product of a projection in @xmath12 and the identity operation in @xmath1 , @xmath28 the weak measurement statistics of all three systems , @xmath0 , @xmath2 , and @xmath1 , is then given by the transient state @xmath29 as illustrated by fig [ fig2 ] , the transient state @xmath30 describes the complete weak measurement statistics between the preparation of the entangled state and the bell measurement .
it is therefore possible to derive the outcomes of weak measurements for any observable @xmath17 by simply taking the product trace of @xmath17 and @xmath30 .
importantly , the weak measurement statistics of the three systems do not factorize , indicating that the transient state describes correlations between the three systems .
these correlations can be observed in weak measurements as the weak values of operators @xmath17 constructed from products of local projection operators on the systems @xmath0 , @xmath2 , and @xmath1 . in the following
, we will take a closer look at these correlations and discuss the physics of teleportation in the light of these results . .
the state @xmath30 describes correlations between all three systems involved in the teleportation process.,width=472 ]
the compact formulation of the transient state @xmath30 given in eq.([eq : tarb ] ) does not make use of the relation between the bell state @xmath7 and the initial entangled state defined in eq.([eq : bellstate ] ) . using this relation and the basis expansion of the initial entangled state in eq.([eq : e ] ) , the transient state can be expressed as @xmath31 note that the appearance of the unitary transform @xmath10 in the remote system @xmath1 is a consequence of the inner product of the projection on @xmath32 with the initial entangled state of @xmath2 and @xmath1 , as shown in eq.([eq : tele ] ) .
the transient operator thus explains the essential connection between the states in @xmath0 and @xmath1 in terms of a statistical correlation selected by the bell measurement .
these correlations can be observed in collective weak measurements of all three systems , where the weak measurements determine the weak values of an operator defined by products of local properties from all three systems .
simplifications are possible if the weak measurement only involves one of the three systems . in this case
, the local transient states can be obtained from the partial traces over those systems that are not involved in the weak measurement . for local measurements of the input system @xmath0 , the result reads @xmath33 thus the weak values of the input system @xmath0 are equal to the expectation values of the input state , confirming that the post - selection of a bell measurement outcome does not `` update '' the information about the input state in any way .
similarly , the state observed in local measurements of the output @xmath1 is given by @xmath34 this is actually the correct output state of the teleportation , since the post - selection did not involve any measurement of system @xmath1 , and weak values without post - selection of a final measurement result ( @xmath35 ) are necessarily equal to the expectation values observed in strong measurements .
significantly , local weak measurements on @xmath0 and on @xmath1 can be performed at the same time , before the post - selection of the bell measurement .
this means that the quantum state @xmath4 appears to co - exist in both @xmath0 and @xmath1 , suggesting a `` weak cloning '' effect as discussed in @xcite
. it may also be interesting to look at the local state observed in the reference system @xmath2 .
here , the trace over @xmath0 and @xmath1 has the curious effect of producing a complex conjugate of the initial state , given by @xmath36 since complex conjugation in the energy basis is often associated with time reversal , it is possible to interpret @xmath37 as a signal running backward in time , as discussed in @xcite . in the context of `` weak cloning
, '' however , the transient state of @xmath2 appears to describe the anti - clone usually associated with optimal cloning processes @xcite .
the local weak measurement results show that the quantum information encoded in the coherence of the input state appears in all three systems equally .
however , the transient state @xmath30 includes a more detailed description of the teleportation process in the form of correlations between the systems . to better understand the transfer of information between @xmath0 and @xmath1
, we should consider the transient state @xmath38 where only the reference system @xmath2 has been traced out .
clearly , the transient state @xmath39 is different from the product of the local transient states , indicating that the sum over @xmath40 describes correlations between @xmath0 and @xmath1 . to find out more about these correlations , we need to consider possible weak measurements of correlated observables .
if we consider a local property @xmath41 with eigenstates @xmath42 , the weak measurement statistics can be expressed in terms of the weak values of the projection operators @xmath43 . in strong measurements ,
these projection operators determine the probability of measuring the outcome @xmath44 .
the weak values of the projectors therefore correspond to the weak measurement equivalent of conditional probabilities @xcite . since the difference between the local transient states in @xmath0 and in @xmath1 is described by the unitary transform @xmath10 , it is convenient to consider a measurement of @xmath41 in @xmath0 and a measurement of @xmath45 in @xmath1 .
the weak values of the joint probabilities for the outcomes @xmath42 in @xmath0 and @xmath46 in @xmath1 is given by @xmath47 although weak values of projection operators are generally complex and can have real values below zero or above one , the results obtained for the projectors of @xmath41 in @xmath0 and @xmath45 in @xmath1 are all positive and correspond to a particularly simple classical probability distribution . specifically , the kronecker delta in eq.([eq : idcorr ] ) indicates that the results in @xmath0 and in @xmath1 must always be equal , since the result is zero unless @xmath48 .
the probability of the outcome @xmath48 corresponds to the probability of the outcome @xmath42 for the input state @xmath4 .
eq.([eq : idcorr ] ) therefore corresponds to the intuitive notion that system @xmath0 and system @xmath1 are identical copies of each other , where the state @xmath4 describes the probability distribution of the perfectly correlated physical properties @xmath41 .
importantly , the result of eq.([eq : idcorr ] ) applies to all physical properties of the teleported system .
thus , the teleportation process can be interpreted as a faithful copy of all physical properties @xmath41 , independent of input state .
note that this is very similar to the phase space picture obtained from the wigner function representation of continuous variable teleportation , where the teleportation process faithfully transfers position and momentum , and the wigner function merely describes a quantum statistical weight attached to each combination of position and momentum by the quantum state @xcite .
the role of the bell measurement is the identification of the unitary @xmath10 that describes the deterministic relation between the physical properties in both systems . by selecting a complete set of @xmath6 bell
states , it is guaranteed that such a deterministic relation can be found in each measurement , indicating a close analogy between @xmath3-dimensional hilbert spaces and discrete phase spaces described by a lattice of @xmath6 points @xcite . effectively , weak measurements in @xmath0 and weak measurements ( or strong measurements ) in @xmath1 can be treated as measurements of the same system .
therefore , the correlations observed between different measurements in @xmath0 and in @xmath1 correspond to correlations of these properties within a single physical system . if we consider the measurement of two non - commuting observables @xmath41 and @xmath49 with eigenstates @xmath42 and @xmath50 , the complex joint probabilities defined by the weak values of the projectors is given by @xmath51 & & \hspace*{7 cm } \langle p_j \mid q_i \rangle \langle q_i \mid \psi \rangle \langle \psi
\mid p_j \rangle.\end{aligned}\ ] ] this result is identical to the joint probabilities reconstructed from weak measurements of @xmath41 followed by a post - selection of @xmath49 @xcite . interestingly , the complex value of the weak measurement outcomes defines a specific measurement sequence .
however , in the case of quantum teleportation this sequence does not correspond to the temporal order of measurements : clearly , it does not matter whether the measurement in @xmath1 is performed before or after the measurement in @xmath0 .
instead , it is relevant that system @xmath0 is directly connected to the input , while @xmath1 is directly connected to the output .
the connection between the two systems is not established by interactions , but by obtaining the necessary information about the accidental relation @xmath10 between the two systems .
since this information relates the past of system @xmath1 to the future of system @xmath0 , it may seem as if the reference @xmath2 carries a signal backwards in time , as indicated by the transient state of the reference @xmath2 .
it is therefore possible to explain the appearance of `` closed time like curves '' discussed in @xcite in terms of the difference between the time at which information is obtained and the time to which this information applies .
according to the analysis presented above , quantum teleportation makes use of the accidental relation @xmath10 between the physical properties of @xmath0 and of @xmath1 to identify the transformation that makes the physical properties in the output of @xmath1 equal to the physical properties in the input of @xmath0 .
the bell measurement merely identifies the relation between @xmath0 and @xmath1 .
this relation is valid for the complete time interval between entanglement generation and bell measurement , so that system @xmath1 is a faithful copy of system @xmath0 during this time .
it has been noted that this situation corresponds to a weak measurement version of cloning ( `` weak cloning '' ) , where the usual limit of cloning fidelity does not apply and a cloning fidelity of one can be achieved @xcite .
however , the bell measurement completely randomizes the state of one of the clones , leaving behind only a single copy of the teleported state .
an alternative approach to accessing the correlations between @xmath0 and @xmath1 in quantum teleportation is to reduce the measurement back - action of the bell measurement to avoid the complete randomization of the input .
however , this requires a reduction of the measurement resolution .
thus , there is a trade - off between the teleportation fidelity and the output fidelity of the original system after the bell measurement , which ensures that the cloning fidelities remain within the established limits .
it has already been shown that the partial teleportation implemented by a finite resolution bell measurement with minimal back - action results in an optimal cloning scheme @xcite . here
, we generalize the discussion by considering coherent superpositions of the bell measurement and the identity operation .
we can then show that the transient state @xmath52 is part of the actual output state of a partial teleportation and contributes to the optimal cloning fidelity for the outputs of @xmath0 and @xmath1 .
a partial bell measurement can be defined as a quantum coherent superposition of a fully projective measurement of @xmath7 and a random assignment of the measurement outcome that leaves the state unchanged .
the measurement operator acting on the initial state can then be written as @xmath53 where @xmath54 and @xmath55 are the coherent amplitudes associated with measurement and identity operation , respectively .
since the positive operator valued measure given by @xmath56 should be normalized to give a total probability of one for all possible measurement outcomes , the coherent amplitudes @xmath54 and @xmath55 can be determined by a single parameter , e.g. @xmath57 here , the parameter @xmath58 determines the reduction of measurement strength , with @xmath59 indicating a fully projective measurement and @xmath60 indicating a random assignment of measurement outcome that does not change the quantum state .
the effects of the partial bell measurement on the input state @xmath61 are found by applying the operators @xmath62 to systems @xmath0 and @xmath2 of the initial state , @xmath63 thus , the output state is a quantum coherent superposition of the initial state and the final state of the ideal teleportation process .
it is possible to write the density matrix of this output state as an effective mixture of three components , corresponding to the density matrix of the initial state , the density of the final state , and the self - adjoint part of the transient state , @xmath64 the statistical weights of the three contributions are given by @xmath65 the quantum statistics of the output can therefore be explained in terms of the statistics of initial state , final state , and transient state .
in particular , the teleportation fidelity @xmath66 ( obtained after the application of @xmath10 ) is equal to one in the final state and in the transient state , while it is @xmath67 in the initial state .
likewise , the fidelity @xmath68 of the output state of system @xmath0 after the partial bell measurement is one for the initial state and for the transient state , but only @xmath67 for the final state .
the fidelities can therefore be given in terms of the statistical weights of the three contributions to the output state , @xmath69 as discussed in @xcite , partial teleportation is an optimal cloning procedure . in particular , the symmetric case of @xmath70
corresponds to optimal @xmath71 cloning .
[ fig3 ] illustrates the transition between optimal teleportation at @xmath59 and optimal cloning at @xmath70 for the case of a two level system ( @xmath72 ) . as the strength of the bell measurement
is reduced , the teleportation fidelity drops and the fidelity of the input system after the bell measurement increases . .
as bell measurement strength is reduced , the fidelity @xmath68 of the input system @xmath0 after the bell measurement increases and the teleportation fidelity @xmath66 decreases until the two values become equal at @xmath70 .
the average fidelity reaches its maximal possible value at this point , resulting in an implementation of optimal cloning as discussed in @xcite.,width=472 ] for the present discussion , the most important aspect of the partial teleportation is that it permits a direct measurement of correlations between @xmath0 and @xmath1 .
in fact , it has already been shown in previous work that optimal cloning of the type realized at @xmath70 includes an ideal cloning term that reproduces the joint probabilities of eq.([eq : cjoint ] ) as predicted from weak measurements @xcite . in the general case of partial teleportation ,
the joint probabilities obtained in measurements of @xmath41 in the output of @xmath0 and @xmath49 in the output of @xmath1 are given by @xmath73 & & + \mbox{re}\left(\langle p_j \mid q_i \rangle \langle q_i \mid \psi \rangle \langle \psi \mid p_j \rangle \right ) \rho_t.\end{aligned}\ ] ] thus , a partial bell measurement preserves a non - vanishing contribution from the real part of the weak measurement statistics that can be observed before the bell measurement is performed .
indeed , the complete weak measurement statistics can even be obtained in the limit of @xmath74 , where @xmath75 is negligibly small . in this case , the weak value given in eq.([eq : cjoint ] ) can be obtained from the small differences between the joint probabilities @xmath76 for different values of @xmath44 .
the logic of this measurement is surprisingly similar to that of the weak measurement , with the essential difference that the low resolution of the measurement originates from the nearly maximal disturbance of the state by the bell measurement , and not from the weakness of the measurement interaction .
the picture of quantum teleportation that has emerged from the weak measurement analysis and its application to partial teleportation indicates that the physical properties of the input system @xmath0 and the remote system @xmath1 are related by the unitary @xmath10 even before the bell measurement has taken place .
this is possible because the initial state already includes all possible relations @xmath10 as a coherent superposition . in the weak measurement analysis
, this superposition can be expressed as an effective mixture of the @xmath6 possibilities according to eq.([eq : complete ] ) .
the bell measurement merely selects the possibility valid in each specific teleportation , without changing the physical properties of the remote system at all .
weak measurements indicate that the physical properties of system @xmath1 originate from the source of the initial entanglement and experience no change or modification during the bell measurement .
since the physical properties of the input state do not change until they are modified by the measurement back - action of the bell measurement , the input properties of @xmath0 and the output properties of @xmath1 appear to coexist in the interval between entangled state preparation and bell measurement .
the relation @xmath10 that relates the physical properties of @xmath1 to the physical properties of @xmath0 is initially random .
in fact , the randomness of the local state of @xmath1 in the entangled state guarantees that each relation @xmath77 is equally likely , regardless of the input state in @xmath0 .
teleportation is not achieved by a physical interaction , but by obtaining information about this random relation between @xmath0 and @xmath1 in the bell measurement . specifically , the combination of the pre - determined correlations in @xmath8 with the correlations in @xmath7 discovered in the bell measurement results in the knowledge that @xmath10 describes the relation between the unmodified physical properties in @xmath0 and in @xmath1 .
the only reason why the perfect correlations between @xmath0 and @xmath1 can never be observed directly is the measurement back - action of the bell measurement , which completely randomizes the local properties of system @xmath0 .
however , a partial bell measurement can be used to avoid complete randomization , and the correlations between @xmath0 and @xmath1 observed after such a partial teleportation correspond to the ones obtained in the weak measurements performed before a full bell measurement .
our analysis thus illustrates the conceptual relation between the `` weak cloning '' observed in teleportation and the optimal cloning realized by partial teleportation @xcite . in both cases ,
the precise relation between the physical properties of the input @xmath0 and the output @xmath1 is determined by the correct result of the bell measurement . in the `` weak cloning ''
limit , the bell measurement is precise , but the state of @xmath0 is randomized after the measurement . in the partial teleportation limit , there is still some randomization in @xmath0 and the reduced measurement resolution results in teleportation errors in @xmath1 , but the error free contribution that remains is precisely described by the self - adjoint part of the transient state observed in weak measurements .
weak measurement statistics provide a powerful tool for the analysis of quantum processes . in the case of teleportation , it is possible to discover close analogies between classical statistics and quantum statistics that are difficult to recognize when the conventional measurement postulate is applied . in particular , weak measurement statistics can separate the information about the past obtained in the bell measurement from the back - action effects associated with the measurement .
it is then possible to show that the change of the quantum state in @xmath1 originates from the selection of a sub - ensemble that was already present before the measurement .
thus , quantum teleportation does not involve any `` spooky '' action at a distance .
the bell measurement at @xmath0 merely identifies the correct transform @xmath10 that relates the physical properties of @xmath1 to the physical properties @xmath0 even before the bell measurement was performed .
significantly , the relation @xmath10 can be applied to all physical properties , and the correlations described by the weak values of projection operators in @xmath0 and in @xmath1 suggest that all properties of @xmath0 are perfectly correlated with the corresponding properties in @xmath1 .
teleportation thus copies all physical properties of the quantum system @xmath0 equally , irrespective of whether the input state is an eigenstate of the property or not .
effectively , the relation defined by @xmath10 is more fundamental to teleportation than the specific values taken by the physical properties themselves . by using weak measurement statistics ,
it is therefore possible to verify that the quantum fluctuations of physical properties @xmath41 in @xmath0 and @xmath1 are perfectly correlated without having to assign counterfactual values to @xmath41 in either system .
teleportation can thus be understood as an operation on the physical properties of the systems , where the bell measurement obtains the relevant information about the relation @xmath10 that maps the properties of @xmath0 onto the properties of @xmath1 .
this work was supported by jsps kakenhi grant number 24540427 . | we explore the mechanism of quantum teleportation by analyzing the weak measurement statistics post - selected by the result of the bell measurement for the joint system composed of the input @xmath0 and the spatially separated output @xmath1 .
it is shown that the weak measurement statistics observed before the bell measurement includes correlations which relate every physical property in the input @xmath0 to a corresponding physical property in the output @xmath1 .
the bell measurement thus identifies the accidental relation between @xmath0 and @xmath1 already present in the quantum fluctuations of the input state .
significantly , this relation applies to all physical properties equally , and is completely independent of the input state .
teleportation therefore copies all physical properties of input system @xmath0 to output system @xmath1 , irrespective of whether the input state is an eigenstate of the property or not . |
in nature there are many examples of glassy systems , _ i.e. _ complex systems that exhibit a very slow dynamics that prevents them to reach the equilibrium state . in this class
there is a large variety of systems , like real glasses , spin glasses , supercooled liquids , polymers , granular material , colloids , ionic conductors , orientational glasses , vortex glasses @xcite .
a common feature of these systems is the _ frustration _ , that is a competition due to geometry or energy constrains
. experiments on glasses @xcite show that in a temperature driven transition precursor phenomena starts at a temperature well above the ideal glass transition .
this temperature depends on experimental conditions and is the onset of some dynamical anomalies , like non - exponential relaxation functions , anomalous diffusion , cooling rate dependent density .
analogous phenomena are present in spin glasses ( sg ) @xcite , that are a generalization of ising model where random distributed ferromagnetic and antiferromagnetic interactions give rise to frustration . in particular , experiments @xcite and numerical simulations @xcite have shown non - exponential autocorrelation functions below a temperature @xmath6 well above the transition temperature @xmath7 , while above @xmath6 only exponential relaxation functions are seen . to explain this phenomenon , randeria _ et al .
_ @xcite have suggested that in the sg the non - exponential regime starts at the griffiths temperature @xmath8 , _ i.e. _ the critical temperature of the ferromagnetic model , due to the presence of randomly large unfrustrated regions , allowed by the disorder in the sg interactions .
_ @xcite have proposed an alternative hypothesis in which the onset @xmath6 of non - exponential behavior coincides with the percolation temperature @xmath9 of fortuin - kasteleyn @xcite coniglio - klein @xcite ( fk - ck ) clusters ( defined below ) .
the idea of ref.@xcite is that in the sg the accessible phase space is simply connected above @xmath9 , while it is not below @xmath9 .
therefore for @xmath10 a local monte carlo ( mc ) dynamics performs a random walk on a ramified percolating - like structure with many time scales , giving rise to non - exponential relaxation functions .
since @xmath9 is less than , but close to , @xmath8 and @xmath6 is difficult to localize , it is not possible to exclude the @xmath11 hypothesis .
however numerical results @xcite on fully frustrated models , without disorder , where the randeria _ et al .
_ argument does not apply , are consistent with the @xmath12 scenario . in refs.@xcite
this result is given both for spin - flip and bond - flip dynamics .
the latter is strictly related to the bond _ frustrated percolation _
( fp ) where _ frustrated loops _ , _
i.e. _ closed paths of connected bonds covering an odd number of negative interactions , have zero weight @xcite . the fp model and the @xmath1 sg model
can be recovered as particular cases of a @xmath13-state potts spin glass ( psg ) @xcite respectively for @xmath14 and @xmath2 .
in this paper we will show numerical results on the static properties of the psg model in 2d . for any @xmath0
the model exhibits a sg transition at a temperature @xmath3 ( in 2d @xmath15 for any @xmath0 ) and a fk - ck percolation transition at higher temperature @xmath4 . for @xmath5
the higher transition corresponds to a real thermodynamical transition of a @xmath0-state potts model @xcite . to make the algorithm faster we use the swendsen and wang ( sw ) @xcite cluster mc dynamics that prevent the slowing down for temperatures near @xmath4 .
dynamical properties of the model for @xmath16 are given in ref.@xcite , where we have shown that autocorrelation times diverge at @xmath9 and , as in sg model , non - exponential relaxation functions are present below @xmath9 .
these results on sg model @xcite , fp model @xcite , fully frustrated systems @xcite and psg model @xcite suggest that the fk - ck percolation may play a role in the context of precursor phenomena , since below @xmath9 the frustration is present at all length scales by means of fk - ck percolating cluster , that can not include frustrated loops . in sec .
ii we define the hamiltonian and review some theoretical results . in sec .
iii we present the numerical results for the model in finite dimensions . in sec .
iv we define the sw dynamics and compere it with the spin flip dynamics , verifying the efficiency of the cluster dynamics . in sec . v we give the conclusions .
the potts spin glass ( psg ) model is defined by the hamiltonian @xmath17 \label{hamiltonian}\ ] ] where to each lattice site is associated an ising spin @xmath18 and a @xmath0-state potts spin @xmath19 .
the sum is extended over all nn sites , @xmath20 is a random quenched variable and @xmath21 is the strength of interaction .
the model is a superposition of a ferromagnetic @xmath0-state potts model @xcite and a @xmath1 ising sg @xcite and for @xmath22 ( _ i.e. _ @xmath2 ) recovers the @xmath1 ising sg hamiltonian . following ref.@xcite it is possible to define fk - ck clusters on this model , activating a bond , between nn sites with both sg interaction and potts interaction satisfied , with probability @xmath23 and defining a cluster as the maximal set of connected bonds . for a given set of interaction @xmath24 it is possible to shown @xcite that @xmath25 can be expressed in terms of bond configurations @xmath26 @xmath27 where @xmath28 if @xmath26 includes any frustrated loop , otherwise @xmath29 where @xmath30 is given in eq .
( [ p ] ) , @xmath31 is the number of clusters in the configuration @xmath26 , @xmath32 is the number of bonds and @xmath33 is total number of interactions .
let us observe that , while the hamiltonian ( [ hamiltonian ] ) is defined only for integer values of @xmath0 , the eq .
( [ partition_function ] ) is meaningful for every values of @xmath0 and that for @xmath14 the eq .
( [ partition_function ] ) gives the partition function of bond fp where a bond configuration without frustrated loops has a weight @xmath34 with @xmath35 , while a bond configuration with frustrated loops has a zero weight .
furthermore in the limit @xmath36 eq .
( [ partition_function ] ) gives the partition function of the _ tree percolation _
@xcite where any bond configuration with a loop is excluded .
exact renormalization group ( rg ) analysis on hierarchical lattice @xcite has predicted for the psg two critical temperatures .
the lower temperature @xmath3 corresponds to a sg transition in the universality class of @xmath1 ising sg , and the higher @xmath4 to a percolation transition in the universality class of a ferromagnetic @xmath0-state potts model .
same results are given for the fully frustrated version of the model studied with a mean field approach @xcite .
looking at the partition function ( [ partition_function ] ) one should expect a singularity at @xmath4 , for any @xmath37 , due to the singularity in the number of cluster @xmath31 .
nevertheless this singularity has never been observed in the case of sg ( @xmath2 ) .
in fact , the rg calculations in the case @xmath2 show a singularity at @xmath7 for the sg free energy and no singularity at @xmath38 .
this result is interpreted in ref.@xcite supposing that the free energy of the hamiltonian ( [ hamiltonian ] ) has the form @xmath39 where @xmath40 is an amplitude which vanishes for @xmath41 and @xmath42 is the specific heat exponent .
we have done our simulation using the swendsen and wang ( sw ) monte carlo cluster dynamics @xcite described in sec .
iv . as we will show ,
the sw dynamics is faster then standard local dynamics , like spin flip dynamics , but suffers of slowing down near the sg critical temperature .
nevertheless , since the sg transition in 2d occurs at @xmath15 @xcite and we are interested in study the system near the percolation transition at @xmath43 , the sw dynamics is particularly indicated .
we have performed numerical simulation of the psg model for @xmath44 on a 2d square lattice with linear sizes ranging from @xmath45 to @xmath46 lattice steps and with quenched random interaction configurations @xmath47 . defining as mc step an update of all the spins of the system
, we have discarded the data of the first 7 500 mc steps and have collected data over 15 000 , 25 000 or 50 000 mc steps , depending on the temperatures and sizes . for each @xmath0
we have calculated the binder parameter for the energy density @xmath48 @xcite defined as @xmath49 where the symbol @xmath50 stands for the thermal average .
this quantity allows to localize the transition and to distinguish between first order and second order phase transition .
in fact , for a second order phase transition in the limit @xmath51 it is @xmath52 for all temperatures , while for a first order phase transition it is @xmath53 where @xmath54 is the minimum value of @xmath55 ( occurring at the phase transition temperature ) and @xmath56 is the energy jump , related to the latent heat , at the same temperature . to estimate the thermodynamical critical exponents for the second order phase transition we have measured the potts order parameter @xmath57 ( where @xmath58 , @xmath59 is the density of potts spins in the @xmath60th state ) , the susceptibility @xmath61^{1/2 } \label{chi}\ ] ] ( where @xmath62 is the total number of potts spins ) and the specific heat @xmath63^{1/2}. \label{cv1}\ ] ] furthermore to estimate the percolation critical exponents we have calculated the percolation probability per spin @xmath64 where @xmath65 is the cluster size and @xmath66 is the density of clusters of size @xmath65 , the mean cluster size @xmath67 and the number of clusters @xmath68 in fig.[v4 ] we show the binder parameter for the case @xmath16 for system sizes @xmath69 .
it is possible to see that @xmath55 for small sizes has a minimum at @xmath70 and that for grater sizes it becomes constant for all temperatures , reveling a second order phase transition .
therefore we can make standard scaling analysis @xcite for the thermodynamical quantities . in particular , by definition of critical exponents
@xmath71 it is @xmath72 where @xmath73 is the correlation length and @xmath74 with @xmath75 finite size critical temperature of the psg model .
analogously from the definitions of the other critical exponents @xmath76 , @xmath77 and @xmath78 we get @xmath79 @xmath80 @xmath81 from standard scaling analysis applied to finite systems @xcite we expect for @xmath82 @xmath83 where @xmath84 is an universal function of the dimensionless variable @xmath85 .
analogous scaling functions are expected for the other thermodynamical quantities .
tuning the values of critical exponents and @xmath86 it is possible to verify the scaling hypothesis , as eq.([scaling_m ] ) , from the mc data .
the values for which the data collapse give the estimates of the critical exponents and of @xmath86 . in figs .
[ sca_s2 ] , [ sca_s2_1 ] , [ sca_s2_2 ] we show the data collapses for system sizes @xmath69 .
the estimated scaling parameters are @xcite given in tab .
[ thermo2 ] the estimated critical exponents are compatible , within the errors , with the expected values for a potts model with @xmath16 state , _
i.e. _ an ising model , in 2d : @xmath87 , @xmath88 , @xmath89 and @xmath90 @xcite .
the estimated critical temperature is @xmath91 .
for the fk - ck percolation quantities we have described the critical behavior of @xmath92 and @xmath93 introducing a percolation set of critical exponents ( @xmath94 , @xmath95 , @xmath96 and @xmath97 ) defined by the relations @xmath98 where @xmath99 is the connectedness length of the clusters and @xmath100 with @xmath101 finite size percolation temperature , @xmath102 @xmath103 standard scaling analysis for finite systems @xcite is applied also in this case and the results are summarized in figs .
[ sca_p ] , [ sca_s ] .
the estimated scaling parameters are given in tab .
[ perco2 ] .
all the estimated exponents are compatible , within the errors , to the corresponding thermodynamical parameters for the 2d ising model and the numerical estimate for the percolation temperature is @xmath104 consistent with the estimates of @xmath86 .
[ v14_100 ] we show the binder parameter @xmath55 for the psg model with @xmath105 and @xmath106 for system sizes @xmath107 .
the fact that @xmath55 has a non vanishing minimum for every size revels that there is a first order phase transition . in this case
there is no diverging length , therefore the scaling analysis can not be applied .
this kind of transition is characterized by the finite size relations @xcite @xmath108 ( where @xmath109 is the euclidean dimension ) for the maximum of finite size @xmath110 ( see tab .
[ maxch ] ) and the relation @xmath111 where @xmath112 is the temperature of the maximum of @xmath113 ( or of the mean cluster size @xmath93 ) for the size @xmath114 and @xmath115 is the corresponding value in the thermodynamical limit , _
i.e. _ the corresponding transition temperature @xmath116 ( or @xmath4 ) .
therefore @xmath116 and @xmath4 can be evaluated by linear fits with one free parameter .
the data are given in tab .
[ max ] and the results are for @xmath105 @xmath117 and for @xmath106 @xmath118 .
the results are summarized in the phase diagram in fig.[phase_d ] : for every @xmath0 the high temperature phase is disordered and non percolating ; decreasing the temperature there is a second or first order phase transition ( depending on @xmath0 ) at @xmath9 corresponding to the percolation of fk - ck clusters and to the ordering of potts variables ; at lower temperature there is the sg transition ( that in 2d occurs at @xmath119@xcite ) . for a fixed realization of @xmath47 it is possible to show @xcite that each critical point is characterized by a diverging critical length . at @xmath4
diverges the linear size of fk - ck clusters associated with the pair connectedness function , while at @xmath3 diverges the linear size of correlated regions .
it is interesting to note that the behavior of @xmath4 can be obtained from the exact expression of the transition temperature of a ferromagnetic @xmath13-state potts model @xcite only by renormalizing the number of states , _
i.e. _ @xmath120 with @xmath121 ( choosing @xmath122 ) .
the mc dynamics used to study the equilibrium properties of the psg model is the swendsen and wang ( sw ) cluster mc dynamics .
the sw dynamics is performed in two steps .
the first steps is to construct the fk - ck cluster configuration @xmath26 , given an ising and a potts spin configuration @xmath123 , activating bonds with the probability in eq .
( [ p ] ) between nn sites when both ising and potts spins satisfy the interaction and with zero probability otherwise .
the second step consists in reversing all the spins in a cluster at the same time with probability @xmath124 , for each cluster .
the sequence of the first and of the second steps applied to the whole system constitutes a mc step , that is the chosen unit of time .
this dynamic completely overcomes the problem of critical slowing down for the unfrustrated spin models @xcite , while turns out to suffer of diverging correlation time if applied to frustrated systems near a critical point .
this inefficiency is a consequence of the fact that the fk - ck clusters used in the sw dynamics do not represent anymore in frustrated models the regions of correlated spins near a critical point @xcite and their percolation temperature @xmath9 is greater than the critical temperature . in particular , this is true for the sg model for which an efficient cluster mc dynamics does not yet exist except for 2d @xcite , while efficient cluster dynamics have been proposed for systems with frustration but without disorder @xcite . nevertheless in sg for temperatures well above the critical temperature @xmath7 and near @xmath9
the sw dynamics is still efficient , consistently with the general observation that the cluster dynamics are efficient at least for temperatures above the percolation temperature @xcite .
on the other hand in ref.@xcite we have shown that the local spin - flip ( sf ) mc dynamics @xcite in the case of psg model with @xmath16 is characterized near @xmath9 by diverging correlation times .
to compare the efficiency of sw dynamics to that of sf dynamics we have studied for the case @xmath16 the correlation functions at the equilibrium , that for a generic observables @xmath125 is defined as @xmath126 , \label{corrm}\ ] ] where @xmath127 and @xmath128 is the equilibration time . as observables we have choose the potts order parameter @xmath82 and the energy @xmath48 of the whole system .
we have also studied the time dependent nonlinear susceptibility for a quenched interaction configuration @xmath129 ^ 2\right\rangle \label{chi_sg_t}\ ] ] ( where @xmath62 is the total number of spins ) .
the normalized correlation function is @xmath130 with @xmath131 . for both the sw and the sf dynamics we have measured the integral correlation time defined as @xmath132 where @xmath133 is the generic correlation function .
we have considered systems with lattice sizes @xmath134 at temperatures above and below @xmath9 .
the data for the sf dynamics are averaged over 32 different quenched interaction configurations , since the local updating of this dynamics strongly depends on the local fluctuation of the frustration . on the other hand the results on global sw dynamics
turns out to be robust " respect the interaction configuration average , in the sense that the fluctuations of @xmath135 are within the errors estimated on the basis of a single interaction configuration analysis .
the simulations have been done with an annealing method , _
i.e. _ with a slow cooling of the system at each temperature . for the sw cluster dynamics @xmath136 mc steps
turn out to be enough to equilibrate the system at the considered temperatures and the averages are done using the the data for the following @xmath137 mc steps . for the sf dynamics
we have discarded the first @xmath138 mc steps ( defined as the local update of any spin in the system ) and recorded the data for @xmath139 mc steps . in tab .
[ tau ] we show the results for @xmath140 .
analogous results have been found for smaller systems .
the data show that , while the sf correlation times for @xmath82 and @xmath48 grow abruptly near @xmath9 where both thermodynamical and percolation transition occur , the sw correlation times only show a slow trend to increase for decreasing temperatures , being smaller than the correspondent sf data at least of an order of magnitude . even for @xmath141 ,
that for sf starts to be non - zero below @xmath9 , the sw dynamics shows smaller correlation times . for temperatures well below @xmath9
it is possible to see that the sw dynamics is characterized by long autocorrelation times , as the sf dynamics .
therefore , at least for temperatures not too much below @xmath9 , the sw dynamics turns out to be more efficient than local sf dynamics .
in particular , near the thermodynamical transition at @xmath9 the sw dynamics completely overcomes the critical slowing down problem in the psg model , as for the unfrustrated models , even if frustration is present via the random interactions of the ising spins .
precursor phenomena characterize the paramagnetic phase of spin glass .
in particular , experiments and local monte carlo simulations show the presence of stretched exponential autocorrelation functions well above the sg transition temperature @xmath7 @xcite .
the relation of the onset @xmath6 of these precursor phenomena to any thermodynamical transition and the localization of @xmath6 are still matters of debate @xcite .
many numerical evidences on disordered @xcite and deterministic frustrated models @xcite have shown that @xmath6 is consistent with the percolation temperature @xmath9 of the the kasteleyn - fortuin and coniglio - klein clusters .
in particular , a generalization of the @xmath1 ising sg to a @xmath13-state potts sg , that recovers the sg for @xmath2 , has shown that for @xmath16 , as for @xmath2 @xcite , the @xmath12 hypothesis is numerically verified @xcite . in this paper , using very efficient cluster monte carlo dynamics , we have shown that in 2d for any @xmath5 the percolation transition corresponds to a real thermodynamical transition in the universality class of the @xmath0-state ferromagnetic potts model . in particular , we have considered the case @xmath16 and the cases @xmath105 , 50 where at @xmath4 a second order and , respectively , a first order phase transition occurs .
exact renormalization group calculations on hierarchical lattices @xcite and mean field analysis for a version of the model without disorder @xcite have shown the same scenario .
all these results suggest that the percolation transition may play a role in the precursor phenomena even in the sg case ( @xmath2 ) , where no thermodynamical transition occurs at @xmath9 .
this idea arises from the observation that , even in sg , below @xmath9 the frustration starts to be manifested on all the length scales by means of the fk - ck clusters , that can not include frustrated loops .
therefore the scenario presented is that for @xmath5 at the percolation temperature @xmath9 there is a thermodynamical transition with associated dynamical anomalies that `` vanishes '' for @xmath2 leaving the dynamical behavior unchanged .
ediger , c.a .
angell , and s.r .
chem _ * 100 * , 13 200 ( 1996 ) ; c.a .
angell , _ science _ * 267 * , 1924 ( 1995 ) and _ relaxation in complex systems _ , ed . k.l .
ngai , g.b .
wright ( office of naval research , washington , 1984 ) .
a. fierro , g. franzese , a. de candia , and a. coniglio , _ percolation transition and the onset of non exponential relaxation in fully frustrated models _ preprint cond - mat/9803202 submitted to _ phys . rev .
e_. .estimated critical exponents and critical temperatures @xmath86 for thermodynamical quantities @xmath82 , @xmath142 and @xmath113 for the psg model with @xmath16 . [ cols="<,^,^,^,^,^",options="header " , ] | we have studied the potts spin glass with 2-state ising spins and @xmath0-state potts variables using a cluster monte carlo dynamics .
the model recovers the @xmath1 ising spin glass ( sg ) for @xmath2 and exhibits for all @xmath0 a sg transition at @xmath3 and a percolation transition at higher temperature @xmath4 .
we have shown that for all values of @xmath5 at @xmath4 there is a thermodynamical transition in the universality class of a ferromagnetic @xmath0-state potts model .
the efficiency of the cluster dynamics is compared with that of standard spin flip dynamics . |
it is well - known that the perturbative vacuum of qed is trivial , i.e. it is characterized by the vanishing expectation value of the electromagnetic vector potential .
our goal is to investigate the vacuum non - perturbatively in the framework of the continuum theory in minkowski spacetime . as a working hypothesis
we assume the presence of a periodic electric background field .
the interaction of the fermions with the background field is treated exactly by integrating out the fermion fields , whereas the quantum fluctuations of the electromagnetic field are taken into account at one - loop order .
the non - perturbative treatment of the interaction with the background introduces infinitely many non - renormalizable , i.e. irrelevant interaction vertices .
it may happen that such vertices generate non - trivial , periodic ( otherwise anti - ferromagnetic ) vacuum structure .
similar examples are known for yang - mills theories on the lattice @xcite , where the existence of various anti - ferromagnetic vacua have been established due to irrelevant interaction terms .
our problem setting can also be motivated by the following intuitive picture . in the presence of a background scalar potential @xmath2 with overcritical amplitude @xmath3 ( with the rest mass of the electron @xmath4 ) localized electrons of the dirac sea
may tunnel into localized electron states of the same energy above the mass gap ( see fig .
[ tunnel ] ) .
therefore , the creation of a certain amount of electron - positron pairs with the spatial separation of @xmath5 is imaginable .
this might stabilize itself with a periodic charge density ( but zero net charge ) and a periodic mean electric field .
the expected effect might be destroyed due to the uncertainty principle .
localisation in an interval @xmath6 means a momentum and , consequently , an energy spread of order @xmath7 and the effect may occur only if @xmath8 .
it is not completely trivial to define an effective action depending on a background field @xmath9 .
the simple shift of the integration variable @xmath10 to @xmath11 - @xmath9 is not the answer requested , since a dependence on the background @xmath9 shall only occur due to terminating the loop expansion .
below we introduce an effective action exhibiting a genuine dependence on the background field .
our method of defining the sector of qed for field configurations belonging to quantum fluctuations around a given vacuum expectation value @xmath12 consists of the following steps : 1 .
_ projection to the sector with a given background field .
_ the generating functional of qed in minkowski spacetime has the form : @xmath13 + with the vector potential @xmath10 @xmath14 and the electron - positron field @xmath15 , where @xmath16 and @xmath17 are the action of the electromagnetic field and the dirac action , resp .
, in the covariant gauge with the gauge parameter @xmath18 .
let the vector @xmath19 be introduced in the space of the vector potential configurations .
multiply the integrand of the generating functional by the factor 1 written in the form @xmath20 with @xmath21 the spacetime volume . then
, we find @xmath22 where the functional @xmath23 can be concieved as the generating functional of the sector belonging to vector potential configurations in a hypersurface orthogonal to @xmath19 .
the projected generating functional @xmath24 is gauge invariant for the choice of @xmath19 satisfying the condition @xmath25 .
the projected effective action @xmath26 is defined by @xmath27 + for the choice @xmath28 with a given background field @xmath9 satisfying the lorentz condition @xmath29 , we obtain the effective action for the sector belonging to the given background @xmath9 . introducing the shifted integration variable @xmath30 , the projected generating functional
can be rewritten as @xmath31 2 .
_ identification of the background with the vacuum expectation value of the vector potential .
_ as to the next , it is required that @xmath32 this condition is used to determine the constant @xmath33 as the functional of the vacuum expectation value of the vector potential @xmath34 . for later use
it is useful to introduce the external sources @xmath35 , @xmath36 , and @xmath37 coupled to the quantum fluctuation @xmath38 of the vector potential , and to the fermion fields @xmath15 and @xmath39 , resp .
then , we find instead of eq .
( [ zqedp ] ) the expression @xmath40 with @xmath41 and @xmath42
the projected generating functional ( [ zqedp ] ) defines the projected effective action via eq .
( [ seffdf ] ) . we determine it in one - loop approximation ,
i.e. we replace the full vector potential @xmath10 in the dirac action @xmath43 by the background field @xmath44 .
thus , the generating functional is factored into an electromagnetic part and a fermionic part , @xmath45 the explicit forms of the actions are given as @xmath46 and @xmath47 with the inverse of the photon propagator in lorentz gauge , @xmath48 and the inverse of the fermion propagator in the background field @xmath9 , @xmath49 with the dirac matrices @xmath50 @xmath51 .
for the sake of simplicity , the notation @xmath52 is used , where @xmath53 and @xmath54 are either lorentz or spinor summation indices . in the one - loop approximation
the path integrals are gaussian ones and can be performed explicitly , leading to @xmath55 and @xmath56 with the modified photon propagator @xmath57 now we restrict our considerations to @xmath58 dimensional systems and time independent periodic backgrounds of the form @xmath59 satisfying the lorentz condition . the constant @xmath33 can be determined by using the fact that @xmath60 from this we find that @xmath61 one establishes now for the one - loop effective action : @xmath62 in the infinite volume limit one finds @xmath63 here the first and the second terms represent the energy density of the dirac vacuum and that of the free electromagnetic field , resp .
( see e.g. @xcite ) .
@xmath64 denotes the summation over all non - negative single fermion energies @xmath65 , being the energy eigenvalues of the dirac equation in the external field @xmath9 . the third term on the r.h.s .
( [ actden ] ) is just the negative of the energy density of the periodic electric background field .
the last term of the effective action ( [ effac ] ) gives a vanishing contribution to the action density in the infinite volume limit .
the meaning of the first two terms of eq .
( [ actden ] ) might lead one to the false conclusion that the negative of the action density is equal to the energy density of the vacuum , as it would happen if the background field were constant @xcite . determining the energy density from the energy - momentum tensor we will show that the energy density does not equal the negative of the action density for inhomogeneous background field . as 1 + 1
dimensional qed is a superrenormalizable theory , the action density is uv finite after subtracting the action density of the free vacuum @xmath66 , i.e. that of the vacuum in the absence of the background field . this difference @xmath67 has been calculated numerically by solving the dirac equation for the single - fermion energies .
these energy eigenvalues show the same band structure plotted against the momentum in fig .
[ felhasad ] as the electrons in the kronig - penney model @xcite because of the periodic potential .
[ effacden ] shows how @xmath68 changes by modifying the amplitude @xmath53 and the wavelength @xmath69 of the potential .
( the fermion rest mass is chosen for @xmath70 . )
one can recapitulate from fig .
[ effacden ] that the surface @xmath71 has only a single stationary point at the origin of the parameter space @xmath72 , i.e. the path integral defining the vacuum - vacuum transition amplitude is dominated by the trivial , identically vanishing field configuration @xmath73 .
the vacuum expectation value of the current can be written at one - loop order in the following form : @xmath74 the first term on the r.h.s . of eq .
( [ aram ] ) is just the first functional derivative of the effective action @xmath75 . in the infinite volume limit
the second term gives : @xmath76 with the field strength tensor @xmath77 evaluated from the background @xmath78 .
here we used that @xmath79 due to eq .
( [ cnull ] ) .
thus , one obtains the equation @xmath80 for the background field configuration making the effective action extremum , one recovers the vacuum expectation value of the maxwell equation .
thus , the effective action has an extremum at the background field configuration coinciding with the mean field solution . in our numerical search , this is the trivial extremum found at @xmath81 .
satisfying the necessary condition of the extremum of the effective action , eq .
( [ mfeqact ] ) results in the poisson s equation for the mean field @xmath82 , which must be considered together with coupled set of operator equations for the quantum fields @xmath83 , @xmath84 and @xmath85 . the latter equations should be solved and the result substituted in the l.h.s .
( [ mfeqact ] ) , in order to make the charge density explicit .
the symmetric energy momentum tensor is defined by @xcite : @xmath86 with the lagrange density @xmath87 corresponding to the action @xmath88 and with the covariant derivative @xmath89 . substituting the ansatz ( [ ansatz ] ) one obtains for the component @xmath1 of the energy - momentum tensor : @xmath90 where @xmath91 @xmath92 are the terms independent of the fermion field and being of the order @xmath93 . due to the constraint ( [ constraint ] ) , the expectation value of the first order term vanishes ,
therefore it can be neglected .
the second order term @xmath94 is the expression for the free electromagnetic field , whereas the zeroth order term is given by @xmath95 and represents the energy density of the periodic background field . due to the projection to a particular sector of the theory , the additional term @xmath96 & = & ( c_0 - { \bar a}^\mu ( x){\bar a_\mu } ( x ) ) \lambda \nonumber\\ & = & \biggl ( ( \omega^{-1 } \int du { \bar a}^\mu ( u){\bar a_\mu } ( u ) - { \bar a}^\mu ( x){\bar a_\mu } ( x ) \biggr ) \lambda \nonumber\\ & = & \frac{1}{2 } a^2 ( 1 - 2 \cos^2 ( \ell x_1 ) ) \lambda\end{aligned}\ ] ] occurs .
it is easy to see that this term does not contribute to the energy , but gives a non - vanishing periodic contribution to the energy density .
finally , the term @xmath97 represents the contribution of the fermions to the energy density . the first term on the r.h.s .
gives vanishing contribution to the total energy if charge conservation is required , whereas the second term accounts for the kinetic energy of the fermion system . at one - loop order
, we have to substitute @xmath98 in the ` hamilton operator ' , i.e. write @xmath99 since @xmath100 in our case , we obtain @xmath101 , i.e. the free ` hamilton operator ' .
the vacuum expectation value of @xmath1 in the presence of the background field @xmath102 is defined as @xmath103 similarly , the vacuum expectation value of @xmath104 in the absence of the background field @xmath105 is given by substituting @xmath106 in eq.([37 ] ) and replacing @xmath107 by @xmath108 .
then , the casimir energy of the vacuum due to the background field @xmath109 is @xmath110\end{aligned}\ ] ] this expression of the casimir - energy reminds one on the expression for the energy of a system of electric charges in classical electrodynamics .
there , the energy is the sum of the energy of the electromagnetic field and the kinetic energy of the charges @xcite . in our case
the casimir energy is the sum of the background electric field and the change of the relativistic kinetic energy ( including the rest mass ) of the dirac - sea due to the presence of the background field .
eq.([37 ] ) can be rewritten by the help of the generating functional @xmath111 as @xmath112 where @xmath113 denotes the operator obtained from @xmath1 by replacing the fields @xmath38 , @xmath15 , and @xmath39 by the operators @xmath114 , @xmath115 , and @xmath116 and the variable @xmath117 by @xmath118 .
then , we find @xmath119 furthermore , the expectation value @xmath120 is equal to the energy density of the free electromagnetic field . indeed , it holds for the second derivatives @xmath121 where @xmath122 tends to the free propagator @xmath123 in the infinite volume limit , since the last term on the r.h.s .
( [ modpro ] ) is of the order @xmath124 .
consequently , the pure electromagnetic contribution to the casimir energy density is given by @xmath125 it is more cumbersome to evaluate the fermionic contribution @xmath126 to the casimir energy density , where the energy density of the free dirac vacuum is subtracted .
since qed in dimensions @xmath58 is superrenormalizable , the difference turns out to be uv finite without further renormalizations .
we perform the evaluation of the fermionic part of the casimir energy in the second quantized formalism
. then we can write @xmath127 where @xmath128 is the ` interacting ' vacuum ( in the presence of the periodic background field ) and @xmath129 denotes normal ordering with respect to the normal vacuum ( in the absence of the background field ) .
the evaluation is performed in the following steps : 1 .
the fermion field @xmath130 is expanded in terms of the eigenspinors @xmath131 and @xmath132 of the dirac hamiltonian @xmath133 belonging to the energy eigenvalues @xmath134 and @xmath135 , resp . here
the quasi - momentum @xmath136 and the integer @xmath137 enumerating the bands are introduced .
2 . then the creation and annihilation operators @xmath138
, @xmath139 and @xmath140 , @xmath141 of these stationary single particle states are expressed as linear combinations of the creation and annihilation operators @xmath142 , @xmath143 and @xmath144 , @xmath145 of the free fermion states @xmath146 and @xmath147 of energies @xmath148 and @xmath149 , resp . in terms of the latter , the normal ordering is performed .
3 . finally ,
the normal ordered operator is reexpressed in terms of the operators @xmath138 , @xmath150 and @xmath140 , @xmath141 and the vacuum expectation value with respect to the vacuum in the presence of the background field is taken .
thus , one arrives to the following expression @xmath151 where , with @xmath152 , @xmath153 , @xmath154 and @xmath155 with the constant coefficients @xmath156 the latter are the overlap integrals of the eigenspinors in the presence of the background and those in the absence of the background .
furthermore , the time dependent creation - annihilation operators are introduced : @xmath157 the constant term @xmath158 represents the casimir energy density of the fermion vacuum .
the last two terms of eq .
( [ ef ] ) cancel . performing the calculation results in : @xmath159 rr^\prime\in n } % \end{sb } \biggl[\epsilon^0_{-kr } ( { \cal f}_1^{(krr^\prime)}+{\cal f}_2^{(krr^\prime ) } )
\cos((r - r^\prime)\ell x ) \nonumber\\ & & + ( \epsilon^0_{k - r^\prime}-\epsilon^0_{kr } ) { \cal f}_3^{(krr^\prime ) } \cos((r+r^\prime)\ell x ) \biggr],\end{aligned}\ ] ] where we introduced the following notations : @xmath160 @xmath161 and @xmath162 denotes the eigenspinors of the free dirac - equation for the positive and negative energy eigenvalues , respectively .
it is straightforward to establish from eq .
( [ enden ] ) that the volume integral of the energy density is not negative : @xmath163 ) that this expression of the energy only vanishes in case of @xmath164 .
this means that the casimir energy @xmath165 of the vacuum is always positive if a non - vanishing periodic electric background field is assumed .
thus , the vacuum of @xmath166 does not favour a periodic mean field energetically with respect to the normal vacuum .
the vacuum expectation value @xmath167 is defined by the minimum of the energy functional @xmath168 the second term vanishes due to the explicit value of @xmath169 .
thus , the necessary condition of the energy minimum takes the form @xmath170 the functional derivative of the first term gives @xmath171 at one - loop order the fermionic contribution to the energy can be rewritten as @xmath172 taking its functional derivative we find @xmath173 \left ( { \bar \eta } \frac { \delta g } { \delta { \bar a}^0 ( x ) } \eta \right ) e^ { { \rm i } ( { \bar \eta } g \eta ) } \biggl|_0 \nonumber\\ & = & - { \rm i } { \mbox { tr } } \left ( \gamma^0 g(x , x ) \right ) + c^0 ( x ) \end{aligned}\ ] ] with @xmath174 .\end{aligned}\ ] ] thus , we find the following equation for the field @xmath167 minimizing the energy : @xmath175 the first term on r.h.s . is the expectation value of the charge density @xmath176 , determined via the propagator @xmath177 as a given functional of @xmath178 , and a similar statement holds for @xmath179 .
a similar equation appears in the classical case when a certain charge distribution moves in an electromagnetic field . integrating out the effect of the charged particle distribution it will result in a polarisation charge density term beside the common charge density in the equation of motion for the scalar potential . in eq .
( [ mfeq ] ) the term @xmath179 corresponds to a polarisation charge density , so minimizing the energy with respect to @xmath180 leads to an equation similar to that of poisson s equation in a polarised medium .
( [ mfeq ] ) can be solved directly for @xmath178 in principle without the need for solving any other equations .
on the contrary to this the mean field equation ( [ mfeqact ] ) obtained by extremizing the effective action does not take the polarisation of the vacuum into account , in order to do this we have to solve a system of operator equations as well .
the energy density of @xmath166 in the presence of a periodic mean field is determined from the energy - momentum tensor . for this purpose a projection method is worked out which is applied to treat the mean electromagnetic field self - consistently .
the projected effective action and the energy density of the vacuum are derived at one - loop order , whereas the interaction of the electron - positron field with the periodic mean field is treated exactly .
it is established , that the negative of the effective action must not be regarded as the energy of the system if the background field is not constant .
it was shown that the necessary conditions of the extremum of the effective action and the minimum of the energy lead to different equations for the vacuum expectation value of @xmath82 .
( [ mfeqact ] ) obtained by extremizing the effective action is the vacuum expectation value of the poisson s equation that does not include the polarisation of the vacuum due to one - loop radiation corrections .
those are accounted for by separate operator equations for the quantum fluctuations of the electromagnetic field and for the dirac field . on the other hand eq .
( [ mfeq ] ) obtained by minimizing the energy functional includes the polarisation effects of the vacuum .
the expectation value of the component @xmath1 of the energy - momentum tensor was determined as the function of the amplitude @xmath53 and wave number @xmath181 of the static periodic scalar potential @xmath182 .
it is found that the vacuum configuration with this periodic electric mean field is not favoured energetically compared to the normal vacuum .
the volume integral of the energy density plotted against the amplitude @xmath53 in fig . [ thens ] shows that the energy of the system increases with increasing @xmath53 monotonically .
the result obtained contradicts to our naive expectation discussed in the introduction .
possibly , the reason is that the naive picture mentioned there does not take into account the uncertainty principle i. e. for a certain @xmath69 the energy spread of a wave packet localized in an interval of @xmath183 could be much larger than the amplitude of the potential and of course in this case our naive picture is not valid any more .
we set now forth our work looking for periodic ground states at finite chemical potential .
to get the projected effective action at one - loop order in eq .
( [ effac ] ) we must find the fermionic single - particle energies , so we have to solve the dirac equation in the presence of the potential ( [ ansatz ] ) : @xmath184 we look for the solution of eq .
( [ diracegy ] ) in the form of bloch waves corresponding to the energy eigenvalues @xmath185 : @xmath186 inserting ( [ megoldas ] ) into ( [ diracegy ] ) we find @xmath187 we get non - trivial solutions when the determinant of the matrix appearing next to the dirac - spinors equals zero .
if we had solved the schrdinger equation in the presence of this sinusoidal potential , the form of the matrix would have been tridiagonal which means that the diagonal and the neighouring diagonal elements are nonzero . in relativistic case the matrix elements are replaced by @xmath188 dimensional @xmath189 matrices but the structure of the matrix remains unchanged .
we can not get the eigenvalues analitically because of the complicated structure of the matrix but we can determine them as precisely as we wish . in numerical calculations we work with matrices with finite dimension .
using the well - known identity @xcite @xmath190={\rm det}\pmatrix { p & q \cr r & s \cr } = { \rm det}[p ] \ { \rm det}[s - rp^{-1}q]\label{lemma}\ ] ] we can reduce the problem of calculating the determinants of these @xmath191 dimensional matrices to calculating four dimensional matrices by identifying the upper left matrix element of the matrix under investigation with @xmath192 in eq .
( [ lemma ] ) .
( @xmath193 denotes the number of @xmath194-s taken into account in eq .
( [ matrix ] ) . )
using identity ( [ lemma ] ) @xmath195 times we get a product of @xmath195 determinants of four dimensional matrices .
we computed the determinant of eq.([matrix ] ) as the function of @xmath185 and determined its zeros , corresponding to the energy eigenvalues @xmath196 in order to evaluate @xmath68 we have to sum the eigenvalues @xmath197 and extract from it the sum of the negative eigenvalues of the free dirac equation .
this difference will depend on the accuracy of the eigenvalues , the number of the eigenvalues taken into account in the sum , the size of the chosen matrix and , of course , the parameters of the potential , the amplitude @xmath53 and the wave number @xmath69 .
we have to be convinced of the stability of the numerical calculation .
we increased the number of members in the sum and the numerical accuracy of the determination of eigenspinors by choosing larger matrices as far as we have seen that the energy difference does not change significantly . to get the expectation value of the component @xmath1 of the energy - momentum tensor we also have to determine the eigenspinors of the dirac equation ( [ diracegy ] ) .
we calculated them with the help of the eigenvalues by solving a system of homogeneous linear equations ( [ matrix ] ) for the eigenspinors .
the authors would like to thank j. polnyi for consulting this work and g. plunien for the valuable discussions .
s.n . thanks g. soff for his kind hospitality .
k.s . expresses his gratitude for the follow - up grant of the alexander von humboldt foundation and w. greiner for his kind hospitality .
this work was supported by the projects otka t023844/97 , daad - mb 27/1999 and nato grant pst.clg.975722 .
99 -1 mm j. fingberg , j. polonyi , nucl . phys .
* b486 * ( 1997 ) 315 r. rajaraman , _ solitons and instantons _ ( north - holland , amsterdam , 1989 ) l. landau , m. lifshitz , _ course of theoretical phys .
relativistic quantum theory _
( pergamon press , oxford , 1971 ) j. d. jackson , _ classical electrodynamics _
( wiley , new york , 1975 ) w. h. press , b. p. flannery , s. a. teukolsky , w. t. wetterling , _ numerical recipes . the art of scientific computing _
( cambridge university press , cambridge , 1986 ) | it is shown that the vacuum of qed@xmath0 in minkowski spacetime does not favour a periodic electric mean field .
the projected effective action exhibiting a genuine dependence on the non - vanishing background field has been introduced .
the functional dependence of the energy density of the vacuum on the assumed periodic vacuum expectation value of the vector potential is determined from the component @xmath1 of the energy - momentum tensor at one - loop order . treating the background field non - perturbatively ,
the energy of the vacuum in the presence of a periodic mean field is found not be equal to the negative of the effective action .
+ + _ pacs : _ 12.20.ds s. nagy and k. sailer _ department for thoretical physics , kossuth lajos university , debrecen , hungary _ |
understanding the origins of the chemical elements is a fundamental goal of contemporary astrophysics . as this understanding has advanced in recent years ,
a few elements have been rather neglected .
boron is one such element .
a primary reason for its neglect has been that stellar absorption lines of the boron atom and ions are not detectable in spectra obtainable from the ground ; the low abundance of boron means that this element is detectable only through its strong resonance lines which are in the ultraviolet . the advance in ultraviolet stellar spectroscopy made possible by the _ hubble space telescope _ ( hst ) and its goddard high resolution spectrograph ( ghrs ) has led to new data on the boron abundance of cool and hot stars . pioneering work on stellar boron abundances was presented by boesgaard & heacox ( 1978 ) who analyzed the @xmath0 1362 resonance line in spectra of early - type stars observed with the _ copernicus _ satellite . to map the evolution of boron with metallicity it is necessary to observe unevolved metal - poor stars .
since these are cool , boron detection has to be made using the atomic resonance lines .
no advance in our knowledge of the boron abundance came with the _ international ultraviolet explorer _ ( iue ) satellite ( an attempt to detect the resonance lines near @xmath0 2497 in a metal - poor star was made by molaro 1987 ) .
duncan , lambert , & lemke s ( 1992 ) detection and analysis of the @xmath0 2497 lines in three metal - poor stars observed with the ghrs has been followed by other studies of these lines see edvardsson et al .
( 1994 ) , rebull et al .
( 1996 ) , and duncan et al .
( 1997a , 1998 ) .
kiselman ( 1994 ) and kiselman & carlsson ( 1996 ) showed that non - local thermodynamic equilibrium ( nlte ) effects have to be considered in analyses of the 2497 lines .
standard big bang nucleosynthesis calculations do not predict a significant production of @xmath13li , @xmath14be , @xmath15b , and @xmath16b , which are thought to be produced by spallation reactions between cosmic rays and cno nuclei ( see ramaty , kozlovsky , & lingenfelter 1996 and ramaty et al .
1997 for recent discussions on the theoretical yields ) .
models of inhomogeneous big bang nucleosynthesis suggest that observable amounts of be and b could be produced ( e.g. malaney & fowler 1989 ; kajino & boyd 1990 ; jedamzik et al .
1994 ) , but this possibility has been questioned on the theoretical grounds ( e.g. terasawa & sato 1990 ; reeves et al .
1990 ) . in this paper
, we derive the boron abundances of two metal - poor stars observed by us to fill in specific places of the study of the evolution of boron .
bd @xmath173130 is a cool very metal - poor star ( [ fe / h]@xmath5 ) selected to extend b abundance determinations to a star of a factor of two more deficient than hd 140283 ( [ fe / h]@xmath18 ; previously studied by duncan et al .
1992 and edvardsson et al .
1994 ) , and is probably the most metal - poor slightly evolved star for which boron can be measured with hst .
duncan et al . (
1997a ) has also observed two stars ( bd @xmath193442 and bd @xmath13740 ) with metallicities [ fe / h]@xmath20 ; however , they are much warmer and , hence , boron is mainly ionized making more difficult to obtain a precise measurement of its abundance from the lines . our second star , hd 84937 , is of special interest because @xmath13li ( another expected product of spallation and fusion reactions driven by cosmic rays ) has been traced in its atmosphere ( smith , lambert , & nissen 1993 ; hobbs & thorburn 1994 , 1997 )
. both stars also have previous measurements of @xmath21li in the literature , and have been included in the samples for deriving beryllium and oxygen ( from uv oh lines ) abundances in metal - poor stars by garca lpez et al .
( 1998 ) and israelian , garca lpez , & rebolo ( 1998 ) , respectively , using the same stellar parameters . to better study the dependence of the derived abundances on the tools usually employed in the analyses , we present here the first detailed comparison of boron abundances derived using the two sets of model photospheres widely employed in the analyses of stellar spectra : those calculated with the atlas program , constructed by r. l. kurucz , and those from the marcs program , developed at uppsala astronomical observatory
. in both cases there have been considerable recent developments in these codes and their underlying input data .
this careful analysis of boron abundances in very metal - poor stars , in which we also re - assess published abundances of old disk and halo unevolved stars , is aimed at better constraining observationally the nucleosynthetic origins of boron .
observations of the stars bd @xmath173130 and hd 84937 were carried out using the ghrs in the low resolution g270m / ssa mode ( @xmath22 ) . the g270 m grating was centered at @xmath0 2497 to observe the doublet located at @xmath23 2496.772 and 2497.723 ( air values ) .
the standard fp - split mode , which provides spectra in slightly different carrousel positions , was used to obtain high signal - to - noise ratios ( s / n ) and minimize the effects of the photocathode glitches and the geomagnetically - induced image motion problem ( gimp ) .
exposure times were estimated from the uv fluxes measured by iue at @xmath0 2500 for our targets .
bd @xmath173130 is a cool metal - poor giant with @xmath24 and was observed with hst during ten orbits ( 21434 sec exposure time ) on 1995 august 4 , while hd 84937 is a hotter metal - poor dwarf with @xmath25 which required three orbits ( 5440 sec ) on 1996 march 21 .
data reductions were performed by standard procedures using the `` stsdas '' package of the iraf suite of programs . final s / n values of 100 and 125 per diode ( 50 and 63 per pixel ) were achieved for bd @xmath173130 and hd 84937 , respectively .
their corresponding measured continuum fluxes at @xmath0 2500 were @xmath26 and @xmath27 erg @xmath28 s@xmath29 @xmath29 .
these values are about 3040% larger than those measured by iue at that wavelength , and these differences are reduced to 2025% when the hst data are degraded to match the iue spectral resolution .
differences in the flux calibrations of both instruments should explain these discrepancies .
the wavelength calibration provided a dispersion of 0.024 pixel@xmath29 , and the final spectra were shifted in wavelength to correct for stellar and earth motions using the @xmath0 2496.533 line as the reference .
continuum normalization is a difficult task in a crowded spectral region like this one . in principle , the hst spectra have been corrected for the spectrograph efficiency versus wavelength .
a soft curve was fitted to the upper points of the spectra , assigning a value of 1 to those pixels corresponding to the region just shortwards of @xmath0 2500 , considered by duncan et al .
( 1992 ; 1997b ) and edvardsson et al .
( 1994 ) as the most line - free region in the data .
spectral syntheses performed afterwards showed that this criterion agrees well with the predictions of synthetic spectra .
figure 1 shows a portion of the final spectra for both stars .
effective temperatures ( @xmath30 ) were estimated initially using the alonso , arribas , & martnez - roger ( 1996a ) calibrations versus @xmath31 and @xmath32 colors , which were derived applying the infrared flux method ( blackwell et al .
1990 ) , and cover a wide range in spectral types and metal content . both stars , bd @xmath173130 and hd 84937 , belong to a large sample of metal - poor stars for which garca lpez et al .
( 1998 ) have derived beryllium abundances .
all the details on photometric measurements , as well as on the selection of metallicity values for these two stars are provided in that work .
surface gravities ( @xmath33 ) were determined by comparing the observed strmgren @xmath32 and @xmath34 indices with synthetic ones generated using the corresponding filter transmissions and a grid of kurucz ( 1992 ) blanketed model atmospheres fluxes .
this is also explained in detail by garca lpez et al .
( 1998 ) , where a comparison is given with the spectroscopic estimates by other authors .
in particular , nissen , h@xmath35 g , & schuster ( 1997 ) have recently derived surface gravities for a sample of 54 metal - poor stars using _ hipparcos _ parallaxes to determine luminosities . for ten stars in common , the gravities derived by garca lpez et al .
are slightly lower systematically , with a mean difference of @xmath36 dex in @xmath33 .
as we will see below , a difference of this magnitude does not affect the boron abundances derived from the @xmath0 2497 lines in metal - poor stars .
bd @xmath173130 had been generally considered in the past as a subgiant star with @xmath37 , but the comparison of its strmgren indices with our synthetic @xmath38 photometry indicates that it is a giant star with @xmath39 .
garca lpez et al . (
1998 ) computed synthetic spectra in the region around the doublet at @xmath0 3131 ( used to derive beryllium abundances ) for different combinations of the stellar parameters , within the error bars provided by the photometric calibrations , and table 1 lists the values which best reproduce the observations .
these values also reproduce well the region observed in bd @xmath173130 and hd 84937 , with the exception of the derived metallicity which is slightly higher than that found from optical and near - uv analyses , as explained in the next section . apart from our observations , we have also derived boron abundances for procyon and for the other metal - poor stars with available measurements of the @xmath0 2497 doublet . for procyon
we choose @xmath40 k , @xmath41 and [ fe / h]@xmath42 , to be consistent with the analysis of duncan et al .
( 1997b ; see below ) . for consistency
, we also estimated the stellar parameters for the other stars by the methods applied to our stars .
most of them were studied by garca lpez et al .
all metal - poor stars considered for boron abundances analysis in this work are listed in table 1 .
due to the general lack of accurate oscillator strengths for the lines in the region , we used a line list computed by comparing lte synthetic spectroscopy with observed spectra of some stars .
the initial line list was taken from duncan et al .
( 1997b ) who analyzed dwarfs and giants of approximately solar metallicity .
the list was drawn from the kurucz ( 1993 ) lowlines list with wavelengths measured in the laboratory but the @xmath43 values were adjusted to reproduce the spectra of procyon , the sun , @xmath12 cen a & b , and the metal - poor star hd 140283 .
several lines predicted by kurucz as well as other lines of the same element studied by johansson & cowley ( 1988 ) and johansson ( 1992 ) were added .
duncan et al.s ( 1997b ) line list is very similar to that compiled by duncan et al .
( 1992 ) .
synthetic spectra were obtained by us using the code wita2 , a unix - based version of abel6 code ( pavlenko 1991 ) , which computes lte atomic and molecular synthetic profiles for a given model atmosphere .
we employed initially a grid of models provided by kurucz ( 1992 ) , which are interpolated for given values of @xmath30 , @xmath33 , and [ fe / h ] . to test the behavior of the duncan et al .
( 1997b ) line list using wita2 and a high - resolution hst spectrum , we compared a @xmath44 spectrum of procyon ( observed by lemke , lambert , & edvardsson 1993 ) with synthetic spectra obtained using the stellar parameters cited above , a microturbulent velocity @xmath45 km s@xmath29 , a wavelength step of 0.009 , and convolved with a gaussian with a fwhm of 0.027 to reproduce the observed spectral resolution .
the stronger line at @xmath0 2497.723 is seriously blended with other lines , and the abundance determination relies only on the line at @xmath0 2496.772 , more isolated and weaker , which is blended with lines of and at @xmath23 2496.708 and 2496.870 , respectively .
obrian & lawler ( 1992 ) and johansson et al . ( 1993 ) provide accurate values ( which differ by only 0.03 dex ) for the oscillator strength of this line . to improve the fit of the synthetic to the observed spectrum in the region surrounding the weaker line , we varied the @xmath43s of 13 of the 20 lines present in the 2496.42497.4 region until a good fit was obtained .
in particular , we adopted the @xmath43 value provided by obrian et al .
( 1991 ) for the @xmath0 2496.533 line ( with an uncertainty of 9% ) , the line @xmath0 2496.933 was not taken into account , and we included two weak lines at @xmath23 2497.153 and 2497.420 from the list used by duncan et al .
( 1992 ) , modifying also their @xmath43s by 0.1 and 0.25 dex , respectively .
figure 2 shows the comparison between the observed spectrum of procyon and five synthetic spectra computed for different values of the boron abundance .
the set of @xmath43s employed to compute them reproduces the spectrum of procyon with a similar quality to those illustrated by lemke et al .
( 1993 ) and duncan et al .
( 1997b ) .
the boron abundance which best matches the observed spectrum is a value between log n(b)@xmath46 and 2.1 ( in the customary scale where log n(b)@xmath47log ( b / h)@xmath112 ) .
the sensitivity of the observed feature to the boron abundance of the computed line is high , as can be seen in figure 2@xmath48 : changes of 0.1 dex around log n(b)@xmath49 can be appreciated .
taking into account the uncertainty in the continuum placement and a change of 50 k in the assigned effective temperature , the boron abundance provided by this analysis is log n(b)@xmath50 . since the possibility of unidentified lines located close to the @xmath0 2496.772 line
can not be excluded , this abundance should be considered strictly as an upper limit . our lte boron abundance for procyon is 0.15 dex smaller and 0.15 dex higher than those obtained by duncan et al .
( 1997b ) and lemke et al .
( 1993 ) , respectively . in the latter case , the model photosphere employed was computed using the marcs code with the opacity sampling method ( osmarcs ) as described by edvardsson et al .
( 1993 ) . using the lte code spectrum included in the uppsala synthetic spectrum package , we have computed synthetic spectra for procyon with the line list previously compiled and an osmarcs model atmosphere . in this case , the boron abundance which best matches the observed spectrum is between log n(b)@xmath51 and 2.0 , just 0.1 dex smaller than the value obtained using a kurucz model .
figure 3 shows the fits to the observed spectrum using both sets of model photospheres and a common boron abundance log n(b)@xmath52 .
it can be seen that they are very similar , being the larger differences in the wavelength range 2496.8 to 2497.2 ( very slight changes in the @xmath53 values used would be needed to obtain a similar good fit with the osmarcs model ) , but not affecting significantly the abundance derived from the @xmath0 2496.772 line .
kurucz models used in our analyses were computed with a mixing length parameter @xmath54 , and we have checked ( by running the program atlas9 using the unix - based version assembled by m. lemke and distributed through the ccp7 database ) that these models do not include overshooting .
osmarcs models do not include overshooting either and were computed with @xmath55 . a very good agreement between spectra computed with both sets of models
was found also by garca lpez , severino , & gomez ( 1995 ) for the analysis of the @xmath0 3131 region , showing negligible differences in beryllium abundances ( using marcs models treating the line blanketing by means of opacity distribution functions and applying wavelength - dependent scaling factors to the total h@xmath56 opacity ) . the very good agreement between these independent analyses in a metal - rich star like
procyon gives us a certain level of confidence to proceed to derive boron abundances in other stars .
it should be noted , however , that gross similarities among the analyses does not preclude the possibility of systematic errors .
furthermore , the presence of greater uncertainties in the model photospheres for metal - poor stars ( e.g. gratton , caretta , & castelli 1996 ) will constrain also the validity of abundances derived in those stars . using our line list and the stellar parameters listed in table 1 , we computed synthetic spectra of the two stars observed by us using both kurucz and osmarcs models .
computations were made with a step of 0.024 in wavelength and convolved with a gaussian to account for the spectral resolution of the observations .
we had initially assigned a metallicity of @xmath57 to bd @xmath173130 following the be analysis by garca lpez et al .
( 1998 ) ; however , a better overall fit to the region is obtained when values of [ fe / h]@xmath58 and @xmath59 are used for kurucz and osmarcs models , respectively .
figure 4 shows the comparison between observed and synthetic spectra ( computed for different boron abundances ) of this star using a kurucz model .
the major features surrounding the @xmath0 2496.772 line are reproduced , although they are not fully fitted .
the sensitivity of the synthetic spectra to boron abundance is seen in detail in figure 4@xmath48 .
the boron abundance of bd @xmath173130 is log n(b)@xmath60 , and changes of 0.1 dex around this value are clearly appreciated .
the spectrum computed with log n(b)=@xmath61 ( no boron ) shows much smaller features for both lines of the doublet , indicating that we have a positive detection of boron in this star .
the errors in the boron abundance associated with variations of @xmath62 k in @xmath30 , @xmath63 dex in @xmath33 , and @xmath63 dex in [ fe / h ] are @xmath64 , 0 , and @xmath64 , respectively .
the line is too weak to be sensitive to the uncertainty of the adopted microturbulence ( @xmath65 km s@xmath29 ) .
placement of the continuum was checked by computing a synthetic spectrum in the 24992501 region using the duncan et al .
( 1997b ) line list at those wavelengths . at @xmath0 2499.96
both synthetic and observed spectra reach a value of 1 .
taking into account the s / n and the possible errors in normalizing the observed spectrum , an additional error of @xmath66 dex can be considered .
combining all these uncertainties , the final lte boron abundance of bd @xmath173130 is log n(b)@xmath67 .
the analysis carried out with the same stellar parameters ( except [ fe / h ] ) using osmarcs models provided an abundance log n(b)@xmath68 , showing the same sensitivity to changes in boron abundance as well as similar uncertainties related to the choice of the stellar parameters and continuum location .
a comparison between the best fits to the observed spectrum obtained with both sets of models is shown in figure 5 .
a similar procedure was followed to obtain the boron abundance of hd 84937 .
a comparison between observed and synthetic spectra computed with a kurucz model is shown in figure 6 .
the line is weaker for a given abundance in hd 84937 than in bd @xmath173130 because boron is more ionized .
the observed spectrum is reproduced with a value between log n(b)@xmath69 and @xmath70 ( figure 6@xmath48 ) but , at this s / n level and with the uncertainties present in the parameters of the lines surrounding the line , a fit to the synthetic spectrum computed without boron ( log n(b)@xmath71 ) , or with other intermediate boron abundance , can not be ruled out .
this means that log n(b)@xmath72 must be considered an upper limit to the boron abundance in this star . as in the previous case , it was necessary to increase the adopted metallicity from [ fe / h]@xmath73 to @xmath74 to obtain an overall good fit .
figure 7 shows the comparison between the fits to the observed spectrum obtained using kurucz and osmarcs models with the same stellar parameters and boron abundance .
it can be seen how the overall behavior is very similar in both cases , as well as the upper limit at log n(b)@xmath72 .
the uncertainties of @xmath75 k in @xmath76 , @xmath63 dex in @xmath33 , @xmath77 in [ fe / h ] , and continuum placement will change this value within @xmath78 dex . taking these uncertainties into account ,
table 1 reflects a conservative upper limit log n(b)@xmath79 for this star .
spectral lines appear to be broader in bd @xmath173130 with respect to hd 84937
. the instrumental profile should be the dominant contributor to the line widths and this ought to be the same for the two stars .
additional broadening would arise if bd @xmath173130 were a spectroscopic binary such that a weaker line was present slightly shifted to lower wavelengths . however , this star does not appear as a binary or suspected binary star in the literature , especially in the survey of proper motions of about 500 stars studied by carney et al .
( 1994 ) . on the other hand ,
hd 84937 is listed as a suspected binary in the catalog of carney et al . and classified as a certain binary by stryker et al .
( 1985 ) from radial velocity measurements , although is not considered as a possible binary by nissen et al .
duncan et al . (
1997b ) observed two giant k - type stars belonging to the hyades , as well as the field giant @xmath80 gem .
the composite spectrum of the hyades stars and the spectrum of @xmath80 gem show similar broad lines which are , in general , deeper than those observed in bd @xmath173130 due to the big differences in metallicity . in any case , the spectral syntheses carried out by us and by duncan et al .
( 1997b ) show a very high sensitivity of the feature to the boron abundance in giant stars .
a complementary test to check the validity of the model photospheres employed in the analysis consists in comparing predicted and observed stellar fluxes at the wavelength region under study .
this is especially important in the ultraviolet , where an enormous number of lines are present such that the models may not account for the line blanketing .
we exploit the absolute flux calibration provided with the hst spectra by following the duncan et al.s ( 1992 ) procedure , using the stellar magnitude at the @xmath81 ( 2.2 @xmath82 m ) band to normalize the fluxes .
@xmath81 magnitudes for our stars ( 6.98 for bd @xmath1233130 and 7.10 for hd 84937 ) were taken from laird , carney , & latham ( 1988 ) .
theoretical fluxes corresponding to the kurucz models were obtained by interpolating the grid of models currently available at the ccp7 database , while they were computed individually for the osmarcs models used in the analyses . below @xmath0 4500 osmarcs models
are computed using opacity sampling techniques , while at longer wavelengths opacity distribution functions ( odf ) are employed .
the latter method is used for computing kurucz models at all wavelengths .
comparisons between both set of models show a very good agreement for ir wavelengths ( @xmath83 @xmath82 m ) , which decreases slightly with decreasing wavelength , and a significant difference appears for @xmath84 .
figure 8 shows the comparison in the uv ( surrounding the region ) for the kurucz and osmarcs models used for computing the synthetic spectra of figure 7 .
it can be seen in the upper panels the differences associated with both ways of considering the line blanketing .
these differences are strongly reduced by degrading the osmarcs fluxes to the same resolution as kurucz ones .
that is shown in the lower panels , where we have convolved the osmarcs fluxes with a gaussian with a sigma of 7 .
this behavior is consistent with the very small differences observed between their synthetic spectra , and indicates that the continuum fluxes are independent of model family .
comparisons between observed and predicted fluxes were carried out at @xmath0 2500 .
the ratios of the observed to predicted fluxes obtained using osmarcs models for bd @xmath17 3130 and hd 84937 are 1.24 and 1.20 , respectively .
these values are larger than the mean ratio @xmath85 resulting from the three metal - poor stars analyzed by duncan et al .
however , a ratio of 1.2 is compatible with the combined uncertainties in the models and in the hst flux calibrations , and we conclude that the models correctly reproduce the flux at @xmath0 2500 . with the aim of tracing the evolution of boron from the early times of the galaxy to the present , we decided to include in our study all metal - poor stars observed for their boron abundances .
these stars are analyzed in a recent work by duncan et al .
( 1997a ) , including their own observations as well as the three stars observed by duncan et al .
( 1992 ) .
rebull et al .
( 1996 ) derived the boron abundance in a very interesting metal - poor star : bd @xmath193442 .
these authors found an lte abundance of log n(b)@xmath86 , considering the following stellar parameters : [ fe / h]@xmath87 , @xmath88 , and @xmath89 k. this value was revised to @xmath90 by duncan et al .
( 1997a , 1998 ) using slightly different parameters ( [ fe / h]@xmath91 , @xmath92 , and @xmath93 k ) .
given that we had found only an upper limit to the abundance of the less metal - poor star hd 84937 with similar @xmath30 , we decided to carry out the analysis of bd @xmath193442 using our tools .
we took the observed spectrum from figure 2 of rebull et al .
( 1996 ) and computed synthetic spectra with different boron abundances using the same stellar parameters .
the fit obtained is very similar to that of rebull et al .
( 1996 ) and the boron abundance derived is log n(b)@xmath42 . with respect to the sensitivity of the synthetic spectra to the abundance , we found the same behavior as in hd 84937 , i.e. , there are very small differences between spectra computed within a range of three orders of magnitude in boron abundance . in order to have the abundance for this star on our @xmath30 scale
, we computed new spectra using the stellar parameters listed in table 1 , but increasing the metallicity by 0.1 dex to obtain a better overall fit ( with both kurucz and osmarcs model atmospheres ) .
our best estimate for the boron abundance of this star is an upper limit at @xmath94 dex , and considering the usual uncertainties in stellar parameters , s / n , and continuum placement this value could change by @xmath95 dex .
the adopted upper limit taking into account this uncertainty is then log n(b)@xmath96 .
a similar behavior can be seen for the star bd @xmath973578 in figure 2@xmath98 of duncan et al .
( 1997a ) .
there is but a very small difference between their synthetic spectra computed with log n(b)=@xmath99 and without boron .
our analysis of this star gives an upper limit of log n(b)@xmath100 , using our adopted @xmath30 and @xmath33 but increasing the metallicity to [ fe / h]@xmath73 and @xmath101 for kurucz and osmarcs models , respectively ( instead of the value @xmath102 listed in table 1 ) .
the upper limit adopted taking into account the uncertainties in the analysis is log n(b)@xmath103 .
these spectral syntheses show that the combination of high effective temperature and low abundance makes it difficult to obtain accurate estimates of boron in metal - poor stars with @xmath104 k using the @xmath0 2496.772 line .
therefore , we consider the available spectra of these stars ( including also bd @xmath105740 , for which we have not computed synthetic spectra ) to provide upper limits to the boron abundances .
this consideration removes four important objects ( in terms of their low metallicities ) from the sample used to delineate the boron evolution .
a systematic effect present in all our spectral syntheses is that the metallicity indicated by the far uv ( boron ) spectra seems to be higher than that needed to reproduce the near uv ( beryllium ) and optical regions . in two cases ( hd 84937 and bd @xmath193442 ) these differences are within the error bars of the adopted metallicities , in another case ( bd @xmath173130 )
the difference is 0.1 dex higher than the error bar , and for bd @xmath973578 this difference is 0.100.15 dex larger than the adopted error bar .
a poor fit to the metallic lines surrounding the region is also observed in the spectral synthesis provided by rebull et al .
( 1996 ) for bd @xmath193442 .
spectral lines in the region are not resolved with the resolution of these ghrs observations .
a line like the @xmath0 2496.533 is very deep and saturated , and thus the core of the line is formed at the top of the atmosphere .
we have checked that the model atmospheres are sufficiently extensive that the line is computed correctly .
there remains the question as to whether the model represents adequately the real upper photosphere .
errors in the uv opacity , departures from lte , or effects of convective overshoot and thermal inhomogeneities might contribute to the fact that the far uv synthetic spectra seem to need a higher metallicity to reproduce the observations .
although there is no clear correlation between these differences and the adopted metallicities for the four stars considered , this effect could also be metallicity dependent and so could cause a slight change of slope in the boron[fe / h ] relation discussed in this paper .
to adapt the abundances given by duncan et al .
( 1997a ) to our stellar parameters for those stars for which we have not carried out a detailed spectral synthesis , we computed the @xmath0 2496.772 line with the boron abundances and stellar parameters listed in that work using kurucz model atmospheres , and measured the corresponding synthetic equivalent widths .
the adopted lte abundances were those which matched these equivalent widths with the stellar parameters listed in table 1 .
the sensitivity of our spectral synthesis to the stellar parameters is similar to that carried out by duncan et al .
( 1997a ) , and we adopted their final errors to the abundances .
table 1 shows the final abundances obtained in this way .
the differences between lte boron abundances obtained using our stellar parameters and those of duncan et al .
( 1997a ) are in the range @xmath106 to 0.08 dex for all the stars , except for bd @xmath105740 with a difference of 0.25 dex .
these differences are mainly due to differences in @xmath30 ( up to 170 k for bd @xmath105740 ) , and for stars with low metallicity the changes in abundance are larger for a given difference in @xmath30 with respect to more metal - rich stars .
we also derived lte abundances for these stars using osmarcs models and the equivalent widths previously computed with kurucz model atmospheres .
these abundances are also listed in table 1 . as expected , there is a systematic slight difference between both sets of abundances with a mean value @xmath107 . in all abundance estimates
we have used a microturbulent velocity of 1 km s@xmath29 . for the most metal - poor stars analyzed the synthetic spectra have shown that the dependence of boron abundance on the microturbulence is negligible . in the most metal - rich stars of the sample , with higher boron abundances , the effect of changing the microturbulence by one additional km s@xmath29 induces a change of 15% in the equivalent widths of the lines .
changes of this order affect the boron abundance by 0.010.07 dex for stars with @xmath108[fe / h]@xmath109 .
as explained in section 3 , effective temperatures were estimated using the @xmath110 and @xmath111 calibrations of alonso et al .
( 1996a ) , and corrected afterwards until a good fit to the @xmath0 3131 region was obtained ( garca lpez et al . 1998 ) . with this procedure , two of the stars considered in this paper (
bd @xmath105740 and hd 140283 ) show @xmath30 values with significant differences with respect to those obtained using other calibrations . for bd
@xmath105740 the temperature is directly a mean value between 6180 and 6410 k from @xmath31 and @xmath32 colors , respectively .
however , given that it is hotter than 6000 k we are considering its boron measurement as an upper limit and it is not relevant for the discussion in the next section .
the temperature adopted for hd 140283 is lower than other values usually quoted in the literature .
using e.g. @xmath112 k ( carney 1983 ; alonso , arribas , & martnez - roger 1996b ) , the lte boron abundance for this star would rise by about 0.05 dex only .
kiselman ( 1994 ) showed that the approximation of line formation in local thermodynamic equilibrium leads to significant errors in abundance determinations for solar - type metal - poor stars using lines .
these lines are especially susceptible to nlte effects because the overexcitation and overionization effects work in the same direction in making the lines weaker than if boltzmann and saha equilibria were valid .
this is not the case for the lines that are used for beryllium abundance analysis here
overionization and overexcitation work in opposite directions on the line strengths as was shown by garca lpez et al .
( 1995 ) and kiselman & carlsson ( 1995 ) .
the nlte behavior of lithium is more complicated , partly because the generally higher lithium abundances make the lithium atoms susceptible for various nlte effects typical for strong lines .
carlsson et al . (
1994 ) have published nlte abundance corrections for an extensive grid of photospheric parameters , and pavlenko et al .
( 1995 ) and pavlenko & magazz ( 1996 ) provide nlte curves of growth which extend to very low effective temperatures .
kiselman & carlsson ( 1996 ) provide nlte corrections for lte boron abundances computed using osmarcs models .
these should in principle not be applied to lte abundances acquired with other models .
the fact an lte analysis of a line gives similar results for two different models is not in itself a guarantee that the nlte analysis would .
line formation in nlte involves coupling between different frequency regions and thus a more complicated dependence on the photospheric temperature and pressure structures and the background opacities than what is the case in lte .
the different treatment of uv opacities ( see our discussion on stellar fluxes ) makes it possible that consistent nlte line modeling using the kurucz models could give results differing from the osmarcs results , even though the lte abundances are so similar . in order to be fully consistent , we choose as our final estimates of boron abundances our osmarcs lte results with the nlte corrections of kiselman & carlsson applied .
correcting the lte abundances obtained with kurucz models using the kiselman & carlsson prescriptions would mean that the final abundances would be formally on the osmarcs nlte scale and slightly overestimated .
the resulting abundances are given in table 1 .
we will adopt for the nlte abundances the same error bars derived in the lte analysis .
the sensitivity on uncertainties in stellar parameters is not significantly different in lte and nlte .
the dependence on atomic data is much more difficult to asses ( especially on @xmath43 ) , and the sensitivity on other data is of second order .
furthermore , possible systematic errors are not easy to estimate .
the errors in the stellar radiation fields are probably the most pertinent ones which bother the nlte corrections , but we are considering that the continuous opacities employed in the osmarcs models reproduce the observed fluxes .
hd 84937 was selected as a key target for this work because it is the only star for which there exist previous measurements of @xmath113li and @xmath14be . however , due to its high @xmath30 , our observations failed to provide a convincing detection of the @xmath0 2496.772 line and consequently we were forced to set an upper limit to the boron abundance .
this disappointment led us to reexamine published spectra of similarly warm extremely metal - poor stars and to conclude that upper limits to rather than certain determinations of the boron abundance may be reached .
this result has some influence on determinations of the evolution of the boron abundance with metallicity . in the case of bd @xmath173130 , a cool giant
, a concern may be that boron has been depleted below its initial abundance . to investigate this concern , we compare its li and be abundances with those of some other stars in our sample .
equivalent widths of the @xmath0 6708 line were taken from thorburn ( 1994 ) and ryan et al .
( 1996 ) , and the abundances were derived using the stellar parameters adopted in table 1 and nlte curves of growth provided by ya .
v. pavlenko ( obtained as described in pavlenko et al .
1995 and pavlenko & magazz 1996 ) .
beryllium abundances were derived by garca lpez et al .
li and be abundances are listed in table 2 , where the latter include an upwards correction of 0.1 dex to consider the mean difference with the gravities derived by nissen et al .
( 1997 ) from _ hipparcos _ parallaxes ( which does not affect either boron or lithium abundances ) . evidently but not surprisingly li is depleted in bd @xmath173130 by more than 0.7 dex .
a part of the observed depletion may have occurred as a dilution when the star evolved to the giant phase .
boron and beryllium are less fragile elements than lithium , and it is unlikely that their abundances have been reduced by nuclear reactions .
it can not be completely ruled out that these abundances have been slightly reduced by dilution .
it will be shown below that the be and b abundances of the star fit the trends with metallicity defined by the other stars , and a small increment of the observed abundances of bd @xmath173130 would not change the general conclusions about the synthesis of boron .
initial results on the evolution of the b abundance with metallicity ( [ fe / h ] ) were provided by duncan et al .
( 1992 ) from just three lte b abundances of halo stars .
the linear trend between log n(b ) and [ fe / h ] was confirmed in essence by kiselman & carlsson ( 1996 ) who included non - lte effects for boron , and by duncan et al .
( 1997a ) who analyzed additional stars .
our rediscussion adds two stars and recognizes that only upper limits to the b abundance are presently obtainable for the warmest stars .
in addition , we give b and be abundances for stellar parameters derived in a uniform way for the entire sample .
figure 9 shows the lte and nlte boron abundances of those stars listed in table 1 against [ fe / h ] . in the upper panel
the symbols are also labeled with the adopted effective temperatures . as explained in section 4 , we are only considering as real measurements those made in metal - poor stars with @xmath114 k. nlte corrections also affect more strongly to the hotter stars , inducing a larger scatter in the boron - metallicity plot when including the abundances associated with the upper limits . for comparison ,
we have also included in figure 9 the boron abundances of two solar - metallicity stars : @xmath115 peg and @xmath116 uma , taken directly from lemke et al .
nlte effects on the b abundances were taken into account for these stars following kiselman & carlsson ( 1996 ) .
two sets of fits are shown in figure 9 .
dashed lines represent linear regressions performed using all metal - poor stars with @xmath117 k , and solid lines restrict the fits to cool stars with [ fe / h]@xmath118 .
this distinction is done to compare with the behavior of beryllium abundances , for which a change of slope at [ fe / h]@xmath119 is apparent from the observations ( garca lpez et al . 1998 ) .
all regression analyses were carried out taking into account the error bars in boron abundances and metallicities , using the routine _ fitexy _ of press et al .
( 1992 ) available in the idl astronomy user s library .
these are the results : * lte , @xmath117 k : + @xmath120 } , \ , n=9 , \ , \chi^2_\nu=0.86 , \ , q=0.54 $ ] , + where @xmath121 is the number of stars included in the fit , @xmath122 is @xmath123 divided by the number of degrees of freedom @xmath124 ( also called _ reduced _ @xmath123 ; @xmath125 in this case ) , and @xmath126 is the probability that a correct model would give a value equal or larger than the observed chi squared ( a small value of @xmath126 indicates a poor fit ; @xmath127 provides an acceptable fit ) . * lte , @xmath117 k and [ fe / h]@xmath118 : + @xmath128 } , \ , n=7 , \
, \chi^2_\nu=0.42 , \ , q=0.84 $ ] . *
nlte , @xmath117 k : + @xmath129 } , \ , n=9 , \ , \chi^2_\nu=0.86 , \
, q=0.54 $ ] .
* nlte , @xmath117 k and [ fe / h]@xmath118 : + @xmath130 } , \ , n=7 , \ , \chi^2_\nu=0.34 , \ ,
q=0.89 $ ] .
the plots make evident that a linear increase of the boron abundance with increasing metallicity describes very well the trend observed in the data , and a change of slope may occur at the halo - disk transition at [ fe / h ] @xmath131 .
duncan et al . (
1997a ) found slopes of @xmath132 and @xmath133 for their linear fits to 11 stars in lte and nlte , respectively .
our corresponding slopes of @xmath134 and @xmath135 , obtained using 9 stars ( without upper limits ) are consistent with those of duncan et al . within
the error bars .
if we include the data associated with the metal - rich stars @xmath115 peg and @xmath116 uma , which have not been adjusted to our @xmath30 scale , the lte and nlte slopes do not change appreciably ( @xmath136 and @xmath137 , respectively ) .
the slopes are slightly steeper and outside the error bars when compared with the results of duncan et al . , if we consider only those 7 stars with [ fe / h ] @xmath118 .
we will consider our nlte abundances and the fit for stars with @xmath117 k and [ fe / h]@xmath118 as the actual values to be used in the discussion performed below this point . in studies of boron evolution ,
oxygen abundances may be a preferable alternative to the iron abundances for astrophysical reasons .
in particular , oxygen is synthesized primarily by the most massive supernovae whose ejecta ( see below ) may be the source of the energetic particles responsible for boron production by spallation . unfortunately , observational determinations of oxygen abundances in metal - poor stars have been a matter of controversy during the last years ( e.g. abia & rebolo 1989 ; bessell , sutherland , & ruan 1991 ; nissen et al . 1994 ; cavallo , pilachowski , & rebolo 1997 ) .
we examined the boron - oxygen trend using the oxygen abundances found by cavallo et al .
( 1997 ) from high - resolution spectra of the ir triplet in a sample of 24 stars with @xmath138\leq -0.48 $ ] .
these results were employed to define the relation between [ o / h ] and [ fe / h ] which was then used to establish the b versus o trend .
we find : @xmath139}$ ] .
this slope is steeper than the comparable trend found by duncan et al .
( 1997a ) , with a slope of 0.82@xmath140 when using a mean [ o / h ] versus [ fe / h ] relation established using abundances extracted from the literature for the stars in their sample .
almost in parallel with our analysis of boron , israelian et al .
( 1998 ) are deriving oxygen abundances in a wide sample of metal - poor stars , including those considered here , using oh lines in the 30853256 region and the same stellar parameters employed in this paper .
until controversy about oxygen in metal - poor stars is clearly resolved , it must suffice to note that the boron and oxygen abundances are approximately linearly related .
two modes of nucleosynthesis of boron compete ( in theory ) for primacy : spallation reactions driven by energetic ( cosmic rays ) particles in extra - stellar environments , and neutrino - induced spallation of @xmath141c occurring in type ii supernovae .
a measure of the relative contributions of the two modes is offered by the b / be ratio ( at a given [ fe / h ] ) because beryllium is synthesized ( in theory ) only by spallation reactions with high energy particles .
moreover , in these spallation reactions , the b / be production ratio is essentially independent of the composition and the energy spectrum of the high energy particles .
thus , the observed b / be ratio reveals the relative roles of the two modes of boron nucleosynthesis .
figure 10 shows the comparison between boron ( nlte ) and beryllium abundances against metallicity .
beryllium data were taken from garca lpez et al .
( 1998 ) , who carried out a detailed spectral synthesis of the @xmath0 3131 region for 28 stars with @xmath142\leq -3 $ ] , using high - resolution observed spectra .
these authors also included all metal - poor stars with accurate be abundance available , and repeated measurements on the same star were averaged both in be and in [ fe / h ] .
also shown for comparison are the be abundances given by lemke et al .
( 1993 ) for the metal - rich stars @xmath115 peg and @xmath116 uma , estimated from boesgaard s ( 1976 ) equivalent widths measured on photographic spectra . a linear regression analysis for stars with [ fe / h ] @xmath118 , without considering the upper limits ( not shown in fig .
10 ) , provides : @xmath143 } , \
, n=29 , \ , \chi^2_\nu=0.81 , \ ,
q=0.75 $ ] .
these values of @xmath122 and @xmath126 indicate an excellent fit over a significant sample of metal - poor stars . these are lte abundances but corrections for non - lte effects are small ( @xmath144 dex in general see garca lpez et al .
1995 and kiselman & carlsson 1995 ) .
the slopes of the observed trends against metallicity are very similar for be and b for stars with [ fe / h ] @xmath118 .
indeed , the slopes are identical within the error bars .
duncan et al .
( 1997a ) found a similar behavior when comparing their nlte b abundances and published be abundances for their sample of stars .
using the be abundances given by garca lpez et al .
( 1998 ) , derived with the stellar parameters listed in table 1 and shown in table 2 , the average b / be ratio for the seven stars with @xmath117 k and [ fe / h ] @xmath118 is b@xmath145/be @xmath146 ( 1 @xmath147 ) .
for all of these stars , with the exception of bd @xmath173130 , independent beryllium measurements are available in the literature based on high - resolution ccd spectra , slightly different stellar parameters , and employing spectral synthesis or curves of growth .
garca lpez et al .
( 1995 ) showed that these independent lte analyses provide very similar be abundances when adjusted for differences in the stellar parameters ( especially the surface gravity ) .
the average be abundances for these seven stars differ by @xmath148 dex from the individual values listed in table 2 . using the average
be values we obtain a mean ratio b@xmath145/be @xmath149 .
figure 11 shows the behavior of log ( b@xmath145/be ) against metallicity .
the solid line represents the mean value and dotted lines show the @xmath150 uncertainties : @xmath151 .
all ratios are compatible with this mean value and its uncertainty within the error bars .
a linear regression fit taking into account the errors in both coordinates provides : @xmath152 } , \ , n=7 , \
, \chi^2_\nu=0.32 , \ ,
q=0.90 $ ] , which is fully consistent with the mean value and is shown as a dashed line in figure 11 .
duncan et al .
( 1997a ) obtained a mean value b@xmath153/be @xmath154 using 9 stars .
be abundances adopted by duncan et al .
( 1997a ) are , in general , larger than those adopted here and that is the main reason why they obtain a smaller b / be ratio .
it is worthwhile to consider the trends given by the regression analyses of boron and beryllium against metallicity , which better represent their overall behavior .
molaro et al .
( 1997 ) derived beryllium abundances for a sample of metal - poor stars and , combining them with measurements available in the literature , obtained a linear fit of be abundance against metallicity with a slope of @xmath155 for 19 stars in the interval @xmath156 <
-0.8 $ ] . this value is compatible with the slope @xmath157 found by garca lpez et al .
( 1998 ) for 29 stars in the interval @xmath158 < -1 $ ] . using this relation and the fit obtained in this work between log n(b)@xmath153 and [ fe / h ] for stars with @xmath117 k and [ fe / h ] @xmath118 , we find a mean ratio of b@xmath145/be @xmath159 .
using our boron relation and the beryllium fit provided by molaro et al .
( 1997 ) , this ratio is just 17 . two observational facts about the b and
be abundances stand out : ( i ) metal - poor ( [ fe / h]@xmath160 ) stars have a constant boron - to - beryllium ratio @xmath10 , and ( ii ) the relation between boron ( or beryllium ) and metallicity has a slope of @xmath9 .
the linear dependence , first seen for be vs fe by gilmore et al .
( 1992 ) and then by duncan et al .
( 1992 ) for b vs fe , was a surprise .
simple models of spallation of ambient cno nuclei in the interstellar medium by cosmic ray protons and @xmath12s had predicted a quadratic dependence ( b @xmath161 fe@xmath162 ) .
several explanations for the observed linear dependence have been advanced .
in essence , most explanations invoke spallation of cno nuclei in energetic collisions with ambient protons and @xmath12s . if the composition of the energetic particles is greatly enriched in cno , as could be assumed for supernovae ejecta or winds from wolf - rayet stars , a quasi - linear relation between b ( and be ) and metallicity would result ( duncan et al .
1992 ; vangioni- flam et al .
other possibilities , such as local spallation by high - energy protons and @xmath12s in enriched regions around supernovae , seem less probable ( feltzing & gustafsson 1994 ) . at any rate , fact ( ii ) is to be explained largely by astrophysical issues .
the simplest interpretation of fact ( i ) is that b and be are made by the same nucleosynthetic process .
this process is spallation of cno nuclei by collisions with protons and @xmath12s .
early calculations of yields from spallation gave a b / be ratio near 20 : reeves ( 1994 ) gives the ratio as 17 for production by high energy ( @xmath163 100 mev nucleon@xmath29 ) cosmic rays .
the primary factors setting the ratio are the cross - sections for the spallation reactions , i.e. , nuclear physics rather than astrophysics .
theoretical yields are thoroughly discussed by ramaty et al .
( 1996 ) who investigate their dependence on the energy spectrum and composition of the energetic particles ( cosmic rays ) , and on the composition of the ambient medium .
this investigation shows that a ratio of b / be @xmath10 from spallation does not constrain the composition of either the energetic particles or the ambient medium , and , provided the energy spectrum of the particles is not biased to threshold energies for the proton and @xmath12 on cno nuclei spallation reactions , the ratio does not constrain the energy spectrum .
ramaty et al .
stress that the various combinations of energy and composition of energetic particles and composition of the ambient medium call for different amounts of energy to be put into energetic particles : certain combinations are very inefficient .
if the particles have the predicted composition of the ( most massive ) supernovae ejecta or wolf - rayet winds , ramaty et al.s condition on the energy available for high energy particles is met .
the condition is very difficult to impossible to meet if the high energy particles have the composition of the sun , the galactic cosmic rays or of the ejecta of low mass type ii supernovae .
it is intriguing that this argument about the composition of the high energy particles is consistent with that adduced from the linear relation between b ( and be ) and fe .
this constancy of be / fe and b / fe for metal - poor stars led ramaty et al .
( 1997 ) to relate the production rates of be and b to the fe production rate by type ii supernovae via cosmic - ray acceleration .
considering that the @xmath16b/@xmath15b value in metal - poor stars is probably not lower than the solar system ratio of @xmath164 ( chaussidon & robert 1995 ) , a b / be @xmath10 would imply an initial acceleration energy of about 60 mev nucleon@xmath29 ( see fig .
13@xmath165 of ramaty et al .
1997 ) , and a rapid increase in the initial energy needed ( @xmath166 mev nucleon@xmath29 ) for slightly lower values ( 1718 ) of b / be .
beryllium is thought to be synthesized only by spallation reactions but boron synthesis by spallation may be supplemented by type ii supernovae in which boron is made by neutrino - induced spallation of carbon ( the @xmath124-process ) .
timmes , woosley , & weaver ( 1996 ) in a detailed study of galactic chemical evolution suggest that predicted yields of boron by supernovae reproduce well the b versus fe trend of the halo and disk _ without _ the introduction of spallation processes .
this is incompatible with the attribution of be synthesis to spallation processes and the observed ratio b / be @xmath10 which implies that boron synthesis is also by spallation processes .
we conclude that yields of boron from the @xmath124-process have been overestimated .
vangioni - flam et al . (
1996 ) , on the basis of few published b abundances , concluded that the @xmath124-process is at least a factor of 5 less efficient than supposed by timmes et al .
( 1995 ) .
alternative models of chemical evolution of the galactic halo ( e.g. , casuso & beckman 1997 ) also give consistent results with the observed linear trends of b and be vs. fe at very low metallicity .
these authors consider an exponentially increasing flow of gas from the halo to form the bulge
. the strong be and b enrichment in the halo is attributed to the presence of more star - forming gas at very low metallicity , and to increased yields of carbon in intermediate and low - mass metal - poor stars .
the be / b ratios that they predict are also similar to the observed ones .
it seems that under certain hypothesis for the evolution of the gas and for the initial mass function , conventional cosmic ray spallation processes may then provide an explanation to the observations . in short ,
the ratio b / be @xmath10 is strong evidence that b and be are products of spallation reactions between protons and/or @xmath12s and cno nuclei . the beryllium abundances ( figure 10 ) indicate that there is a change of slope in the run of be with fe at about [ fe / h ] @xmath119 , the canonical transition between halo and disk ( a more detailed discussion based on the be data is presented in garca lpez et al .
the very limited data on the boron abundance in disk stars suggests that the b versus fe trend follows the be versus fe trend such that b / be remains constant at @xmath10 .
additional data for boron at these metallicities is crucial to confirm this and for a better understanding of the halo - disk transition .
the meteoritic abundances ( grevesse , noels , & sauval 1996 ) give b / be @xmath167 in agreement with our result from metal - poor stars .
bd @xmath168 3130 observed and analyzed here is the most metal - poor star for which a firm measurement of the boron abundance presently exists ( two slightly more metal - poor stars
bd @xmath169 and bd @xmath170 740 have been analyzed but we consider that the @xmath0 2497 line has not been positively identified in published spectra . ) our result and slightly revised results from other metal - poor stars establish the run of boron with iron for halo stars : the boron abundance increases linearly , with a slope @xmath171 , with the iron abundance . with beryllium abundances
consistently derived for the same stars using the same atmosphere parameters , the b / be ratio is found to be @xmath10 .
a direct interpretation of these results is that b and be in the halo were synthesized by spallation reactions between ambient protons and @xmath12s and high energy particles with a composition enriched in cno .
the b / be ratio is in fine agreement with the predictions ( vangioni - flam et al .
1996 ; ramaty et al . 1996 , 1997 ) .
this is a significant result because the predicted b / be ratio is largely independent of the details of the modeling of the spallation process being largely determined by laboratory measurements of spallation cross - sections .
the linear trend of b ( and be ) with fe is dependent on the astrophysics of spallation in the halo but can be reproduced with plausible assumptions ( vangioni - flam et al .
1996 ) .
a consistent picture of boron synthesis has emerged for the halo of our galaxy .
stars more metal - poor than bd @xmath168 3130 are now known but it will not be easy to extend measurements of boron abundance to these stars
. the @xmath0 2497 line will be too weak in the warmest of these stars , as we have found for less extreme stars .
some extension to lower metallicity will be possible for cool stars but for the fact that the candidate stars are faint requiring prohibitively long exposures with hst .
almost certainly , the boron plateau signifying detection of primordial boron will elude us ; boron ( and beryllium ) from homogeneous and inhomogeneous big bangs is predicted to have an abundance 7 or 8 dex below the abundance found here for bd @xmath168 3130 ( thomas et al .
1993 , 1994 ) .
it is a pleasure to acknowledge the following people for their help with the analyses described in this work : martin asplund for computing several osmarcs models and their corresponding fluxes , ya .
v. pavlenko for providing nlte li curves of growth , and the staff at stsci and ecf for their support during all phases of these hst observations .
the comments of an anonymous referee have been of value in improving the content of this article .
this work was partially supported by the space telescope science institute ( grant go-05862 ) , and the spanish dges under projects pb92 - 0434-c02 - 01 and pb95 - 1132-c02 - 01 . be and bg acknowledge support from the swedish natural science research council ( nfr ) and the swedish space board .
lccccccc bd @xmath173130 & 5170@xmath172150 & [email protected] & @xmath173 & @xmath174 & @xmath175 & 0.05 & 1,a hd 84937 & 6210@xmath172120 & [email protected] & @xmath176 & @xmath79 & @xmath79 & @xmath177 & 1,a bd @xmath193442 & 6200@xmath172150 & [email protected] & @xmath178 & @xmath96 & @xmath96 & @xmath179 & 2,a bd @xmath973578 & 6080@xmath172120 & [email protected] & @xmath180 & @xmath103 & @xmath103 & @xmath181 & 3,a hd 142373 & 5795@xmath172100 & [email protected] & @xmath182 & @xmath183 & 1.96 & 2.03 & 3,b hd 184499 & 5810@xmath172100 & [email protected] & @xmath184 & @xmath185 & 1.71 & 1.83 & 3,b hd 76932 & 5800@xmath172100 & [email protected] & @xmath186 & @xmath187 & 1.68 & 1.85 & 3,b hd 201891 & 5870@xmath172100 & [email protected] & @xmath188 & @xmath189 & 1.52 & 1.75 & 4,b hd 194598 & 5960@xmath172120 & [email protected] & @xmath190 & @xmath191 & 1.38 & 1.67 & 3,b hd 64090 & 5385@xmath172100 & [email protected] & @xmath192 & @xmath193 & 1.02 & 1.21 & 3,b hd 19445 & 5810@xmath172150 & [email protected] & @xmath194 & @xmath195 & 0.18 & 0.53 & 4,b hd 140283 & 5550@xmath172100 & [email protected] & @xmath196 & @xmath197 & @xmath198 & 0.23 & 4,b bd @xmath105740 & 6295@xmath172120 & [email protected] & @xmath199 & @xmath200 & 0.42 & 1.31 & 3,b lccc hd 76932 & @xmath186 & @xmath201 & @xmath202 hd 201891 & @xmath188 & @xmath203 & @xmath204 hd 194598 & @xmath190 & @xmath205 & @xmath206 hd 64090 & @xmath192 & @xmath207 & @xmath208 hd 19445 & @xmath194 & @xmath209 & @xmath210 hd 140283 & @xmath196 & @xmath211 & @xmath206 bd @xmath173130 & @xmath173 & @xmath212 & @xmath213 | we have observed the @xmath0 2497 line to derive the boron abundances of two very metal - poor stars selected to help in tracing the origin and evolution of this element in the early galaxy : bd @xmath1233130 and hd 84937 .
the observations were conducted using the goddard high resolution spectrograph on board the _ hubble space telescope_. a very detailed abundance analysis via spectral synthesis has been carried out for these two stars , as well as for two other metal - poor objects with published spectra , using both kurucz and osmarcs model photospheres , and taking into account consistently the nlte effects on the line formation .
we have also re - assessed all published boron abundances of old disk and halo unevolved stars .
our analysis shows that the combination of high effective temperature ( @xmath2 k , for which boron is mainly ionized ) and low metallicity ( [ fe / h ] @xmath3 ) makes it difficult to obtain accurate estimates of boron abundances from the @xmath0 2497 line .
this is the case of hd 84937 and three other published objects ( including two stars with [ fe / h ] @xmath4 ) , for which only upper limits can be established .
bd @xmath1233130 , with [ fe / h ] @xmath5 and log n(b)@xmath6 , appears then as the most metal - poor star for which a firm measurement of the boron abundance presently exists .
the evolution of the boron abundance with metallicity that emerges from the seven remaining stars with @xmath7 k and [ fe / h ] @xmath8 , for which beryllium abundances were derived using the same stellar parameters , shows a linear increase with a slope @xmath9 .
furthermore , the b / be ratio found is constant at a value @xmath10 for stars in the range @xmath11<-1 $ ] .
these results point to spallation reactions of ambient protons and @xmath12 particles with energetic particles enriched in cno as the origin of boron and beryllium in halo stars . |
arrays of stochastically driven nonlinear dynamical systems may exhibit nonequilibrium phase transitions of continuous or discontinuous type , for a recent review see @xcite , cf . also @xcite .
concepts developed to describe equilibrium phase transitions such as symmetry or ergodicity breaking , order parameter , critical behaviour , critical exponents etc . have been successfully transfered to noise induced nonequilibrium phase transitions .
the structure of the theory will be generically of mean field type , if infinite globally coupled arrays are studied which allows for a number of analytical results .
remarkably , essential characteristics of phase transitions can already be found in the case of a single stratonovich model .
this is mainly due to the multiplicative nature of the noise .
models driven by additive noise do not show this peculiar property .
the langevin equation for the single - site stratonovich model @xcite reads @xmath4 where @xmath5 is a control parameter , @xmath6 denotes the strength of the noise and @xmath7 is a wiener process with autocorrelation @xmath8 .
equation ( [ 1le ] ) is interpreted in the stratonovich sense as indicated by the symbol @xmath9 .
the stratonovich model describes , e.g. , the overdamped motion in a biquadratic potential @xmath10 where the control parameter is stochastically modulated , @xmath11 , with a gaussian white noise @xmath12 . the associated fokker - planck equation ( fpe ) describing the evolution of the probability density @xmath13 is @xmath14p\big\}.\end{aligned}\ ] ] equation ( [ 1fpe ] ) has a weak stationary solution , a dirac distribution @xmath15 located at the common zero @xmath16 of drift and diffusion coefficient , which is also a zero of the stochastic flow in eq .
( [ 1le ] ) .
if the system is initially at @xmath16 it will always stay there .
furthermore , there exist spatially extended strong stationary solutions determined up to a constant factor , @xmath17 .
@xmath18 will live on @xmath19 if the initial distribution lives on @xmath20 , and on @xmath21 $ ] , if the initial distribution lives on @xmath22 .
the constant is determined such that the solution is normalized integrating over the support and can be interpreted as probability density , i.e. @xmath23 provided @xmath24 .
for @xmath25 the normalization @xmath26 diverges since the integrand in ( [ sm1z ] ) scales like @xmath27 as @xmath28 in this case it can be shown ( see [ stratito ] ) that a weakly normalized version converges to the known weak solution @xmath15 and @xmath16 is an absorbing fixed point of the system .
if fractions of the initial distribution of given weights live on @xmath29 , on @xmath30 , and on @xmath31 , all will keep their weight and evolve to the stationary probability densities living on their respective support as guaranteed by a @xmath32-theorem @xcite .
the stratonovich model exhibits a strong ergodicity breaking @xcite depending on the control parameter @xmath5 , since the state space decomposes into regions where the system can not reach one region if it has started in a different one . for @xmath33
the only stationary solution is @xmath15 , i.e. the fixed point @xmath16 of the stochastic dynamics is absorbing .
additionally , for @xmath34 we have the spatially extended solution ( [ sm1p ] ) living on @xmath35 depending on the initial distribution .
this is reflected by the mean value @xmath36 obviously , @xmath37 can serve as an order parameter and shows a critical behavior @xmath38 as @xmath39 with @xmath40 .
note that also the location of the maximum of the spatially extended density undergoes a bifurcation , @xmath41 for @xmath42 and @xmath43 for @xmath44 .
the critical behaviour of an array of _ infinitely _ many globally coupled stratonovich models has been thoroughly investigated in @xcite .
the scaling of higher moments was considered in @xcite , see also @xcite .
the stationary probability density is the solution of a nonlinear fokker - planck equation which depends on the order parameter .
the scaling behaviour of the order parameter is analytically determined , @xmath45 as @xmath46 with @xmath47 and @xmath48 , where @xmath49 is the strength of the harmonic coupling between the systems @xcite .
the strong coupling limit , @xmath50 , of an infinite array of globally coupled systems was analytically treated already in the pioneering paper @xcite , cf .
also @xcite . in this paper
we investigate nonequilibrium phase transitions in _ finite _ arrays of globally coupled stratonovich models in the strong coupling limit .
we introduce center of mass and relative coordinates and exploit that for strong coupling there is a clear separation of time scales .
the relative coordinates relax very quickly to zero and the system behaves as a single entity described by the center of mass coordinate @xmath51 .
thus , we can adiabatically eliminate the relative coordinates .
the stationary probability density of the center of mass coordinate @xmath52 is analytically calculated for a class of nonlinear systems and a scheme to determine the transition manifold in the parameter space is developed . for finite arrays of stratonovich models the mean value @xmath53 of the center of mass coordinate is computed analytically . near a critical value of the control parameter
@xmath5 the stochastic system shows a scaling behaviour similar to the order parameter of the single stratonovich model with the same critical exponent @xmath54 but with a different @xmath1 which is also given analytically . keeping a finite small distance to @xmath1 we recover for @xmath3 the known result of the self consistent theory @xcite with critical exponent @xmath55 ,
see above . for finite @xmath0
we identify a crossover value of the control parameter @xmath56 .
for @xmath57 we have a linear scaling as for @xmath58 whereas for @xmath59 a square root behaviour as for @xmath3 is observed .
the analytical results are coroborrated by numerical simulations . recently
, finite arrays of ( non- ) linear stochastic systems have been investigated also by muoz et al .
@xcite , and by hasegawa @xcite .
muoz et al . tried to obtain for multiplicative noise characteristics of the probability density of the mean field for finite @xmath0 .
they argued that the langevin equation for the mean field variable is of similar form as the langevin equation for a single system . assuming that the multiplicative driving noise and the local field variable are uncorrelated
, they inferred the scaling behaviour of the variance of an effective multiplicative noise with @xmath0 , and of the critical value @xmath1 of the control parameter .
they also predicted a crossover from a critical exponent @xmath54 near @xmath1 to the critical exponents for @xmath3 for larger distances to @xmath1 .
note that in @xcite the langevin equation was treated in the ito - sense which leads to a shift of the critical control parameter compared to the same equation in the stratonovich - sense .
hasegawa considered finite systems with additive and multiplicative noise using his augmented moment method which is applicable for small noise strength .
he emphasized that multiplicative noise and the local field variable are not uncorrelated in contrast to the assumption in @xcite and demonstrated some consequences of such a simplification .
our approach , though similar in spirit to @xcite , is controllable , valid in leading order for strong coupling @xmath49 , and provides explicit analytic results which are confirmed by independent numerical simulations .
it may serve as a starting point to calculate next order corrections @xmath60 .
the paper is organized as follows . in the next section
we consider two harmonically coupled stratonovich models and show that for strong coupling the center of mass coordinate @xmath61 is the relevant degree of freedom .
the mean value of @xmath61 shows a critical behavior which is analytically characterized .
section [ arbn ] deals with a class of @xmath0 globally coupled systems of general kind . for strong coupling
we compute analytically the stationary probability distribution @xmath52 after eliminating the relative coordinates .
further , we determine the transition manifold in the parameter space where @xmath52 undergoes a transition from a delta - distribution to a spatially extended solution . in sec .
[ nsm ] we specialize to the case of @xmath0 globally coupled stratonovich models and determine the critical behaviour of the order parameter and of higher moments of @xmath61 for strong coupling .
conclusions are drawn and a summary is given in section [ con ] .
in [ weaknorm ] we introduce the concept of weak normalization for the case that a spatially extended solution of the stationary fpe can not be normalized in the naive sense .
[ stratito ] shows that the langevin approach both in stratonovich- and in ito - interpretation leads to the same results as the fokker - planck approach used in the main part of the paper .
we consider a pair of particles with coordinates @xmath62 and @xmath63 in a biquadratic potential which are coupled harmonically with positive coupling strength @xmath49 and each subjected to independent gaussian white noise of strength @xmath6 .
the system of langevin equations reads @xmath64dt + \sigma x_i\circ dw_i(t ) , \;\ ; i=1,2 \ ; , \label{2le}\end{aligned}\ ] ] where @xmath65 denotes independent wiener processes with @xmath66 .
in contrast to eq .
( [ 1le ] ) no exact solution of system ( [ 2le ] ) is known .
the joint probability density @xmath67 is governed by the fpe @xmath68 , \label{2fpe}\end{aligned}\ ] ] where , adopting the notation of @xcite , @xmath69 denote drift and diffusion coefficients , respectively .
one can show that the system ( [ 2fpe ] ) exhibits no detailed balance .
hence , there is no easy way to obtain analytically the stationary solution @xmath70 . for strong coupling , however ,
a systematic analytical approach is possible . with increasing coupling strength
the particles become tightly glued together and move as a single entity .
therefore it appears natural to introduce center of mass and relative coordinates .
simulations of eq .
( [ 2le ] ) show that indeed the stationary distribution of the relative coordinate @xmath71 becomes very sharp for large values of @xmath49 , cf .
[ fig : r_r - hist_d ] .
( left ) and relative coordinates @xmath72 ( right ) for two coupled systems .
the distribution of relative coordinates is symmetric with respect to zero and becomes very sharp with increasing strength of the coupling @xmath49 .
the symbols show histograms from @xmath73 realizations obtained by solving eq .
( [ 2le ] ) with a stochastic runge - kutta scheme @xcite .
parameters are @xmath74 ( squares ) , @xmath75 ( circles ) , and @xmath76 ( triangles ) ; @xmath77 and @xmath78 .
initial values were all in the positive sector .
entries of several bins are omitted to avoid overloading ; the lines are guides to the eye . ]
the stationary distribution of the center of mass @xmath79 shows a behaviour which is similar to the distribution of a single stratonovich model . for large values of @xmath5
we have a monomodal distribution which vanishes at the boundaries of the support , cf .
[ fig : r_r - hist_d ] . for small values of @xmath5
the distribution @xmath79 diverges as @xmath80 in a normalizable way , cf .
[ fig : r_r - hist_a ] . for even smaller values of @xmath5
all trajectories @xmath81 approach zero and the distributions of both @xmath82 and @xmath61 are @xmath83-distributions .
accordingly , the mean value @xmath84 undergoes a continuous transition at a critical value of @xmath5 . in the following we analytically calculate @xmath79 and @xmath85 and its scaling characteristics in the strong coupling limit , @xmath50 .
( left ) and relative coordinates @xmath72 ( right ) for two coupled systems in a semilogarithmic plot . for sufficiently small control parameter
, @xmath52 diverges for @xmath80 in a normalizable way .
@xmath72 becomes essentially sharper compared to fig .
[ fig : r_r - hist_d ] . parameters are @xmath86 , @xmath87 , @xmath78 ( triangles ) and @xmath88 ( circles ) .
the symbols show data from @xmath89 realizations generated by a stochastic runge - kutta algorithm @xcite .
initial values were chosen in the positive sector .
the lines are guides to the eye . ]
we introduce the center of mass coordinate @xmath90 and the relative coordinate @xmath91 by @xmath92 with the inverse transformation @xmath93 with @xmath94 the langevin equations ( [ 2le ] ) then transform to @xmath95dt + \frac{\sigma}{\sqrt{2}}\left(r\circ d\widetilde w_1(t)\label{2r}\ ! + \!r\circ d\widetilde w_2(t)\right),\end{aligned}\ ] ] where the transformed wiener processes @xmath96 are defined as @xmath97 with @xmath98 .
the fpe associated to ( [ 2r],[2r ] ) governing the probability density of center of mass and relative coordinates @xmath99 reads @xmath100 where the fokker - planck operators are @xmath101 , \\ { \cal l}_r&=&-\partial_r \big[\big(a-2d+\frac{\sigma^2}{2}\big)r - r^3 - 3rr^2\!-\!\frac{\sigma^2}{4}\partial_r\left(r^2+r^2\right ) \big],\\ { \cal l}_{r r}&= & \;\sigma^2 \partial^2_{rr } \ ; rr\;.\end{aligned}\ ] ] note that only @xmath102 depends on @xmath49 . in the strong coupling limit @xmath103 the relative
coordinate vanishes , @xmath104 , on a very fast time scale of the order @xmath105 , cf .
( [ 2r ] ) .
hence , the stationary probability density factorizes to @xmath106 with a dirac distribution for the relative coordinate . in this case
there is no flow related to the relative coordinate @xmath82 , i.e. @xmath107 , since for any suitable function @xmath108 @xmath109 \equiv 0\;.\end{aligned}\ ] ] integrating eq .
( [ 2fpe_r ] ) with respect to @xmath82 yields in the stationary case @xmath110\,.\label{2dfp_red}\end{aligned}\ ] ] similarly as for the single stratonovich model , there is always a weak solution @xmath111 . for initial values @xmath112 ( or @xmath113 ) the spatially extended solution of eq .
( [ 2dfp_red ] ) lives on the support @xmath114 ( or on @xmath115 $ ] ) and can be normalized provided @xmath116 . for @xmath117 the weakly normalized version of the spatially extended solution converges to @xmath118 .
thus , we have @xmath119 there is a strong ergodicity breaking when @xmath5 crosses @xmath120 .
the mean value @xmath121 calculated with ( [ ps2 ] ) is @xmath122 and scales like @xmath123 with @xmath54 as @xmath124 .
in the following we demonstrate that the strategy sketched above can be generalized for a class of @xmath0 coupled systems .
we consider @xmath125dt + g(x_i)\circ dw_i(t),\label{ndle}\end{aligned}\ ] ] with @xmath126 and where @xmath127 and @xmath128 are smooth ( with no singularities ) and twice differentiable chosen such that the stochastic process @xmath129 has natural boundaries at infinity @xcite . both @xmath127 and @xmath128 may depend on a @xmath130-dimensional set of control parameters @xmath131 .
@xmath132 is the coupling strength of the harmonic attraction .
note that we have absorbed a factor @xmath133 , the strength of the noise , in the function @xmath128 .
the fpe for the joint probability density @xmath134 , @xmath135 , associated to ( [ ndle ] ) reads @xmath136.\end{aligned}\ ] ] using the shorthands @xmath137 , @xmath138 , and @xmath139 , drift coefficient and diffusion matrix are given by @xmath140 it is advantageous to introduce center of mass and relative coordinates @xmath141 , @xmath142 , by the linear transformations @xmath143 the inverse transformation is given by @xmath144 observing the rules for linear transformations we have @xmath145 drift and diffusion coefficients in the new coordinates are given by , cf .
also @xcite , @xmath146 where @xmath147 and @xmath148 stand for the new coordinates @xmath61 , @xmath149 , and @xmath150 , respectively .
again , the fpe determining @xmath151 has the form @xmath152 , with @xmath153 where @xmath154 explicitly , the new drift and diffusion coefficients are @xmath155 all arguments in @xmath156 , @xmath157 , and @xmath158 have to be expressed by @xmath159 , see eqs .
( [ inversex1],[inversexk ] ) .
note that only @xmath160 depends on the coupling strength @xmath49 explicitly . for large times
the probability density @xmath151 converges to a stationary probability density , cf .
@xcite , determined by @xmath161 . for @xmath103 this enforces @xmath162=0\;,\ ] ] which has a weak solution @xmath163 in the strong coupling limit all fluctuations of the relative coordinates vanish .
the system is concentrated on the center of mass and moves stochastically as a whole , combined particle .
the probability density of the center of mass @xmath52 can be determined by integrating @xmath164 over all relative coordinates . performing this integration we obtain from the stationary fpe @xmath165
provided that the boundary terms associated with the relative coordinates vanish . in the strong coupling limit
we have @xmath166 , ( [ fpint ] ) holds in any case and leads to @xmath167 where @xmath168 is given by ( [ fpinrr ] ) . from ( [ dr ] ) and ( [ drr ] ) we infer drift and diffusion for @xmath169 as @xmath170 the spatially extended strong solution of ( [ fpr ] ) is given by @xmath171 provided that the normalization constant @xmath26 is finite . whether or not this is the case depends on the scaling behaviour of the functions @xmath172 and @xmath173 near a common zero @xmath174 which , if existing , will build a boundary of the support .
this is explained in detail in the next subsection .
equation ( [ diffcoeffn ] ) shows that in the strong coupling limit the diffusion coefficient @xmath175 scales like @xmath176 , cf . also @xcite . for the infinite system and finite noise strength
@xmath6 the stationary probability density of the center of mass @xmath52 is a dirac measure located at one of the attractive zeros of the drift coefficient ( [ driftcoeffn ] ) , depending on the initial conditions .
for @xmath50 all particles are strongly correlated .
the variance of the coordinate @xmath177 of an arbitrary system @xmath178 calculated with @xmath179 is @xmath180 due to the strong correlations , the variance of the center of mass scales like @xmath181 in contrast to the case of uncorrelated systems where the central limit theorem predicts a scaling like @xmath182 .
there will be a strong ergodicity breaking if the state space decomposes into different regions with the property that one region will not be accessible if we start in a different one @xcite . for multiplicative noise , zeros of the stochastic flow separate the state space into mutually non - accessible regions .
if we place the system initially on such a zero , i.e. on a fixed point of the stochastic dynamics , it will stay there forever .
accordingly , the fpe has a weak solution , a @xmath83-distribution living on that fixed point .
if any trajectory in the neighborhood asymptotically reaches the fixed point ( the fixed point is absorbing ) , there will be no spatially extended probability density in this neighborhood .
the spatially extended stationary solution of the fpe can not be normalized in the naive sense .
the weak normalization procedure leads to the weak solution . if trajectories can not reach the fixed point , the stationary solution of the fpe will be normalizable and we will have a spatially extended probability density living on the support bounded by the fixed point
this properties can be exploited to determine the transition manifold in the parameter space .
we suppose that @xmath183 and @xmath184 near a common zero @xmath174 have the following scaling behaviour @xmath185 where @xmath186 .
near @xmath174 we have for the stationary solution ( [ ps ] ) of the reduced fpe ( [ fpr ] ) @xmath187 for @xmath188 the integral in ( [ p(eps ) ] ) gives a contribution @xmath189 at the upper boundary which vanishes for @xmath190 so that in leading order latexmath:[$p_s(r_0+\varepsilon ) \propto negative only for @xmath58 . in this case , if @xmath192 the singularity of @xmath193 at @xmath194 is not normalizable and we have only a weak stationary solution @xmath195 .
note that for @xmath196 coupled systems of this kind the singularity of @xmath193 does not occur .
for @xmath197 the integral gives a logarithmic contribution @xmath198 which leads to @xmath199 if the exponent in ( [ p(eps)2 ] ) is smaller than @xmath200 the density @xmath52 will diverge for @xmath201 and will not be normalizable in a naive way .
the weak normalization procedure leads to a dirac measure located at @xmath174 .
if the exponent is larger than @xmath200 the density @xmath52 will be normalizable and we will have a spatially extended probability density .
the transition manifold @xmath202 in the control parameter space @xmath203 is determined by the condition that the exponent in ( [ p(eps)2 ] ) is @xmath200 , @xmath204 for @xmath205 the integral in ( [ p(eps ) ] ) diverges like @xmath206 as @xmath207 .
accordingly , @xmath208 for @xmath209 and @xmath210 for @xmath211 . in the first case
@xmath52 is normalizable and we have a spatially extended stationary probability density . in the latter case
the weak normalization procedure yields a dirac measure at @xmath174 .
a change in the sign of @xmath212 induced by tuning a control parameter is associated with a change of the stability of the fixed point @xmath174 of the _ deterministic _ flow @xmath172 and leads to a significant alteration of the ergodic properties . in a vicinity of @xmath174 the behaviour of the stochastic system
is dominated by the deterministic flow .
the transition manifold @xmath213 does not depend on the system size and the amplitude @xmath214 in contrast to ( [ mani ] ) .
if the system lives on the @xmath215 dimensional transition manifold ( [ a_fzero ] ) , that is @xmath216 , the scaling of the deterministic flow is not described by ( [ fscaling ] ) but by @xmath217 with @xmath218 .
the systems behaviour , now depending on @xmath219 and @xmath0 , could be classified in more detail repeating the above procedure .
now we return to our specific example and consider @xmath0 globally coupled stratonovich models in the strong coupling limit . for drift and noise function
we have @xmath220 the common zero of @xmath127 and @xmath128 is @xmath221 with @xmath222 , and @xmath223 , @xmath224 .
inserting this in ( [ mani ] ) we obtain an explicit representation of the transition curve in the 2-dimensional parameter space , @xmath225 given the noise strength @xmath6 we have @xmath226 which reproduces the results for @xmath58 , for @xmath227 ( see above ) , and for @xmath228 .
transition point @xmath2 vs system size @xmath0 for @xmath229 .
simulation results for @xmath230 obtained by maximizing the linear correlation coefficient @xcite ( gray bullets ) and by short time relaxation @xcite ( black bullets ) are both in good agreement with the asymptotic result ( [ ac ] ) for @xmath231 ( solid line).,scaledwidth=50.0% ] figure [ ac_vs_n ] compares results from simulation and the asymptotic result ( [ ac ] ) for @xmath1 and illustrates the finite size scaling @xmath232 for strong coupling @xmath233 . for initial values @xmath234 ( or @xmath235 ) @xmath236 the stationary distribution for the center of mass ( [ ps ] ) lives on @xmath30 or @xmath29 , respectively , and is given by @xmath237 for initial values @xmath238 we have @xmath239 for all values of @xmath5 .
similar to the single stratonovich model , there is a qualitative change in the shape of the spatially extended probability density .
the maximum of @xmath52 undergoes a bifurcation at @xmath240 .
figure [ pvsr ] compares for different system sizes the asymptotic result ( [ pssm ] ) with histograms obtained by simulations for large @xmath49 .
probability density of the center of mass obtained by simulation for @xmath87 ( symbols ) compared with analytic results ( lines ) for @xmath231 given by eq .
( [ pssm ] ) . on the left
we show results for different system sizes @xmath227 ( circles ) , @xmath241 ( squares ) , and @xmath242 ( triangles ) for @xmath243 .
the histograms are obtained from @xmath244 realizations generated by a stochastic euler method @xcite . on the right
we show only results for @xmath245 for @xmath246 ; here @xmath247 realizations were generated by a stochastic runge - kutta algorithm @xcite . @xmath248 . ] for @xmath249 the @xmath250th moment of the center of mass can be evaluated as @xmath251 keeping @xmath0 finite , the first moment scales as @xmath252 like @xmath253 since @xmath254 as @xmath255 @xcite .
note that also the higher moments @xmath256 scale linear with @xmath257 . keeping a finite distance to @xmath1 we obtain for @xmath3 , observing @xmath258 as @xmath259 @xcite , @xmath260 which reproduces the result in @xcite for @xmath261 .
higher moments of order @xmath250 scale with @xmath262 .
we define the crossover value @xmath56 by @xmath263 . for @xmath264
we have a linear scaling as for @xmath58 whereas for @xmath59 a square root behaviour as for @xmath3 is observed , cf . fig .
[ crossover ] .
our results are analytically derived for the strong coupling limit in a controllable approach .
we note that both the critical and the crossover value of the control parameter are in accordance with the values proposed on different grounds in @xcite for weak and intermediate noise , provided the shift due to the ito interpretation used there is taken into account .
crossover in the scaling behaviour of @xmath85 as a function of @xmath265 as predicted by eq .
( [ r^n ] ) for @xmath58 ( dash - dotted line ) , @xmath266 ( dashed line ) , and @xmath76 ( solid line ) ; @xmath78 .
the symbols show averages over @xmath267 realizations generated by a stochastic runge - kutta scheme @xcite for @xmath227 ( circles ) and @xmath76 ( squares ) ; @xmath78 and @xmath87 .
the dotted straight lines have slope @xmath268 ( left ) and slope @xmath269 ( right ) , respectively .
the arrows indicate the crossover points @xmath270 .
, scaledwidth=75.0% ]
in this paper we have determined the characteristics of a continuous nonequilibrium phase transition in a finite array of globally coupled stratonovich models in the limit of strong coupling @xmath50 . in this limit
there is a clear separation of the time scales governing the evolution of the center of mass coordinate and the relative coordinates : the characteristic time of the relative coordinate scales as @xmath271 and thus becomes short in the strong coupling limit .
the slow center of mass coordinate enslaves the fast relative coordinates , its mean value serves as order parameter .
this allows a controllable and consistent treatment both in the fokker - planck and the langevin description which is corroborated by numerical simulations .
the reduction of a high - dimensional problem to a problem of low dimension is of course inspired by generalizations of slaving and adiabatic elimination techniques and the concept of center manifolds to stochastic systems developed in a different context @xcite , cf .
also @xcite .
we have analytically determined both the critical value of the control parameter @xmath1 and the scaling behaviour of the order parameter and of higher moments . with increasing distance from @xmath1 a crossover from linear to square root behaviour is found . for @xmath3
the known scaling behaviour is reproduced .
the formal results , i.e. , the computation of the stationary distribution of the center of mass coordinate ( up to a quadrature ) and the determination of the transition manifold are given for a general class of systems .
our approach may serve as a starting point to calculate next order corrections in @xmath105 . in a multiscale analysis
we have to take into account that for finite but large @xmath49 the distribution of the relative coordinates is , though very sharp , of finite width .
the observation that a solution of the stationary fokker - planck equation which is not normalizable in a naive way converges to the weak solution if weakly normalized is certainly of interest in a broader context .
the fpe for multiplicative noise has two types of stationary solutions : weak solutions , i.e. dirac - distributions living on the zeros of the stochastic flow and spatially extended strong solutions which live on a support which is bounded by zeros of the stochastic flow or by natural boundaries at infinity . under certain conditions
the spatially extended solution may diverge at a zero of the stochastic flow so strongly that it is not normalizable and therefore can not be considered as a probability density . here
we introduce the concept of weak normalization and show that in the latter case the weakly normalized solution converges to the dirac distribution living on that zero .
we assume that the unnormalized solution @xmath272 lives on @xmath273 where @xmath274 is a zero of the stochastic flow and scales for @xmath275 as @xmath276 the normalization integral diverges at the lower boundary and scales like @xmath277 introducing a test function @xmath278 which can be expanded near @xmath274 as @xmath279 we have as @xmath280 @xmath281 now we obtain for the hereby defined weakly normalized probability density @xmath282 @xmath283 which implies that @xmath284 .
in sections [ neq2 ] and [ arbn ] we used the fokker - planck approach in the center of mass and relative coordinates to calculate @xmath1 for @xmath285 . in this limit
the relative coordinates @xmath286 , and it is easy to calculate the reduced stationary probability density of the center of mass coordinate @xmath52 .
we determined @xmath1 such that @xmath52 is a dirac measure at @xmath287 for @xmath288 and it is spatially extended for @xmath289 .
the same result can be obtained in the langevin approach , both in stratonovich and ito - interpretation as explained in the following for the special case @xmath227 .
the generalization to @xmath290 is straightforward .
we exploit that for large @xmath49 the characteristic time scale of the relative coordinate @xmath91 is @xmath105 so that @xmath91 becomes very fast .
then the ( slow ) center of mass coordinate @xmath90 feels only the average of the fast process @xmath91 and it is justified to replace in the stratonovich - langevin equation ( [ 2r ] ) for @xmath61 the terms associated with @xmath82 by their average , @xmath291 since for @xmath50 , @xmath292 we have @xmath293 . however , the second average @xmath294 is not zero as one could naively expect . with the help of the furutsu
novikov theorem @xcite we obtain @xmath295 note that the averages here are with respect to realizations of @xmath296 .
we now observe that the resulting equation for @xmath61 does not depend on @xmath296 , therefore @xmath297 , and obtain @xmath298dt + \frac{\sigma}{\sqrt{2 } } r\circ d \widetilde w_1(t ) \,,\end{aligned}\ ] ] from which the threshold @xmath299 follows .
the system ( [ 2r],[2r ] ) in stratonovich sense can be written in a compact form as @xmath300 , where @xmath301 .
the equivalent ito system is @xmath302 , where the drift term is modified by the ito shift ; @xmath303 is the shorthand of a jacobian , cf .
e.g. @xcite . for our system
we have @xmath304 and @xmath305 .
the ito shift amounts to @xmath306 so that the equivalent ito version of ( [ 2r ] ) reads @xmath307\!dt \!+\!\frac{\sigma}{\sqrt{2}}\ ! \left(r \,d
\widetilde w_1(t)\!+\!r \,d\widetilde w_2(t)\right).\end{aligned}\ ] ] again , @xmath61 feels only the average of the terms associated with the fast process
@xmath82 , we have @xmath308 but now also the second average vanishes since in the ito calculus @xmath309 which results in @xmath310dt + \frac{\sigma}{\sqrt{2 } } r \,d \widetilde w_1(t).\end{aligned}\ ] ] this is indeed the ito eqivalent to eq .
( [ 2rs ] ) which can be seen observing that in the single variable case the ito shift is simply @xmath311 . for arbitrary @xmath0
the same procedure leads to @xmath312 as obtained above .
99 we used the stochastic runge - kutta method in a version proposed in @xcite which is an explicit algorithm for stochastic ordinary differential equations in the stratonovich sense that converges with weak order @xmath268 , cf .
the step size was always @xmath313 .
trajectories of @xmath81 are numerically determined by an adapted heun method @xcite with time step @xmath313 for different control parameters @xmath314 .
values of @xmath5 for which @xmath81 comes very close to zero are discarded to exclude that @xmath315 .
after the transient period the steady state temporal averages are build . supposing a power law @xmath316 the data are displayed for different test values of @xmath2 in a double logarithmic plot . after a visual control we determine @xmath2 by maximizing the linear correlation coefficient as proposed in @xcite , @xmath317 is the slope of the corresponding line .
the transition points @xmath2 are determined observing numerically the behaviour of @xmath90 for an initial state where all @xmath318 with @xmath319 for a short period of time such that the distribution of the @xmath81 becomes not too broad .
then , the evolution is essentially governed by the linear part of the langevin - equation of @xmath90 , cf . [ stratito ]
. trajectories of @xmath81 are generated by a stochastic runge - kutta method with step @xmath313 , @xcite .
generically , for @xmath288 the trajectory of @xmath90 will relax towards zero , whereas for @xmath289 it will increase provided @xmath174 was smaller than the saturation value . finally , the estimates of @xmath1 obtained from different trajectories are averaged .
a similar procedure has been exploited in @xcite for @xmath3 ; cf .
also @xcite . | a finite array of @xmath0 globally coupled stratonovich models exhibits a continuous nonequilibrium phase transition . in the limit of strong coupling
there is a clear separation of time scales of center of mass and relative coordinates .
the latter relax very fast to zero and the array behaves as a single entity described by the center of mass coordinate .
we compute analytically the stationary probability and the moments of the center of mass coordinate .
the scaling behaviour of the moments near the critical value of the control parameter @xmath1 is determined .
we identify a crossover from linear to square root scaling with increasing distance from @xmath2 .
the crossover point approaches @xmath2 in the limit @xmath3 which reproduces previous results for infinite arrays .
the results are obtained in both the fokker - planck and the langevin approach and are corroborated by numerical simulations . for a general class of models
we show that the transition manifold in the parameter space depends on @xmath0 and is determined by the scaling behaviour near a fixed point of the stochastic flow . |
one of the most productive and perspective ideas in nuclear many - body theory is the concept of independent single - particle motion in the mutually generated mean field .
it explains many nuclear observables qualitatively and provides a basis for quantative descriptions of nuclei within various versions of the shell model . over the last decade
, the experimental study of single - particle motion has shifted towards nuclei near the edge of stability with the emphasis on the occupation probabilities of single - particle orbits as indicators of the evolution of nuclear shell structure near the drip line .
these occupancies are usually determined by comparing the measured cross sections of one - nucleon transfer , breakup , knockout and coulomb dissociation reactions , to those predicted theoretically . in these calculations ,
the single particle motion is represented by the overlap integral between the many - body wave functions of neighbouring mass @xmath3 and @xmath4 nuclei .
most theoretical analyses assume these overlaps can be calculated using potential models that employ standard woods - saxon potentials . with these potentials , at radii outside the binding interaction ,
the overlap integrals achieve their asymptotic form given by an exponential decrease with a decay constant @xmath5 determined by the nucleon separation energy @xmath6 , with @xmath7 is the @xmath8+@xmath9 reduced mass .
this region gives important contribution to the amplitudes of one - nucleon removal reactions and is crucial to predict direct nucleon capture reactions relevant to nuclear astrophysics .
it has been shown recently in ref .
@xcite that there exist a class of nuclei for which one - nucleon overlaps can reach their asymptotics much later than usual .
this class includes the nuclei at the limits of nuclear stability with very similar one- and two - nucleon separation energies , @xmath6 and @xmath10 , both of which are significantly smaller than for stable nuclei .
such nuclei , for example @xmath0be , @xmath1c , @xmath2ne , can be considered as cluster systems of a core and two valence nucleons .
the correlations of the two nucleons outside of the nuclear core can lead to non - standard behaviour of the overlap because even when one nucleon is removed far from the center - of - mass of the residual mass @xmath4 nucleus , it will still be affected by its interaction with the remaining loosely - bound nucleon in @xmath4 .
the possibility for such an effect follows from the feynman diagram approach in which the asymptotics of the overlap integrals for many - body systems are represented by a sum of different feynman diagrams .
it has been noticed by blokhintsev @xcite that , although in most cases the standard point diagram gives the exponential decrease with the smallest decay constant @xmath11 , sometimes other diagrams can give a much slower decrease than that governed by the differences in initial and final binding energies .
the feynman diagram technique , applied in @xcite to one - neutron overlaps for core+2n cluster systems , suggests that although the standard term , @xmath12 , dominates at very large @xmath13 , the contribution @xmath14 from the generalised triangle diagram , corresponding to the @xmath15 virtual process , is determined by the decay constant @xmath16 that can be larger than @xmath11 by only 40@xmath17 .
it means that if the strength of the generalised triangle diagram is large then its contribution can be noticeable for range of @xmath13 where the standard behaviour @xmath12 is traditionally assumed to be achieved .
the abnormal asymptotic behaviour was modelled in @xcite for the @xmath18be@xmath19be@xmath20 overlap by adding a long - range potential to a local potential model .
the resulting overlap was then used to predict longitudinal momentum distributions for one - neutron knockout from @xmath0be .
comparison of the latter to available experimental data suggests such abnormalities would take place for @xmath21be distances between 5 and 10 fm .
however , the contribution from the generalised triangle diagram can be calculated reliably only for @xmath22 , which for @xmath0be is @xmath23 fm , therefore , to understand the overlap behaviour at @xmath24 fm exact dynamical three - body calculations should be carried out . here
, we calculate one - nucleon overlaps using a three - body core+2n model and compare them to the predictions of the standard potential model ( sec .
we discuss effects of non - standard overlap behaviour on nuclear reactions ( sec . 3 ) and challenge they pose for both theory and experiment ( sec.4 ) .
effective local @xmath21be potentials for pseudo @xmath0be fitted by a sum of two woods - saxon potentials.,title="fig : " ] effective local @xmath21be potentials for pseudo @xmath0be fitted by a sum of two woods - saxon potentials.,title="fig : " ] effective local @xmath21be potentials for pseudo @xmath0be fitted by a sum of two woods - saxon potentials.,title="fig : " ] effective local @xmath21be potentials for pseudo @xmath0be fitted by a sum of two woods - saxon potentials.,title="fig : " ]
we have performed three - body calculations for @xmath0be ( @xmath25be+n+n ) , @xmath1c ( @xmath26be+p+p ) and @xmath2ne ( @xmath27o+p+p ) assuming no structure for the @xmath25be , @xmath26be and @xmath27o cores but taking antisymmetrization into account using the pauli projection technique @xcite .
the three - body schrdinger equation has been solved by expanding the wave functions onto a hyperspherical harmonics basis .
the overlap integrals @xmath18be@xmath19be@xmath20 , @xmath28b@xmath29c@xmath20 and @xmath30f@xmath31ne@xmath20 have been obtained by overlapping the three - body and two - body wave functions .
the @xmath9-core woods - saxon potentials have been fitted to chosen separation energies and the gpt potential @xcite has been used for the @xmath32 interaction .
the coulomb @xmath33-core and @xmath34 potentials have been also included for @xmath35c and @xmath2ne .
* pseudo @xmath0be*. first of all the convergence of the overlap in the asymptotic region has been tested for a simplified case of @xmath0be in which the deformation and excitation of the @xmath25be core have been neglected and the @xmath21be interaction was zero in all partial waves except for @xmath36 .
this has enabled us to include terms up to @xmath37 in the hyperspherical expansion and to obtain converged values for @xmath18be@xmath19be@xmath20 with high accuracy up to 30 fm .
the @xmath38be potential was fitted either to the experimental value of @xmath39be ) , 496 kev , or to a much smaller value of 102 kev to check whether the effect from the @xmath40 correlations increases with weaker binding . in this particular case ,
the @xmath40 potential was represented by just one gaussian that gives a scattering length of @xmath41 fm .
the three - body energies obtained for @xmath0be are shown in table 1 .
the overlap integrals @xmath18be@xmath19be@xmath20 , renormalized to unity , are plotted as ratios to their asymptotics in fig.1a where they are compared to those calculated in a potential model with standard geometry @xmath42 = 1.25 fm and @xmath43 = 0.65 fm .
also , the r.m.s .
radii of these overlaps are given in table 1 in comparison with standard values .
the abnormal behaviour is clearly seen for both three - body calculations and , as expected , the abnormalities are stronger for a smaller @xmath39be ) .
fig.2 shows effective local potentials ( open circles ) obtained by inversion of the two - body schrdinger equation .
they can be fitted by a sum of two woods - saxon potentials one of which has unusually large diffuseness ( see table 2 ) .
* realistic @xmath0be*. to take into account the deformation and the excitation of the @xmath25be core , we have repeated the calculations from @xcite for the case exc1 .
the model space have been increased up to @xmath44 = 34 which has provided a reasonable stability for the overlap @xmath18be@xmath19be@xmath20 at @xmath45 fm .
the abnormal surface behaviour in @xmath18be@xmath19be@xmath20 is seen for all three states of the @xmath46be residue ( see fig .
1b ) and the effective local @xmath47be potentials , the parameters of which are shown in table 2 , have long - range terms although their diffuseness are smaller than in the pseudo @xmath0be case .
the abnormalities arise due to the correlations between the two neutrons which are visualised in the plot of the wave function densities in fig .
3 . when the c.m .
of two neutrons is far away from the @xmath25be core , the probability that these neutrons are close to each other is high .
probability densities of the wave functions obtained in three - body calculations for realistic @xmath0be , @xmath2ne and pseudo @xmath1c . ]
lllllll nucleus @xmath3 & nucleus @xmath3@xmath48@xmath49 & @xmath50@xmath48@xmath51 & @xmath6 & @xmath10 & @xmath52 & @xmath53 + pseudo @xmath0be & @xmath46be(@xmath54 ) & 0.102 & 0.56 & 0.656 & 7.650 & 6.983 + pseudo @xmath0be & @xmath46be(@xmath54 ) & 0.496 & 1.25 & 1.76 & 5.807 & 5.423 + real @xmath0be & @xmath46be(@xmath54 ) & 0.500 & 3.141 & 3.641 & 4.412 & 4.298 + & @xmath46be(@xmath55 ) & 0.180 & 3.461 & 3.641 & 3.588 & 3.473 + & @xmath46be(@xmath56 ) & @xmath481.09 & 4.731 & 3.641 & 3.738 & 3.244 + pseudo @xmath35c & @xmath57b & 0.149 & 4.526 & 4.675 & 3.175 & 3.016 + @xmath2ne(@xmath58 ) & @xmath59f(@xmath56 ) & 0.603 & 2.974 & 3.577 & 3.602 & 3.337 + & @xmath59f(@xmath54 ) & 0.176 & 3.401 & 3.577 & 4.068 & 3.602 + @xmath2ne(@xmath60 ) & @xmath59f(@xmath56 ) & 0.603 & 2.974 & 1.141 & 3.511 & 3.426 + & @xmath59f(@xmath54 ) & 0.176 & 3.401 & 1.141 & 4.486 & 3.947 + * pseudo @xmath1c*.
our attempts to use the proper spin @xmath61for the @xmath62be core have been successful only for small model spaces that can not provide convergence for the overlap in the region of interest .
therefore , we have used @xmath63 for @xmath62be and neglected the @xmath33-@xmath62be spin - orbit interaction .
this overbinds @xmath1c but enables us to increase the model space up to @xmath44 = 50 , thus providing high accuracy for @xmath28b@xmath29c@xmath20 up to 20 fm .
the overlap converges to its asymptotic form slower than in the standard case but not as slow as for @xmath0be .
the two protons stay close to each other when both are far away from the @xmath26be core but the weight of this configuration is smaller than in @xmath0be ( see fig.3 ) .
* @xmath2ne*. the @xmath2ne wave functions have been calculated for the ground and first excited @xmath64 states in a model space with @xmath44 = 50 .
the @xmath34 correlations outside the nuclear interior are clearly seen in the @xmath2ne wave function and the overlaps differ from the standard ones in the surface region .
however , this difference is much smaller than for @xmath0be , despite the long - range character of the coulomb force .
the three - body dynamics results mainly in an increase of the overlap r.m.s .
radius , which is stronger in the @xmath65-wave where the centrifugal barrier is absent .
lllllllllll nucleus @xmath3 & @xmath3@xmath48@xmath49 & @xmath66 & @xmath42 & @xmath67 & @xmath68 & @xmath69 & @xmath70 & @xmath71 & @xmath72 & @xmath73 + pseudo @xmath0be & @xmath46be(@xmath54 ) & 53.5 & 1.20 & 0.70 & 4.8 & 0.989 & 1.85 + pseudo @xmath0be & @xmath46be(@xmath54 ) & 55.9 & 1.07 & 0.82 & 1.5 & 1.124 & 3.1 + real @xmath0be & @xmath46be(@xmath54 ) & 50.0 & 1.33 & 0.75 & 6.9 & 1.17 & 1.4 + & @xmath46be(@xmath55 ) & 50.0 & 0.99 & 0.79 & 9.8 & 1.12 & 1.55 & 8.85 & 1.08 & 0.80 + pseudo @xmath35c & @xmath57b & 54.24 & 1.10 & 0.81 & 2.7 & 1.2 & 1.5 & 8.91 & 1.10 & 0.81 +
the knockout cross sections populatingthe final states @xmath46be(@xmath54 ) ( in black ) , @xmath46be(@xmath55 ) ( in red ) and @xmath46be(@xmath56 ) ( in blue ) calculated with two - body @xmath21be wave functions calculated for a range of different geometries of the woods - saxon potentials ( circles ) .
the cross sections corresponding to the standard woods - saxon geometry are shown by open circles while those calculated with a sum of two woods - saxon potentials from table 2 are shown by open squares . ] * nucleon knockout reactions * have recently become a popular tool to determine spectroscopic factors .
we have calculated the cross sections and the parallel momentum distributions for @xmath1be(@xmath0be,@xmath46be)x at 90 mev / nucleon using the @xmath18be@xmath19be@xmath20 overlap obtained in three - body calculations .
the parallel momentum distributions from the three - body @xmath18be@xmath19be@xmath20 overlap agree with those from the woods - saxon potential , provided they have the same r.m.s . radius .
the slow convergence of @xmath18be@xmath19be@xmath20 to its asymptotic form can not be seen in such an experiment .
the momentum distributions depend on the radius of the overlap and thus can be used to determine this if measured with high precision .
the cross sections scale almost linearly with the r.m.s .
radius of the overlap ( see fig.4 ) and , therefore , at these energies the effect from surface abnormalities in the @xmath18be@xmath19be@xmath20 overlap manifests itself only as a change in the overall norm of the calculated cross sections .
this will therefore influence spectroscopic factors deduced from comparisons between calculated and measured cross sections .
* transfer reactions * are sensitive to the surface part of the overlap and thus could be a good tool to study non - standard single - particle motion in the core+2n cluster systems .
a subclass of such reactions , peripheral transfer reactions , determine asymptotic normalization coefficients in the overlap tails thus constraining the overlap r.m.s .
radius . * low - energy direct radiative capture*. proton capture cross sections are often sensitive only to the asymptotic region of the one - nucleon overlap and thus possible abnormal surface behaviour will not influence energy dependence of such reactions .
however , the contribution of the surface region to the neutron capture amplitude can be more important .
we have performed two - body calculations of the @xmath46be(n,@xmath74)@xmath0be cross sections using three different @xmath21be interactions : ( i ) the standard wood - saxon potential ; ( ii ) the effective local potential from table 2 presented by two woods - saxon potentials and ( iii ) a single woods - saxon potential that gives exactly the same r.m.s .
radius for the @xmath21be wave function as the effective local one .
both the absolute values and the energy dependence of the cross sections , shown in fig . 5 ,
are very sensitive to the shape of the overlap . the cross sections for the @xmath46be(n,@xmath74)@xmath0be reaction divided by square root of the @xmath21be energy calculated with three different @xmath21be potentials .
the @xmath75 is shown in logarithmic scale . ]
the exact dynamical three - body calculations have confirmed the possibility of slower radial convergence of the one - nucleon overlaps to their asymptotic form in core+2n cluster systems due to the strong correlations between the two valence nucleons .
this phenomenon has the strongest effect for final core+n systems in relative @xmath76 states and when the coulomb interaction is absent .
the non - standard behaviour has consequences for the determination of spectroscopic factors from nucleon removal reactions and predictions of neutron capture rates on weakly - bound @xmath65-wave nuclei at stellar energies .
therefore , theory must make reliable predictions for non - standard overlaps .
the main challenge for theory is to treat explicitly the three - body dynamics within a many - body object .
however , even when internal structure of the core is neglected , the model space , needed to describe the nucleon motion at large distances in coupled - channel hyperspherical calculations , becomes huge .
convergence accelerating methods should be developed to describe properly the core deformation and excitations within the three - body model .
unsolved is the question of how antisymmetrisation influences non - standard behaviour of the overlap .
the widely used phenomenological shell model does not generate single - particle wave functions at all so no abnormalities can be seen there .
no mean field based theory can reproduce this effect . on the other hand , ab - initio approaches do not yet have sufficient accuracy to study such effects at large distances .
for example , keeping in mind that to see unambiguously abnormalities at @xmath45 fm in three - body calculations the @xmath44 should be at least 30 , means that within models of the no - core shell model type at least 30 additional major shells are needed which do not influence the total binding energy .
this does not appear feasible at the moment .
three - body dynamics can be included explicitly in microscopic cluster models by hand @xcite .
however , at present , such models can cope only with the simplest oscillator internal core structure and very simple nn interactions .
observation of predicted three - body effects in one - nucleon overlaps can be a very difficult task because such effects can be obscured by unsufficient knowledge of reaction mechanisms and uncertainties of other inputs to reaction amplitudes .
however , unambigious experimental confirmation either of their presence or absence is very important for our understanding of nuclear dynamics in near - drip line nuclei .
this work was performed under the uk grants ep / c520521/1 and ep / e036627/1 and in the lawrence livermore national laboratory under doe contract de - ac52 - 07na27344 .
3 timofeyuk n k , blokhintsev l d and tostevin j a 2003 _ phys .
c _ * 68 * 021601(r ) blokhintsev l d 2001 _ bull .
sci . phys_. * 65 * 77 thompson i.j et al 2000 _ phys .
c _ * 61 * 24318 gogny d , pires p and de tourreil r 1970 _ phys .
lett . _ * 32 * 571 nunes f m , thompson i j and tostevin j a 2002 _ nucl .
a _ * 703 * 593 korennov s and descouvemont p 2004 _ nucl .
phys . _ a 740 249 | there exists a class of nuclei that are obtained by adding one nucleon to a loosely - bound nucleon - core system , for example @xmath0be , @xmath1c , @xmath2ne . for
such nuclei , one - nucleon overlap integrals that represent single - particle motion can strongly differ from the standard ones due to the correlations between the two nucleons above the core .
the possible non - standard overlap behaviour should be included in the interpretation of the experimental data derived from one nucleon removal reactions such as knockout , transfer and breakup , as well as the predictions of low - energy nucleon capture that leads to these nuclei .
we investigate the non - standard behaviour within a three - body model and discuss the challenges associated with this problem . |
electromagnetic ( em ) properties of particles are fundamental observables for their internal structure . for hadrons
it is still not possible to reveal this structure analytically from qcd first principles .
two prominent ways to obtain information on baryon em properties are chiral perturbation theories ( @xmath5pt ) and lattice qcd ( lqcd ) which both allow to investigate the quark / pion mass dependence of observables . at present , finite volume lqcd results are mainly compared to infinite volume @xmath5pt ones @xcite where discrepancies in the small pion mass region are seen @xcite .
em finite volume effects on the @xmath5pt side are therefore of interest , however , reveal certain subtleties .
special care has to be taken for the decomposition of the vector - current matrix element in form factors as well as for their lorentz invariance @xcite .
an alternative approach , e.g. to investigate the magnetic moment , is to use the particle s self - energy in an external em field .
this was done in @xcite for pions and nucleons in the non - relativistic heavy baryon @xmath5pt .
one motivation for this work is to derive the corresponding covariant infinite volume @xmath5pt self - energies of the nucleon and @xmath3 that would be needed for a future finite volume study .
another one is to investigate further the non - analytic em field dependence of self - energies @xcite .
we see that beside the found condition for expanding a resonance self - energy in the field strength there exist two more . for this
, we modernize the em background field technique ( bft ) of @xcite and apply it to stable and unstable particles . in 1958 c. m. sommerfield used the bft to obtain the electron s amm up to the fourth order , @xmath6 , correctly for the first time . however , it is the well known three point function method , i.e. the one - photon approximation , that became the preferred method to calculate em moments in field theories .
we will use both techniques to check our formulas .
it turns out that the self - energies of particles , stable as well as unstable , generally depend non - analytically on the magnetic field @xmath0 .
this is e.g. seen for the nucleon and @xmath7-isobar in the covariant @xmath4 b@xmath5pt with @xmath8 loops where the @xmath9 renormalized results with @xmath10 are : @xmath11\,\,\,,\\ \sigma_{\delta^{+}}\left(\mathcal{b}\right ) & = & \frac{m_{\delta}c_{\delta}}{\left(1-\mathcal{b}\right)^{4}}\big[d_{1}+d_{2 } \ln\mu+d_{3}\ln r+d_{4}\ln\left(1-\mathcal{b}\right){\nonumber}\\ & & + \frac{\left(\mathcal{b}-1\right ) d_{5}\mbox{arctanh \ensuremath{\omega}}_{1}}{\sqrt{\lambda^{2}+4\mu^{2}\mathcal{b } } } + \frac{\left(\mathcal{b}-1\right)d_{6}\mbox{arctanh \ensuremath{\omega}}_{2}}{\sqrt{\lambda^{2}+\mathcal{b}\left(\mathcal{b } -2 + 2r^{2}+2\mu^{2}\right)}}+d_{7}\ln\left(r^{2}-\mathcal{b}\right)\big],\end{aligned}\ ] ] with certain polynomials @xmath12 , @xmath13
, @xmath14 and further non - analytic functions @xmath15 and notations given later . one difference of the self - energies is that the square root in @xmath16 can be expanded in a weak magnetic field @xmath0 for all pion masses @xmath17 whereas for the @xmath7-isobar only if the condition @xmath18 between the nucleon , @xmath3 and pion masses is met .
this is a general situation for unstable particles and was discussed in @xcite within a simpler field theory . in the work @xcite
only the one feynman graph leading to the above condition was investigated whereas we derive here the remaining two conditions coming from the two more feynman graphs . in the next section we will give necessary notations for the bft and
will use them in the third section to investigate the self - energies of stable and unstable particles . in the forth and fifth section
we apply the bft to the nucleon and @xmath7-isobar and derive the above expressions .
we consider spin-1/2 and spin-3/2 fields moving in a constant electromagnetic field given by the potential @xmath19 with @xmath20}$ ] as the electromagnetic field strength tensor .
the dirac equation for a particle @xmath21 of mass @xmath22 with the em minimal substitution is : @xmath23\psi_{a}(x)=0\,\,\,,\label{eq : dirac_eq_a}\ ] ] where we take @xmath24 and define the momentum @xmath25 .
this operator is non - commutative with @xmath26=\frac{1}{i}f_{\mu\nu}\,\,\,.\label{eq : commutation_relation}\ ] ] the spin @xmath27 field @xmath28 satisfies the equation @xmath23\psi_{a}^{\mu}(x)=0,\ ] ] with the subsidiary conditions @xmath29 because of the non - commutativity we have to symmetrize occurring expressions and use for the propagators : @xmath30\,\,,\ ] ] where we introduce the notation @xmath31 @xcite . with this
we calculate the particle s self - energy @xmath32 and obtain its anomalous magnetic moment ( amm ) @xmath33 through the linear energy shift : @xmath34 in fig .
[ fig : feynman_graphs ] we list all feynman graphs used in this work for the bft and three point function method .
the upper row shows all types of graphs that appear to the one - loop level in the bft , tadpole graphs not considered .
for better reading we use the notation @xmath35 , i.e. @xmath36 , with which the linear approximation to the self - energy reads : @xmath37 where in the last equation the @xmath38 terms cancel out . the self - energy formulas in this work
omit some , but not all , @xmath39 terms .
in particular non - analytic @xmath0 structures are preserved in the * * bft * * which are not in the one - photon approximation eq .
( [ eq : linear_approximation ] ) .
upper row : feynman graphs contributing to the self - energy in the presence of an external electromagnetic field .
the blue lines indicate which loop - internal particle is affected by the field .
middle and lower row : feynman graphs contributing to the three point function method to obtain the amm .
single solid lines correspond to stable particles ( e.g. the nucleon ) , double solid lines to unstable particles ( e.g. the @xmath7 ) , dashed lines to ( pseudo- ) scalar particles ( e.g. the pion ) and the blue cross to the coupling of the photon.,title="fig : " ] upper row : feynman graphs contributing to the self - energy in the presence of an external electromagnetic field .
the blue lines indicate which loop - internal particle is affected by the field
. middle and lower row : feynman graphs contributing to the three point function method to obtain the amm .
single solid lines correspond to stable particles ( e.g. the nucleon ) , double solid lines to unstable particles ( e.g. the @xmath7 ) , dashed lines to ( pseudo- ) scalar particles ( e.g. the pion ) and the blue cross to the coupling of the photon.,title="fig : " ] upper row : feynman graphs contributing to the self - energy in the presence of an external electromagnetic field .
the blue lines indicate which loop - internal particle is affected by the field
. middle and lower row : feynman graphs contributing to the three point function method to obtain the amm .
single solid lines correspond to stable particles ( e.g. the nucleon ) , double solid lines to unstable particles ( e.g. the @xmath7 ) , dashed lines to ( pseudo- ) scalar particles ( e.g. the pion ) and the blue cross to the coupling of the photon.,title="fig : " ] + upper row : feynman graphs contributing to the self - energy in the presence of an external electromagnetic field .
the blue lines indicate which loop - internal particle is affected by the field .
middle and lower row : feynman graphs contributing to the three point function method to obtain the amm .
single solid lines correspond to stable particles ( e.g. the nucleon ) , double solid lines to unstable particles ( e.g. the @xmath7 ) , dashed lines to ( pseudo- ) scalar particles ( e.g. the pion ) and the blue cross to the coupling of the photon.,title="fig : " ] upper row : feynman graphs contributing to the self - energy in the presence of an external electromagnetic field .
the blue lines indicate which loop - internal particle is affected by the field .
middle and lower row : feynman graphs contributing to the three point function method to obtain the amm .
single solid lines correspond to stable particles ( e.g. the nucleon ) , double solid lines to unstable particles ( e.g. the @xmath7 ) , dashed lines to ( pseudo- ) scalar particles ( e.g. the pion ) and the blue cross to the coupling of the photon.,title="fig : " ] + upper row : feynman graphs contributing to the self - energy in the presence of an external electromagnetic field .
the blue lines indicate which loop - internal particle is affected by the field .
middle and lower row : feynman graphs contributing to the three point function method to obtain the amm .
single solid lines correspond to stable particles ( e.g. the nucleon ) , double solid lines to unstable particles ( e.g. the @xmath7 ) , dashed lines to ( pseudo- ) scalar particles ( e.g. the pion ) and the blue cross to the coupling of the photon.,title="fig : " ] upper row : feynman graphs contributing to the self - energy in the presence of an external electromagnetic field .
the blue lines indicate which loop - internal particle is affected by the field .
middle and lower row : feynman graphs contributing to the three point function method to obtain the amm .
single solid lines correspond to stable particles ( e.g. the nucleon ) , double solid lines to unstable particles ( e.g. the @xmath7 ) , dashed lines to ( pseudo- ) scalar particles ( e.g. the pion ) and the blue cross to the coupling of the photon.,title="fig : " ]
we use the notations of the appendix and consider two spin 1/2 fields @xmath41 , @xmath42 interacting with a scalar field @xmath43 : @xmath44 with the covariant derivative @xmath45 , @xmath46 as the charge of the field with @xmath24 and @xmath47 either @xmath48 or @xmath49 . in the upper row of fig . [
fig : feynman_graphs ] we show all the types of self - energy graphs that can occur for a charged or uncharged external particle @xmath50 .
the corresponding expressions @xmath51 with a loop - internal particle @xmath52 are : @xmath53 } \gamma_{1}\frac{1}{\left[\left(\pi - l\right)^{2}-m^{2}+i\varepsilon\right ] } |\psi_{ex}\rangle\,\,\,,\\ \langle\psi_{ex}|\sigma_{2}\left(\s f\right)|\psi_{ex}\rangle & = & \frac{g^{2}}{i}\langle\psi_{ex}|\int\tilde{dl}\gamma_{2}\frac{1}{\left[\s\pi-\s l - m_{in}+i\varepsilon\right ] } \gamma_{2}\frac{1}{\left[l^{2}-m^{2}+i\varepsilon\right]}|\psi_{ex}\rangle\,\,\,,\\ \langle\psi_{ex}|\sigma_{3}\left(\s f\right)|\psi_{ex}\rangle & = & \frac{g^{2}}{i}\langle\psi_{ex}|\int\tilde{dl}\gamma_{3}\frac{1}{\left[\s\pi-\s l - m_{in}+i\varepsilon\right]}\gamma_{3}\frac{1}{\left[\left(\pi_{+ } + l\right)^{2}-m^{2}+i\varepsilon\right]}|\psi_{ex}\rangle\,\,\,,\end{aligned}\ ] ] with @xmath54 .
for a third - spin projection of @xmath55 we can write these expressions in dimensional regularization as : @xmath56\left[l+\ln\left(z^{2}-\lambda_{i}^{2 } -i\varepsilon\right)+\left(\delta_{i1}+\delta_{i2}\right ) \ln\left(1-\mathcal{b}\right)\right ] + \mathcal{o}\left(\mathcal{b}^{2}\right)\,\,\,,\label{eq : self_energy_ps}\end{aligned}\ ] ] with @xmath57 , @xmath58 given below and @xmath59 for @xmath60 and @xmath61 for @xmath62 .
we see that for @xmath63 these expressions are the same and for @xmath64 several different logarithms occur .
the formula eq .
( [ eq : self_energy_ps ] ) omits some , but not all , @xmath38 terms and the integrated solution for the real and imaginary parts are : @xmath65\,\,\,,\\ \mbox{im}\,\sigma_{i } & = & \frac{g^{2}\pi}{\left(4\pi\right)^{2}}m_{ex}\left(s_{5}\alpha_{i}+r\right)2\lambda_{i}\end{aligned}\ ] ] with @xmath66 and @xmath67
@xmath68 the self - energies obtain imaginary parts if @xmath69 becomes positive together with @xmath70 and/or @xmath71 .
in addition , the non - analytic contributions @xmath72 give constrains on when the self - energies can be expanded for small @xmath73 .
these constrains can generically be written as : @xmath74 for a certain function @xmath75 depending on the type of graph .
one of these constrains , coming from the first graph in fig .
[ fig : feynman_graphs ] , was investigated in @xcite for the situation of @xmath76 . in the following
we investigate the remaining two as well as further applications of eq .
( [ eq : self_energy_ps ] ) .
the first example is the nucleon - pion ( @xmath77 , @xmath78 ) system with pseudo - scalar couplings , @xmath79 . for this
we have @xmath59 , @xmath80 , @xmath81 , @xmath82 and write for the proton and neutron energies : @xmath83 with @xmath84\,\,\,,\end{aligned}\ ] ] and @xmath85 , @xmath86 and @xmath87 .
it is easy to check that we get from this form the same amm as obtained from the usual three point function method : @xmath88 according to this , we can also write for the neutron and iso - vector self - energies : @xmath89 where the difference of eq.([eq : ps_neutron_s12 ] ) to eq.([eq : ps_neutron_s3 ] ) is of order @xmath38
. energy shift of the proton ( left ) and neutron ( right ) with pseudo - scalar coupling for @xmath90 mev .
the upper row corresponds to the shift with @xmath91 and the lower row to @xmath92 .
the linear solid green lines corresponds to the linear approximation eq .
( [ eq : linear_approximation ] ) while the curved solid red lines to eqs .
( [ eq : ps_proton],[eq : ps_neutron_s3 ] ) .
the solid blue line shows the result of eq .
( [ eq : ps_neutron_s12 ] ) .
the dotted lines correspond to the the imaginary parts.,title="fig : " ] energy shift of the proton ( left ) and neutron ( right ) with pseudo - scalar coupling for @xmath90 mev .
the upper row corresponds to the shift with @xmath91 and the lower row to @xmath92 .
the linear solid green lines corresponds to the linear approximation eq .
( [ eq : linear_approximation ] ) while the curved solid red lines to eqs .
( [ eq : ps_proton],[eq : ps_neutron_s3 ] ) .
the solid blue line shows the result of eq .
( [ eq : ps_neutron_s12 ] ) .
the dotted lines correspond to the the imaginary parts.,title="fig : " ] + energy shift of the proton ( left ) and neutron ( right ) with pseudo - scalar coupling for @xmath90 mev .
the upper row corresponds to the shift with @xmath91 and the lower row to @xmath92 .
the linear solid green lines corresponds to the linear approximation eq .
( [ eq : linear_approximation ] ) while the curved solid red lines to eqs .
( [ eq : ps_proton],[eq : ps_neutron_s3 ] ) .
the solid blue line shows the result of eq .
( [ eq : ps_neutron_s12 ] ) .
the dotted lines correspond to the the imaginary parts.,title="fig : " ] energy shift of the proton ( left ) and neutron ( right ) with pseudo - scalar coupling for @xmath90 mev .
the upper row corresponds to the shift with @xmath91 and the lower row to @xmath92 .
the linear solid green lines corresponds to the linear approximation eq .
( [ eq : linear_approximation ] ) while the curved solid red lines to eqs .
( [ eq : ps_proton],[eq : ps_neutron_s3 ] ) .
the solid blue line shows the result of eq .
( [ eq : ps_neutron_s12 ] ) .
the dotted lines correspond to the the imaginary parts.,title="fig : " ] in fig .
[ fig : ps_self_energies ] we plot the nucleon energy shifts together with the linear approximation with @xmath93 mev and @xmath90 mev .
we see that the signs of the amm are in agreement with phenomenology , i.e. @xmath94 and @xmath95 .
to estimate @xmath39 effects , we use the difference of the red and blue lines in the lower right neutron graph .
the combination @xmath92 does not have @xmath39 contributions whereas the difference of the expressions eq .
( [ eq : ps_neutron_s3 ] ) and eq . ( [ eq : ps_neutron_s12 ] ) is of @xmath39 .
plotting the same results for various pion masses shows that the @xmath38 effects are small for magnetic field strengths of @xmath96 for pion masses larger than @xmath97 mev and for @xmath98 with pion masses around @xmath99 mev . within these region the eqs.([eq : linear_approximation],[eq : ps_neutron_s3],[eq : ps_neutron_s12 ] ) give approximately the same results .
we estimate therefore that the self - energy formulas are applicable for magnetic field strengths of @xmath96 with pion masses between the physical point @xmath100 mev up to @xmath99 mev .
this is the applied pion mass range in lattice qcd calculations .
in addition we have also for the nucleon the non - analytic @xmath73 expressions , eq .
( [ eq : omega_contributions ] ) , which would constrain a definition of the magnetic moment by the linear energy shift .
these constrains read : @xmath101 which can always be fulfilled for @xmath102 .
further , we obtain for the nucleon self - energy an imaginary part when a stronger magnetic field is applied .
the imaginary parts come only from @xmath103 and @xmath104 and read : @xmath105 for @xmath106 in @xmath103 and @xmath107 in @xmath104 .
the nucleon results of this section are obtained from the general expression eq .
( [ eq : self_energy_ps ] ) for particles with yukawa couplings as in the lagrangian eq .
( [ eq : ps_lagrangian ] ) and are transcript - able for other stable particles .
for the second example we include a resonance ( @xmath108 ) of mass @xmath109 with the coupling @xmath110 and get with @xmath111 and @xmath112 the self - energies : @xmath113 + \mathcal{o}\left(b^{2}\right)\,\,\,.\label{eq : resonance_self_energy}\end{aligned}\ ] ] from this , we also recover the results of the three point method : @xmath114 since the resonance is unstable we have the following imaginary parts for the self - energies : @xmath115 for @xmath116
. explicitly , these parts are present for magnetic fields of @xmath117 in the case of @xmath118 and @xmath104 with a magnetic field of @xmath119 we have @xmath120 , eq .
( [ eq : imag_resonance ] ) has do be altered accordingly and an additional cusp is present at @xmath121 . in fig .
[ fig : sigma_resonance ] we show all three possible self - energies for the parameters @xmath93 mev , @xmath122 mev and @xmath90 mev together with the linear approximations . for the graph
@xmath103 we see the linear behavior near @xmath123 and a cusp appearing according to eq .
( [ eq : imag_sigma1_resonance ] ) .
this graph was investigated in @xcite . in the case of @xmath118
we see only one cusp at @xmath124 since for the present mass constellation eq .
( [ eq : imag_sigma2_resonance ] ) is fulfilled for all @xmath125 and an imaginary part is steadily present .
however , choosing @xmath126 closer to the mass - gap @xmath127 the second cusp appears on the @xmath128 side .
the occurrence of such two cusps can be seen for @xmath104 .
the cusp for @xmath128 is due to the imaginary part from eq .
( [ eq : imag_sigma3_resonance ] ) and the one for @xmath129 due to @xmath119 . as we approach with @xmath126 the mass - gap @xmath127 , i.e. @xmath130 , all cusps corresponding to eqs .
( [ eq : imag_sigma1_resonance],[eq : imag_sigma2_resonance],[eq : imag_sigma3_resonance ] ) converge on @xmath123 where we have @xmath131 . for a @xmath126 mass larger than the mass gap
the resonance will not be unstable anymore and we get similar results as in the previous example . the explicit conditions to expand the self - energies for small @xmath73 due to the non - analytic contributions @xmath72 , eq .
( [ eq : omega_contributions ] ) , are : @xmath132 for the parameters @xmath133 , @xmath134 and @xmath135 .
the first condition was found in @xcite .
which condition is the most strict one on @xmath73 depends on the charge of the external particle and the actual values of the participating masses . for @xmath136 and @xmath137
it is the first one .
self - energies of a resonance in an external magnetic field @xmath0 .
the linear green lines correspond to the linear approximation eq .
( [ eq : linear_approximation ] ) while the curved red lines to eqs .
( [ eq : resonance_self_energy ] ) .
the solid lines show the real parts and the dotted lines the imaginary parts.,title="fig : " ] self - energies of a resonance in an external magnetic field @xmath0 .
the linear green lines correspond to the linear approximation eq .
( [ eq : linear_approximation ] ) while the curved red lines to eqs .
( [ eq : resonance_self_energy ] ) .
the solid lines show the real parts and the dotted lines the imaginary parts.,title="fig : " ] self - energies of a resonance in an external magnetic field @xmath0 .
the linear green lines correspond to the linear approximation eq .
( [ eq : linear_approximation ] ) while the curved red lines to eqs .
( [ eq : resonance_self_energy ] ) .
the solid lines show the real parts and the dotted lines the imaginary parts.,title="fig : " ]
we apply now the bft to a more involved situation , namely , the nucleon self - energy in the su(2 ) chiral perturbation theory of @xcite : @xmath138 the covariant derivatives are given in the appendix and the nucleon axial - vector and the pion decay constants are @xmath139 and @xmath140 mev .
we consider the two graphs @xmath103 and @xmath118 in fig .
[ fig : feynman_graphs ] and obtain for the nucleon the following unrenormalized self - energies in @xmath141 dimensions : @xmath142\,\,\,,\\ \sigma_{2}^{n}\left(b\right ) & = & i\left(\frac{g_{a}}{2f_{\pi}}\right)^{2}m_{n } \int_{0}^{1}dz\big[-m_{n}^{2}\left(1-z\right)^{3}j_{2}+\left(-6 + 3z\right)j_{1}+\left(3-z\right)\varepsilon j_{1}{\nonumber}\\ & & \,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\,\ , + b\left(2\left(1-z\right)^{4}m_{n}^{2}j_{3}+\left(3 - 8z+6z^{2}-z^{3}\right)j_{2}-\left(3 - 4z+z^{2}\right)\varepsilon j_{2}\right)\big]\,\,\,.\end{aligned}\ ] ] the loop integrals @xmath143 with @xmath144 are listed in the appendix .
the expressions @xmath103 and @xmath118 only differ by @xmath0 dependent terms and we write for the nucleon self - energies : @xmath145 by integrating the feynman parameter we get the @xmath9 renormalized proton self - energy @xmath146\,\,\,,\label{eq : p_self_energy}\\ n_{1}\left(\mathcal{b},\mu\right ) & = & 36 - 156\mathcal{b}+236\mathcal{b}^{2}-135\mathcal{b}^{3 } + 12\mathcal{b}^{4}+7\mathcal{b}^{5}+\left(72 - 252\mathcal{b}\right)\mu^{2 } + \mathcal{o}\left(\mathcal{b}^{2}\right)\,\,\,,\\ n_{2}\left(\mathcal{b},\mu\right ) & = & -24\mathcal{b}^{3}+42\mathcal{b}^{4}-12\mathcal{b}^{5 } + 96\mathcal{b}\mu^{2}+\left(-144 + 348\mathcal{b}\right)\mu^{4}+\left(36 - 72\mathcal{b}\right)\mu^{6 } + \mathcal{o}\left(\mathcal{b}^{2}\right)\,\,\,,\\ n_{3}\left(\mathcal{b},\mu\right ) & = & -24\mathcal{b}^{2}+42\mathcal{b}^{3}-12\mathcal{b}^{4}+120\mathcal{b}\mu^{2 } + \left(-36 + 72\mathcal{b}\right)\mu^{4}+\mathcal{o}\left(\mathcal{b}^{2}\right)\,\,\,,\end{aligned}\ ] ] with @xmath147 .
we give here only terms up to @xmath148 or constant in @xmath149 and list in the appendix the full coefficients @xmath150 . by setting @xmath151
we recover the normal nucleon @xmath9 renormalized b@xmath5pt self - energy @xcite : @xmath152\,\,\,.\label{eq : bchpt_sigma}\end{aligned}\ ] ] the chiral expansion of the nucleon mass to order @xmath153 is @xmath154 where @xmath155 , @xmath156 and @xmath157 are the low - energy constants for the nucleon mass and its amm .
the @xmath158 and @xmath159 terms in eq .
( [ eq : bchpt_sigma ] ) break the usual power counting scheme @xcite where e.g. the eoms renormalization scheme @xcite is one way to deal with this problem . the general behavior of the self - energy in this section is similar to the nucleon self - energy of the last section for pseudo - scalar @xmath8 couplings .
especially the non - analytic @xmath73 term is the same and we get again an imaginary part for @xmath160 .
proton self - energy as function of a magnetic field @xmath0 for @xmath90 mev and @xmath93 mev .
the solid red line is the result of eq .
( [ eq : p_self_energy ] ) and the solid linear green line the linear approximation @xmath161 with @xmath162 .
the dashed line is the imaginary part of eq .
( [ eq : p_self_energy ] ) . ] in fig .
[ fig : proton_sigma ] we show the proton self - energy as function of the magnetic field @xmath73 for the phenomenological values . in case of the linear approximation we use the amm obtained from the three - point method , i.e. defined by @xmath163 via the matrix element @xmath164u(p)\,\,\,,\end{aligned}\ ] ] with @xmath165 as the momentum transfer .
the explicit results for the nucleon graphs in the second row of fig .
[ fig : feynman_graphs ] are : @xmath166\label{eq : pv_kappa1}\\ \kappa_{2}^{n } & = & \frac{1}{i}\left(\frac{g_{a}m_{n}}{2f_{\pi}}\right)^{2}\int_{0}^{1}dz2 \left[\left(3 - 14z+15z^{2}-4z^{3}\right)j_{2}-\left(3 - 7z+5z^{2}-z^{3}\right)\varepsilon j_{2}+2m_{n}^{2}\left(1-z\right)^{5}j_{3}\right]\label{eq : pv_kappa2}\end{aligned}\ ] ] with @xmath167 and @xmath168 @xcite . extracting the amm from the above self - energy yields identical results where in the case of @xmath169 this can be seen literally even before the feynman parameter integration : @xmath170.\end{aligned}\ ] ] the two @xmath171 integrals
add up to the same expression in eq .
( [ eq : pv_kappa1 ] ) . however , in the case of the three - point function method , e.g. , the @xmath171 contributions come purely from tensor integrals whereas in the case of the bft method it is a combination of tensor and scalar loop - integrals . as we see for the infinite volume case this difference is not important .
we consider now the @xmath172-isobar and concentrate on the graphs that give its decay width .
we take the following lagrangian @xcite : @xmath173 and obtain the relevant self - energies as : @xmath174\,\,\,,\\ \sigma_{2}^{\delta}\left(b\right ) & = & \frac{1}{i}m_{\delta}\left(\frac{h_{a}}{2f_{\pi}}\right)^{2}\frac{1}{2 } \int_{0}^{1}dz\left[\left(z+r\right)j_{1}-b\left(z+r\right)\left(1-z\right)^{2}j_{2}\right]\,\,\,,\\ \mathcal{m}_{\delta } & = & z\mu^{2}+\left(1-z\right)r^{2}-z\left(1-z\right)+\mathcal{b}_{i}\,\,\,,\label{eq : delta_self_energies}\end{aligned}\ ] ] with @xmath175 and @xmath86 and all other definitions given in the appendix . integrating
these expressions yield @xmath176\,\,\omega_{1}\left(b\right)\,\,\,,\\ \sigma_{2}^{\delta}\left(b\right)\cdot\frac{144}{m_{\delta}c_{\delta } } & = & \mathcal{a}_{2}\left(b\right)+48\left[2\left(r+\alpha_{2}\right)\lambda_{2}^{2 } + \mathcal{b}\left(3r+3\alpha_{2}-6r\alpha_{2}-\left(3\alpha_{2}^{2}+1\right)\left(2 + r-\alpha_{2}\right)\lambda_{2}^{2}\right)\right]\,\,\omega_{2}\left(b\right)\,\,,\\ \omega_{i}\left(b\right ) & = & \sqrt{\lambda_{i}^{2}}\left(\mbox{arctanh } \frac{\beta_{i}}{\sqrt{\lambda_{i}^{2}}}+\mbox{arctanh } \frac{\alpha_{i}}{\sqrt{\lambda_{i}^{2}}}\right)\,\,,\label{eq : omega_delta}\end{aligned}\ ] ] with @xmath177 and the analytic parts @xmath178 also listed in the appendix . with these two expressions we can also write the self - energy of the different iso - spin states as : @xmath179 the @xmath172-isobar self - energy loop results as function of a magnetic field @xmath0 for @xmath90 mev , @xmath93 mev and @xmath180 mev .
the left pictures shows the real part and the right one the imaginary part .
the curved red solid lines are from eq .
( [ eq : delta_self_energies ] ) while the linear green lines correspond to the linear approximation @xmath181 .
, title="fig : " ] the @xmath172-isobar self - energy loop results as function of a magnetic field @xmath0 for @xmath90 mev , @xmath93 mev and @xmath180 mev .
the left pictures shows the real part and the right one the imaginary part .
the curved red solid lines are from eq .
( [ eq : delta_self_energies ] ) while the linear green lines correspond to the linear approximation @xmath181 .
, title="fig : " ] in fig .
[ fig : delta_sigma ] we show the @xmath172 self - energy as function of the magnetic field @xmath0 for the phenomenological parameters together with the linear approximation as obtained from the three - point function method .
the cusp comes from the @xmath103 while the second cusp of @xmath118 is not present for @xmath90 mev , however it emerges on the @xmath182 side for larger pion masses and both cusps fall on @xmath123 for @xmath183 .
the imaginary parts read @xmath184\,\,\,,\\ \mbox{im}\,\sigma_{2}\left(b\right)\cdot\frac{3}{\pi m_{\delta}c_{\delta } } & = & -2\left(1-\mathcal{b}\right)\left(r+\alpha_{2}\right)\lambda_{2}^{3}{\nonumber}\\ & & -\mathcal{b}\lambda_{2}\left[-6\alpha_{2}^{2}+3\alpha_{2}^{3}-2\lambda_{2}^{2 } + 3\alpha_{2}\left(1+\lambda_{2}^{2}\right)+r\left(3 - 6\alpha_{2}+3\alpha_{2}^{2}+\lambda_{2}^{2}\right)\right]\,\,\,,\end{aligned}\ ] ] where the @xmath172 decay width , experimentally given by @xmath185 mev , is obtained from @xmath186 . the slope of the imaginary part at @xmath123 in fig .
[ fig : delta_sigma ] is consistent with @xcite and we obtain a vanishing decay width at : @xmath187 the magnetic moment from the three - point function method is again defined through the vector matrix element by @xmath163 and @xmath188g^{\alpha\beta } + \left[f_{3}^{\delta}\gamma^{\mu}+\frac{i\sigma^{\mu\nu}q_{\nu}}{2m_{\delta}}f_{4}^{\delta}\right ] \frac{q^{\alpha}q^{\beta}}{4m_{\delta}^{2}}\big\}u_{\beta}(p)\,\,\,,\end{aligned}\ ] ] with the results @xcite : @xmath189j_{2}\,\,\,,\\ \kappa_{2}^{\delta } & = & \frac{1}{i}\left(\frac{h_{a}m_{\delta}}{2f_{\pi}}\right)^{2}\int_{0}^{1}dz\,2 \left(1-z\right)\left[\left(1+r\right)\left(1-z\right)-\left(1-z\right)^{2}\right]j_{2}\,\,\,,\end{aligned}\ ] ] with @xmath190 and @xmath191 .
the results of the bft method agree in this form literally . from @xmath192 in eq .
( [ eq : omega_delta ] ) we obtain the condition @xmath18 for expanding the @xmath7-isobar self - energies in small magnetic fields @xcite .
we investigated the self - energies of particles placed in a constant electromagnetic ( em ) field .
explicitly , we applied the em background field technique ( bft ) once to the situation of stable and unstable spin-1/2 particles coupling to ( pseudo- ) scalar fields and once to the nucleon and @xmath7-isobar baryons in the @xmath193 chiral perturbation theory ( b@xmath5pt ) .
we obtained the self - energies of these particles as function of the external constant magnetic field @xmath0 and calculated from these the anomalous magnetic moments ( amm ) by the linear energy shift .
we summarize our findings as : * self - energies of dirac particles coupling to ( pseudo- ) scalar fields : we investigated all three types of feynman graphs that can appear in the one - loop bft .
the self - energies generally depend non - analytically on @xmath0 where for stable particles the actual non - analytic points are unproblematic for defining the amm by the linear energy shift .
these points depend on the mass and charge constellations in the loop and give three different conditions for resonances on when their self - energies can be expanded for weak @xmath0 .
one of these conditions was reported earlier .
the self - energy formulas contain those @xmath39 terms that keep the non - analytical dependence intact , in contrast to the one - photon approximation .
however , they omit some @xmath39 contributions where we estimated for the nucleon - pion system with pseudo - scalar couplings that these contributions are small for magnetic fields of @xmath96 with @xmath194 as the nucleon mass .
* self - energies of the nucleon and @xmath7-isobar in b@xmath5pt : we derived the formulas for the nucleon and @xmath7-isobar self - energies depending on the pion mass and the magnetic field .
we recover the expressions as obtained from the three point function method as well as the condition on when the @xmath7-isobar magnetic moment is well defined by the linear energy shift .
further , we saw that the amm expressions have different tensor- and scalar - loop integral combinations depending on whether the amm is derived by the three point function method or from the self - energy . in the infinite volume these combinations add up to the same amm expressions .
the author is thankful to b. tiburzi , v. pascalutsa and m. vanderhaeghen for valuable discussions and critical comments .
we use the following notations : @xmath195 with @xmath196 dimensions , @xmath197 the covariant derivatives are : with the operators @xmath199 and @xmath200 . the results for the loop integration in infinite volume and dimensional regularization with @xmath201 is : @xmath202\,\,\,,\,\,\ , & j_{2}\left(\mathcal{m}\right)=\frac{-i}{\left(4\pi\right)^{2 } } \left[l+\ln\tilde{\mathcal{m}}\right]\,\,\,,\,\,\ , & j_{3}\left(\mathcal{m}\right)=\frac{-i}{\left(4\pi\right)^{2 } } \frac{1}{2m_{sc}^{2}}\frac{1}{\tilde{\mathcal{m}}}\,\,\,.\end{aligned}\ ] ] the corresponding parts of the nucleon and @xmath7 self - energies are : @xmath204_{\alpha,\beta,\lambda\to\alpha_{1},\beta_{1},\lambda_{1}}\\ \mathcal{a}_{2}\left(b\right ) & = & \big[\mathcal{a}\left(b\right)+\mathcal{b } \big(-40 - 128r+32\alpha+240r\alpha+36\alpha^{2}-168r\alpha^{2}-48\alpha^{3 } + 132\lambda^{2}+120r\lambda^{2}-48\alpha\lambda^{2}{\nonumber}\\ & & + 12\left(2 - 4\alpha+3\alpha^{2}-3\lambda^{2}+r\left(4 - 6\alpha+6\alpha^{2 } -6\lambda^{2}\right)\right)\ln\left(1-\mathcal{b}\right){\nonumber}\\ & & + \left(72r\alpha+36\alpha^{2}-72r\alpha^{2}-48\alpha^{3}+48r\alpha^{3 } + 24\alpha^{4}+36\lambda^{2}-72r\lambda^{2}-144\alpha\lambda^{2}+72\alpha^{2 } \lambda^{2}\right)\ln\left(\alpha^{2}-\lambda^{2}\right){\nonumber}\\ & & -24(-1+\alpha)^{2}\left(-1 + 2r(-1+\alpha)+\alpha^{2}+3\lambda^{2}\right ) \ln\left(\beta^{2}-\lambda^{2}\right)\big ) \big]_{\alpha,\beta,\lambda\to\alpha_{2},\beta_{2},\lambda_{2}}\end{aligned}\ ] ] j. gegelia , g. japaridze , phys . rev . *
d60 * , 114038 ( 1999 ) .
+ j. gegelia , g. japaridze , x. q. wang , j. phys . *
g29 * , 2303 ( 2003 ) .
+ t. fuchs , j. gegelia , g. japaridze , s. scherer , phys .
rev . * d68 * , 056005 ( 2003 ) . | we investigate the self - energies of particles in an external magnetic field @xmath0 .
the dependence is generally of the type @xmath1 with @xmath2 a polynomial in @xmath0 and the participating masses .
the non - analytic point depends on the mass and charge constellation , is unproblematic for stable particles but constrains the linear energy shift approximation for resonances .
we recover an earlier reported condition on when the energy can be expanded in @xmath0 and derive two more conditions .
further , we obtain the @xmath0 dependent self - energies of the nucleon and @xmath3-isobar in the @xmath4 covariant chiral perturbation theory . |
we performed a @xmath0 hypernuclear spectroscopic measurement , which used the @xmath1 reaction at jlab hall - c in 2009 ( jlab e05 - 115 @xcite@xcite ) .
the scattered electron ( @xmath3 ) and kaon ( @xmath4 ) were measured in coincidence using the he s ( high resolution electron spectrometer ) and hks ( high resolution kaon spectrometer ) , dedicated electron and kaon spectrometers .
the positions and angles of the scattered particles were measured with tracking detectors in the focal planes of each spectrometer and this information was used to reconstruct a missing mass .
however , hks , the kaon spectrometer , had a very large background of protons and positive pions .
an aerogel erenkov detector and a water erenkov detector , used to reject @xmath5s and protons both on - line and off - line , played an important role in @xmath4 identification .
these erenkov detectors were mounted close to the hks dipole magnet ( figure [ fig : hks_setup ] ) in order to minimize loss of kaons due to decay .
therefore , fringe field of up to @xmath2 5 g were observed in the region of the photomultipler tubes ( pmts ) on the erenkov detectors .
it was expected that the @xmath4 identification efficiency would be reduced because of the effects of magnetic fields on pmt performance .
6 g around the erenkov detectors effects on their pmt performance.,width=321 ] a previous experiment ( jlab e01 - 011 @xcite@xcite@xcite@xcite in 2005 ) used iron shields around the pmts .
however , this was not optimal as it is difficult to shield magnetic fields parallel to the axis ( perpendicular to the face ) of the tube and required installation of large amount of iron . therefore , for this experiment we implemented a bucking coil system to more effectively cancel out the fringe fields .
generate erenkov light within the hks momentum acceptance .
the aerogel erenkov detector is used as a veto detector to suppress @xmath6 particles.,width=302 ] from other particles by choosing an optimized threshold of number of photoelectrons.,width=302 ] over the hks momentum acceptance of @xmath21.05 gev/@xmath7 to @xmath21.35 gev/@xmath7 , the number of background particles are 2,000:1 for protons and 10,000:1 for @xmath6s relative to kaons . using aerogel and water , erenkov detector index of refraction values of 1.05 and 1.33 were chosen .
figures [ fig : np_aerogel ] and [ fig : np_water ] show predicted erenkov light photon yields for the three particle species in the two radiator media calculated by : @xmath8 where @xmath9 is the number of photons , @xmath10 is the fine structure constant , @xmath11 is the wave length of the erenkov light , @xmath12 is the refraction index of the medium , and @xmath13 and @xmath14 are the charge and the velocity factor of the incident particle . in the figures , @xmath12 is fixed at 1.05 and 1.33 for aerogel and water , and the yield is integrated over a @xmath11 range between 300 nm and 600 nm , corresponding to the sensitive region of the pmts .
as the kaon and proton velocities are below threshold in aerogel , @xmath6 and lighter particles can be rejected .
the water erenkov detector is sensitive to all three particle species , but can distinguish protons and kaons by the difference in photon yields .
we required that a on - line rejection ratio of the order of 1/100 for @xmath6 and @xmath15 to obtain the desired hypernuclear yield and signal to accidental background ratio .
the goal of a high background rejection ratios must be balanced against the desire to maximize the efficiency of @xmath16 detection , particularly in the water erenkov detectors .
this efficiency is determined by photo - electron collection efficiency of pmts , a property that is strongly affected by magnetic fields .
a monte carlo simulation was done to estimate on - line background rejection ratios and kaon detection efficiency .
this simulation was normalized with a cosmic ray tests which found , for the water erenkov detectors , a mean signal of 54 photoelectrons per cosmic ray . according to the simulation
, two layers of water erenkov cuts that rejects @xmath17 of protons will accept @xmath16 with a @xmath18 efficiency .
however , when the mean number of photoelectrons is reduced by @xmath19 ( mean signal of 13 photoelectrons ) , the @xmath16 acceptance rate drops to @xmath20 for the same proton rejection ratio , directly reducing the yield for @xmath0 hypernuclei . given the limits on beam time , efforts to mitigate the effect of magnetic field on kaon detection efficiency are justified . in order to maximize the yield of @xmath0 hypernuclei and the signal to noise ratio , we set a goal that magnetic fields should not reduce photoelectron yields by more than @xmath21 .
before the setup of jlab experiment e05 - 115 was mounted , studies were made of the effects of magnetic fields on the erenkov pmts and of the ability of bucking coils to cancel these effects .
the test setup is shown in fig .
[ fig : test_setup ] which also defines the coordinate system . a two inches diameter h7195 pmt ( hamamatsu , photocathode of bialkali , dynode stages of 12 ,
supply voltage of @xmath22 v , typical gain of 3.0@xmath2310@xmath24 ) , which was used in the water erenkov detector , was tested .
a helmholtz coil which can generate magnetic fields to @xmath2516 g was constructed . the cross section and the number of turns of each conductor
are 45 mm @xmath23 46 mm and 114 , respectively .
the excitation curve of the coil was measured by a hall probe ( group3 technology , digital teslameter of dtm-151-ds , probe of mpt-141 - 10s , accuracy of @xmath2610@xmath27 ) and compared to a calculation by opera3d - tosca@xcite which is a magnetic field calculation software by using three dimensional finite element method in the figure [ fig : by_coil ] .
the measured values of magnetic field of y - component ( b@xmath28 ) are lower by a few @xmath29 than those of tosca calculation
. if the probe has small angles or / and little displacement from the center , the measured values could be lower than the real center value of @xmath30 . in the present paper , the following values of magnetic field
are derived from the applied current by using the measured relation . ) and current ( i ) correlation at the center of the coils which are measured by the hall probe and calculated by opera3d - tosca@xcite.,width=302 ] the uniformity of the helmholtz coil was also measured . in the range of 400 mm @xmath31 400 mm around the center which we put pmt for the test ,
the uniformity of @xmath30 was @xmath32 the uniformity is checked by tosca as well .
figure [ fig : by_ratio ] is showing @xmath33 by tosca calculation .
the solid line describes the value when the values of @xmath34 and @xmath35 are fixed as 0 mm .
the dashed line describes the value when the values of @xmath34 and @xmath13 are fixed as 0 mm . as seen in figure [ fig : by_ratio ]
, the tosca calculation shows @xmath36 in the range of 400 mm @xmath23 400 mm , which is consistent with the measurement within required accuracy for the bucking coil test .
( center ) vs. displacement from the center of the coils by the tosca calculation .
the solid line describes the value when the values of @xmath34 and @xmath35 are fixed as 0 mm .
the dashed line describes the value when the values of @xmath34 and @xmath13 are fixed as 0 mm .
the calculation is consistent with the measurement within required accuracy for the bucking coil test.,width=340 ] the other components of magnetic field in the region of 400 mm @xmath23 400 mm were @xmath37 according to the measurements .
an led ( stanley 3889s yellow ) was used as the light source for the gain study of pmt .
the peak wave length and luminous intensity of the led are 690 nm and 4 mcd .
the led was covered by a light diffusion cap in order to achieve uniform illumination of the pmt photocathode .
the led was placed at the bottom of a paper cylinder with an inside surface covered by a teflon reflector sheet .
the distance between the bottom of the paper cylinder and the photocathode of pmt was 260 mm .
the cylinder was wrapped with two turns of light - tight black sheets to prevent light leaks .
figure [ fig : circuit ] shows the electronics block diagram for the pmt tests .
the led generates light using transistor - transistor logic ( ttl ) pulses from a clock generator .
the light intensity of the led is controlled by varying the width of the pulses .
the width was set at @xmath2530 ns to generate @xmath25200 photo - electrons .
the charge information from the pmt was digitized by an analog to digital converter ( adc ) with a trigger derived from the led drive signal .
studies were made of the effects of magnetic fields on the pmts before the bucking coils were installed .
the results are presented as `` relative gain '' , which is defined as the ratio of adc value with a magnetic field to the adc value with no magnetic field .
it should be noted that any change in `` relative gain '' would originate from multiple sources as 1 ) changes in the number of photoelectrons collected by the first dynode of the pmt and 2 ) changes in the amplification factor after the first dynode of the pmt .
the @xmath38 is defined as the angle between the magnetic field direction ( the @xmath35-axis ) and the pmt axis ( perpendicular to the face of the pmt ) of the pmt .
2000 v. different markers represent different angles between the tube axis and magnetic field direction ( @xmath39 degrees ) .,width=340 ] figure [ fig : pmt_ang_dep ] shows the angular dependence of the relative gain .
the relative gain is more than @xmath40 with @xmath41 g for all @xmath38 .
however , it drops to @xmath2540@xmath29 with a magnetic field of 5 g parallel to the pmt axis ( @xmath42 ) .
5 g is the magnitude of the fringe field seen on these tubes in the spectrometer .
as seen , the relative gain is strongly affected by magnetic fields along the axis of the pmt .
the relative gain when using the led is strongly affected by the magnetic field as discussed in the previous section . to see changes in the relative gain for a single photo - electron peak ,
data were taken with a self trigger .
figure [ fig : pmt_opemag_dep ] shows relative gain of one photo - electron peak as a function of magnetic field .
2300 v .,width=340 ] the relative gain of one photo - electron is near to 1 , and is slightly increasing with field .
if the overall relative gain reduction originates only from that gain factor reduction after the first dynode , this result should show similar behavior to figure [ fig : pmt_ang_dep ] when we used an led trigger .
in addition , a signal rate reduction was observed . according to these results , the reduction in relative gain originates mainly from a reduction of probability that electron which is converted from photon on photocathode reaches the first dynode of pmt .
it is also noted that the relative gain increase in figure [ fig : pmt_opemag_dep ] does not necessarily imply that gain factor is increased . under higher magnetic fields , only higher energy electrons may be able to reach the first dynode .
these higher energy electrons can knock off more secondary electrons at the first dynode , and what we observed might be biased by these events . 2300 v , @xmath432000 v and @xmath431700 v for the pmt ( h7195 ) . ,
width=340 ] figure [ fig : pmt_hv_dep ] shows the relative gain as a function of the magnetic field which is applied parallel to the axis of the pmt ( @xmath42 ) with different hv settings . as the high voltage is increased , the effects of the magnetic field are eased slightly ( e.g. the relative gain at 5 g are 0.42 for @xmath432300 v and 0.31 for @xmath431700 v ) . however , this improvement is insufficient for our experimental requirements .
figure [ fig : buckingcoil_setup ] shows setup of a bucking coil on a pmt as used for the tests .
a wire with cross section of 4 mm@xmath44 and a resistance of 0.2 @xmath45 m was used to wind the bucking coil .
the wire of 20 turns was rolled onto the pmt , starting at photocathode making a coil with the length of 90 mm .
the current applied to the bucking coil were set to cancel the magnetic field inside the pmt .
the relative gain as a function of current for the bucking coil under magnetic fields of 4 g , 5 g and 6 g ( @xmath42 ) are shown in the fig.[fig : bc ] . )
the length of bucking coil of 20 turns was 90 mm starting at the photocathode of pmt ( h7195 , @xmath432000 v ) .,width=340 ] the relative gains were recovered to @xmath21 for each applied magnetic field on the pmt by choosing the optimal current for the bucking coil .
those curves have a plateau region .
if one chooses bucking coil current of 2.5 a @xmath25 5.0 a , the relative gain is more than 0.95 under the magnetic field of 5 g. 2000 v ) .,width=340 ] in the experiment , jlab e05 - 115 , bucking coil can not be rolled onto the pmt starting at the photocathode because of physical interferences from the frame of the erenkov detector . in anticipation of this constraint
, tests were made with a bucking coil that started 40 mm away from the photocathode .
the results are represented in the fig.[fig : bc_pos_dep ] .
the relative gains are recovered to @xmath2 1 for each applied magnetic field as well as the fig .
[ fig : bc ] .
however , the necessary currents are higher than those of fig . [
fig : bc ] .
this means that the relative gain is more sensitive to the magnetic field around the photocathode of the pmt .
figure [ fig : bc_ang_dep ] shows the relative gain as a function of the bucking coil current with angles of @xmath46 0 , 30 , 60 and 90 degrees . as angles of the magnetic field with respect to the axis of pmt closer to @xmath47 , the maximum relative gains are lower though they are still more than 0.95 for @xmath48 .
this imperfection of relative gain recovery is considered to be caused by a perpendicular component of magnetic field to the axis of pmt which can not be cancelled by the bucking coil .
0 , 30 , 60 and 90 degrees ( h7195 , @xmath432000 v ) .,width=340 ] ) was used , and the bucking coil started at the photocathode of the pmt ( h7195 , @xmath432000 v ) .,width=340 ] a relation between relative gain and the number of bucking coil turns is shown in fig .
[ fig : bc_turns_dep ] . in this test ,
a magnetic field of 5 g was applied on pmt ( @xmath42 ) .
the bucking coil started at the photocathode of the pmt , and a hv setting of @xmath432000 v was used .
less current is needed to recover the relative gain up to @xmath251 with coils with a greater number of turns though the plateau region becomes narrower .
1700 v , @xmath432000 v and @xmath432300v
. a magnetic field of 5 g ( @xmath42 ) was applied , and the bucking coil position started at the photocathode of the pmt ( h7195 ) .,width=340 ] figure [ fig : bc_hv_dep ] shows the relative gain as a function of bucking coil current with different hv settings , @xmath431700 v , @xmath432000 v and @xmath432300 v. in this test , a magnetic field of 5 g was applied to the pmt ( @xmath49 ) , and the bucking coil of 20 turns was wound starting at the photocathode of the pmt .
the necessary currents to recover relative gains up to @xmath25 1 for each hv setting are very similar . in the actual experiment setup ,
different hv settings were used for pmts .
the obtained results imply that a single bucking coil current can work though different hv settings are used in the actual experimental setup .
40 mm away from the photocathode .
unit in the figure is mm.,width=340 ] a schematic drawing of the water erenkov detector is shown in the fig .
[ fig : wc_setup ] .
there are two layers of water erenkov detector with 12 segments for each layer .
a detector segment consists of an acrylic box with its inside surface covered by a teflon sheet , and pmts ( h7195 ) on top and bottom of the box with uv - glass windows .
deionized water ( resistivity of 18 m@xmath50 cm , the refraction index of 1.33 ) was used as the radiation medium .
a bucking coil of 40 turns was rolled around each pmt , starting @xmath240 mm away from the photocathode .
figure [ fig : wc_bg_photo ] is a photograph of bucking coils on pmts of the water erenkov detectors . in the experiment ,
the effects of bucking coil for the water erenkov detector were checked by measuring the counting rates of each pmt , which is equivalent to measuring the relative gains as discussed in section [ sec : ope ] if the discriminator threshold is kept the same .
the ratios of counting rates of each pmt when the hks dipole magnet was on to those when the magnet was off are shown in the fig .
[ fig : wc_real ] .
the counting rates for each pmt are decreased by more than @xmath250@xmath29 when the bucking coil is not used . on the other hand , the counting rates with bucking coil on ( 40 turns ,
the current of 2 a ) are recovered to those with no hks magnetic field . 1 when the bucking coils are on ( 40 turns , the current of 2 a ) .,width=340 ] figure [ fig : ac_setup ] shows a schematic drawing of the aerogel erenkov detector
. there are three layers of aerogel erenkov detectors , and each layer has 7 segments .
a segment consists of a structurally strengthened paper box , 40 pieces of hydrophobic silica aerogel tiles ( sp-50 , matsushita , refraction index of 1.05 ) and two pmts ( xp4572b / d1 , photonis , photocathode of bialkali , dynode stages of 10 , typical supply voltage of + 2100 v , typical gain of 2.0@xmath2310@xmath51 ) , located on the top and bottom of the box without any windows .
it is noted that the pmt which is used for aerogel detector is more sensitive to magnetic fields because its diameter ( 5 inches ) is larger than that for water erenkov detector ( 2 inches ) .
figure [ fig : ac_bg_photo ] shows the bucking coils on pmts for the aerogel erenkov detectors .
40 turns of bucking coils were used for each pmt .
figure [ fig : ac_real ] shows the relative gain of the each pmt in the first layer of the aerogel erenkov detectors as a function of bucking coil current . because the fringe field of the hks dipole is not uniform in the region of the aerogel erenkov detector , the optimal bucking coil currents are different for each segment .
however , a single current of 8 a was chosen since it recovered most of the relative gain reasonably for all pmts . a schematic drawing of the hks detectors is represented in the fig .
[ fig : hks_setup ] .
the hks consists of three layers of plastic scintillation detectors ( ktof1x , ktof1y , ktof2x ) for tof ( time of flight ) measurement , two layers of horizontal drift chambers ( kdc1 , kdc2 ) for particle tracking , and the aerogel and water erenkov detectors ( ac1 , ac2 , ac3 , wc1 , wc2 ) for particle identification .
the hks trigger was made by the following logical condition , @xmath52 where , @xmath53 @xmath54 cp@xmath55 in eq.([eq : cp_trigger ] ) is a charged particle trigger which consists of the coincidence of three layers of tof detectors ( ktof1x , ktof1y and ktof2x ) . in eq .
( [ eq : k_trigger ] ) , ac represents ( ac1 @xmath56 ac2 ) @xmath57 ( ac2 @xmath56 ac3 ) @xmath57 ( ac3 @xmath56 ac1 ) , and wc represent ( wc1 @xmath56 wc2 ) .
the overline on ac means that the ac was used as veto to suppress @xmath5 .
the typical hks trigger rate was @xmath21 khz with a beam current of 2 @xmath58a on a polyethylene target ( ch@xmath59 , the material thickness of 450 mg / cm@xmath44 ) .
data with the hks@xmath55 and the cp@xmath55 conditions were taken during the experiment although a coincidence between the hks trigger ( hks@xmath55 ) and the he s trigger was used for physics data .
@xmath6 and @xmath15 rejection ratios were estimated by comparing the number of those events in the hks@xmath55 and the cp@xmath55 .
the rejection ratios of @xmath6 and @xmath15 with a beam current of 2 @xmath58a on the polyethylene target were @xmath61 and @xmath62 , respectively .
the @xmath4 detection efficiency was 91@xmath29 .
at the on - line ( hardware ) trigger level , thresholds of the erenkov detectors were set slightly loose so as to avoid over - cutting of @xmath4s .
therefore , some @xmath5s and protons remained in the recorded data .
the top plot of fig .
[ fig : msq_aer ] shows the correlation between number of photoelectrons from the sum of the three layers of aerogel erenkov detectors and mass squared of the measured particle .
the bottom plot is an x - projection of the top graph .
mass squared , @xmath63 is calculated by the following equation , @xmath64 where @xmath65 is the momentum reconstructed by transfer matrix of the hks , and @xmath14 is the velocity factor derived from the tof measurement . in fig .
[ fig : msq_aer ] , clusters of @xmath6 , @xmath4 and @xmath15 are clearly seen .
@xmath6s are rejected by applying the cut on the number of photoelectrons in the aerogel erenkov detector .
two types of boxes of water erenkov detectors were used in the experiment .
the main differences between them were reflective materials and pmt choice . on the low momentum side
( segment numbers from 1 to 6 , box1 ) , white acrylic as reflective material and h7195 pmts were used . on the high momentum side where severer particle identification is required ( segment numbers from 7 to 12 , box2 ) , teflon sheets as the reflective material and h7195uv pmts which has uv - glass window were used .
h7195uv has the same responses to the magnetic field as h7195 , but higher efficiency to ultra violet light . in cosmic ray tests , the average number of photoelectrons in box1 and box2 type segments were @xmath250 and @xmath2100 , respectively .
there was difference of number of photoelectrons between box1 and box2 , and thus , normalized number of photoelectrons were introduced to adjust @xmath4 peak to 1 .
the top plot of fig .
[ fig : msq_wat ] shows the correlation between the normalized number of photoelectrons from the sum of two layers of water erenkov detectors and the mass squared , while the bottom plot is the x - projection of the top figure .
protons and other particles ( @xmath6 , @xmath4 ) can be separated by the cut on the number of photoelectrons .
figure [ fig : msq ] shows the mass squared distribution before and after the cuts on the number of photoelectrons in the erenkov detectors .
@xmath4 events are clearly selected after applying these cuts . when the erenkov and mass squared cuts were selected to keep 90@xmath29 kaons in the total off - line events , the contaminated @xmath6 and @xmath15 events were 4@xmath29 and 3@xmath29 of remaining events ( using the 450
mg / cm@xmath44 ch@xmath59 target ) . in this case , the total ( on - line and off - line ) rejection powers of @xmath6 and @xmath15 were 6.5@xmath2310@xmath27 and 6.1@xmath2310@xmath66 , respectively .
events are clearly selected after the cuts .
the width of single gaussian fitting for kaon peak is @xmath67 ( 0.29 [ gev/@xmath68 ) @xmath44.,width=340 ]
the water and aerogel erenkov detectors , which were located a few meters away from the hks dipole magnet , needed to be operated in a fringe field of @xmath25 g. the separation efficiency of @xmath4 from the other background particles ( @xmath6 and @xmath15 ) is deteriorated in these fields as the fields reduce the relative gains .
monte carlo simulations indicated that relative pmt gains of more than 65@xmath29 were required to keep the yield of @xmath0 hypernuclei and the signal to noise ratio reasonably high . to operate the pmts attached to erenkov radiators under this field , bucking coils
were installed to actively and locally cancel the magnetic field on the pmts .
+ tests with a helmholtz coil were performed to study the effects of magnetic fields on the h7195 pmt which was used for the water erenkov detectors . with a magnetic field of @xmath25 g parallel to the axis of the pmt ,
the gain was reduced by 60@xmath29 ( relative gain of 0.40 ) . a bucking coil of 40 turns and @xmath22 a starting at the photocathode of pmt was needed to recover the relative gain up to @xmath21 in this case .
+ in jlab e05 - 115 experiment , bucking coils of 40 turns and current of 2 a were used for the water erenkov detectors , and coils with 40 turns and current of 8 a were used for aerogel erenkov detectors . with these coils , a relative gain of more than 0.70 and 0.75 for each segment of the water and aerogel erenkov detectors , respectively , were achieved .
the on - line rejection ratios of @xmath6 and @xmath15 with a beam current of 2 @xmath69a on the ch@xmath59 target ( the material thickness of 450 mg / cm@xmath44 ) were 7.4 @xmath70 and 3.8 @xmath71 , respectively .
the efficiency for @xmath4 events was 91@xmath29 .
off - line , @xmath4 events were selected by applying cuts on the number of photoelectrons in the erenkov detectors . when erenkov and mass squared cuts were selected to keep 90@xmath29 efficiency in off - line kaon selection , the contaminated @xmath6 and @xmath15 events were 4@xmath29 and 3@xmath29 of remaining events ( using the 450 mg / cm@xmath44 ch@xmath59 target ) . in this case , the total ( on - line and off - line ) rejection powers of @xmath6 and @xmath15 are 6.5@xmath2310@xmath27 and 6.1@xmath2310@xmath66 , respectively . the implementation of bucking coils allowed for a very clean on - line and off - line selection of @xmath4 events .
we would like to thank jlab staff of physics , accelerator and engineer divisions for supports for the experiments .
the program at jlab s hall - c is supported by the specially promoted program ( 12002001 ) , the creative research program ( 16gs0201 ) , grant - in - aid by mext ( 15684005 ) , japan , jsps research fellowships for young scientists ( 244123 ) , us - japan collaboration research program , the jsps core - to - core program ( 21002 ) and the strategic young researcher overseas visits program for accelerating brain circulation ( r2201 ) by the japan society for promotion of science .
this work is supported by u.s .
department of energy contract de - ac05 - 06or23177 under which jefferson science associates , llc , operates the thomas jefferson national accelerator facility .
o.hashimoto , s.n.nakamura , l.tang , j.reinhold et al . , jlab e05 - 115 proposal ( 2005 ) o.hashimoto et al . ,
nuclear physics a835 , 121 - 128 ( 2010 ) l.yuan et al . ,
physical review c73 , 044607 ( 2006 ) l.tang et al . , nuclear physics a790 , 679c-682c ( 2007 ) o.hashimoto et al . ,
nuclear physics a804 , 125 - 138 ( 2008 ) s.n.nakamura et al . ,
physical review letters 100 , 012502 ( 2013 ) http://www.cobham.com , cobham antenna systems , vector fields simulation software | aerogel and water erenkov detectors were employed to tag kaons for a @xmath0 hypernuclear spectroscopic experiment which used the @xmath1 reaction in experimental hall c at jefferson lab ( jlab e05 - 115 ) .
fringe fields from the kaon spectrometer magnet yielded @xmath2 5 gauss at the photomultiplier tubes ( pmt ) for these detectors which could not be easily shielded .
as this field results in a lowered kaon detection efficiency , we implemented a bucking coil on each photomultiplier tubes to actively cancel this magnetic field , thus maximizing kaon detection efficiency . |
although both gravity and gauge theories contain a local symmetry , they have rather disparate physical properties .
qcd , for example , is a confining theory while gravity is not .
similarly , in four dimensions qcd is renormalizable while field theories of gravity are non - renormalizable . nonetheless , within the context of the perturbative expansions some remarkable ` squaring ' relations exist between the tree - level @xmath3-matrix elements of gravity and gauge theories in both string and field theories @xcite .
these squaring relations imply that gauge symmetry is more closely related to the diffeomorphism invariance of gravity than one might suspect based on the respective lagrangians .
recently , there has also been a resurgence of interest in perturbative gravity both in anti - de sitter spaces @xcite and in phenomenological applications of large compact dimensions @xcite , further motivating an investigation of these relations . in this letter
we will show that one can construct a low energy lagrangian for gravity directly from qcd @xmath3-matrix elements and then discuss how this lagrangian relates to the usual einstein - hilbert lagrangian .
the starting point of our investigation is the kawai , lewellen and tye ( klt ) relations expressing closed string tree amplitudes in terms of open string amplitudes @xcite .
these relations arise from the property that the integrand of a closed string amplitude is composed of left- and right - mover open string components ( see e.g. ref .
@xcite ) . in the low energy limit where string theory reduces to field theory , the klt relations express gravity tree amplitudes in terms of a sum of products of
` left ' gauge theory amplitudes and ` right ' gauge theory amplitudes .
the klt relations were first exploited by berends , giele and kuijf to obtain an infinite sequence of maximally helicity violating pure gravity tree amplitudes @xcite using known gauge theory results . by unitarity
, tree - level relations necessarily imply that loop amplitudes in the two theories are also related .
indeed , the method of constructing @xmath3-matrix elements via their analytic properties ( see refs .
@xcite ) provides a means for exploiting this to obtain loop - level ( super ) gravity amplitudes . using these ideas ,
the divergence structure of @xmath4 supergravity has been investigated with the result that it appears to be less divergent than previously thought @xcite .
two infinite sequences of maximally helicity violating one - loop amplitudes in gravity theories have also been constructed @xcite . these may be viewed as results in an effective field theory of gravity which necessarily must match the low energy limit of a more fundamental theory , such as string or @xmath5 theory .
the ability to use the squaring relation between gravity and gauge theory to perform non - trivial gravity computations suggests that one can develop a deeper understanding of perturbative gravity by exploring this relationship .
from the point of view of the field theory lagrangians , the klt relations are rather mysterious : the einstein - hilbert lagrangian does not factorize in any obvious way in terms of the yang - mills lagrangian .
it is not even completely clear what the notion of ` left ' and ` right ' parts of the theory mean given that the graviton is a symmetric tensor .
another obvious difficulty is that the yang - mills lagrangian has only three- and four - point interactions while the einstein - hilbert lagrangian has an infinite set of interactions .
nevertheless , the gravity @xmath3-matrix does have the property that it is composed of products of two gauge theory amplitudes and that a graviton , very roughly speaking , is composed of ` left ' and ` right ' gauge fields , i.e. , @xmath6 .
when one has a property of the @xmath3-matrix , a natural idea is to organize the underlying formalism so that it reflects that property . as a well known example , the general superiority of feynman diagrams as compared to time ordered perturbation theory follows from their preservation of manifest lorentz symmetry . here
we wish to rearrange the gravity lagrangian so it captures properties associated with the factorization of the @xmath3-matrix into sums of products of gauge theory amplitudes .
the analysis of this letter will consist of two parts . in the first part we will systematically construct a gravity lagrangian from qcd gluon amplitudes using the klt relations .
we explicitly carry out the procedure through five graviton interactions ; in principle , the procedure can be carried out to arbitrarily high orders , although as a practical matter the complexity of the computations increases rapidly with the number of gravitons .
nevertheless , the procedure shows that a lagrangian for gravity can be obtained using qcd amplitudes .
the gravity lagrangian obtained in this way has the properties that ( a ) by construction it produces the correct tree - level amplitudes and ( b ) graviton lorentz indices associated with the ` left ' gauge theory do not contract with ones associated with the ` right ' gauge theory .
a lagrangian with the property that the associated feynman diagrams factorize into left and right parts has been presented previously by siegel @xcite . a spin - off from our analysis
is that it can be used to provide a simpler set of gravity feynman rules than the conventional @xcite de donder gauge rules .
moreover , in our construction , the gravity three vertex is given directly as sums of squares of gauge theory vertices .
in the second part , we will find a set of field variables which allows us to interpret the gravity lagrangian obtained from qcd in terms of the conventional einstein - hilbert lagrangian .
non - linear field redefinitions and gauge fixings have been used previously by van de ven @xcite to simplify the computation of two - loop divergences in gravity . here
we wish to consider such field redefinitions and gauge fixings in order to find a particular set of field variables which helps clarify the connection to qcd .
this construction indicates that there is a closer correspondence between ordinary gauge invariance and the diffeomorphism invariance of einstein gravity than one might have suspected .
a basic property of integrands for closed string amplitudes is that they factorize into open string integrands ( except for the momentum zero mode ) . at tree - level ,
klt @xcite used this property to find simple relationships expressing closed string amplitudes in terms of products of open string amplitudes . in the infinite string tension limit where string theory reduces to an effective field theory , for the four- and five - point
amplitudes these relations are , @xmath7\ , , \cr } \equn\label{kltexamples}\ ] ] where , @xmath8 , @xmath9 , the @xmath10 are color - ordered gauge theory partial amplitudes @xcite and the @xmath11 are gravity amplitudes .
color does not appear in these relations because the gauge theory partial amplitudes are independent of color .
the arguments of the amplitudes label the external legs .
these relations hold for any particle states appearing in any closed string theory since they follow from the basic factorization of the string integrands into ` left ' and ` right ' sectors .
in particular , they hold in pure einstein gravity , gravity coupled to a dilaton , anti - symmetric tensor or gauge field , and in supergravity theories .
moreover , these klt relations generalize to an arbitrary number of external legs .
( explicit expressions may be found in appendix a of ref .
@xcite . ) a basic property of the klt formulas is that any lorentz indices associated with the ` left ' gauge theory amplitude do not contract with the lorentz indices of the ` right ' gauge theory amplitude .
the klt relations give tree - level gravity amplitudes directly from known gauge theory amplitudes , which are generally far easier to compute .
this was first applied by berends , giele and kuijf @xcite to obtain an infinite sequence of maximally helicity violating @xmath12-point einstein gravity tree amplitudes using known qcd gluon amplitudes . beyond tree - level , one may exploit the klt relations using the cutting method developed for performing qcd amplitude calculations of phenomenological interest @xcite . with this method
one can reconstruct complete loop amplitudes ( in a given dimensional regularization scheme ) from their @xmath13-dimensional unitarity cuts .
the one - loop amplitudes are specified in terms of tree amplitudes which satisfy the klt relations ; one can then obtain higher loop amplitudes by iterating the procedure . in this way , one can obtain loop - level ( super ) gravity @xmath3-matrix elements without reference to a lagrangian or feynman rules @xcite .
the efficiency of the computational method follows from the fact that one is _ recycling _ previously performed gauge theory calculations to obtain new results in gravity theories . 1.9 truein -.3
cm for example , consider the unitarity cuts of a one - loop four - point amplitude in any gravity theory coming from the low energy limit of a string theory .
as depicted in , a gravity tree amplitude appears on each side of the cut .
thus , on the cut one must evaluate the product , @xmath14 summed over all states in the gravity theory that can cross the cut . from the klt relations ( [ kltexamples ] )
we may re - express this product in terms of gauge theory amplitudes , @xmath15 the sum over the gravity theory states necessarily factorize into a sum over the states of two gauge theories because the underlying string theory has this property .
this is a generic property holding for all cuts . in this way
, one can re - express cut gravity amplitudes in terms of gauge theory ones .
this has led to the construction of a number of non - trivial gravity loop amplitudes @xcite .
we now discuss a procedure for constructing an off - shell low energy lagrangian for pure gravity starting from qcd gluon amplitudes .
we explicitly carry this out for up to five graviton interactions . by construction ,
the lagrangian that we obtain is equivalent to the einstein - hilbert lagrangian in that it produces identical tree - level @xmath3-matrix elements . in field
theory one usually takes a given lagrangian and constructs feynman rules that can then be used to obtain the @xmath3-matrix elements . here
we wish to reverse this process , since from the klt relations we have the gravity @xmath3-matrix in terms of qcd amplitudes , but wish instead to obtain a lagrangian that preserves the ` left'-`right ' factorization of lorentz indices . to obtain the @xmath12-graviton term in the lagrangian we start with the tree - level @xmath12-graviton amplitudes , given in terms of the qcd amplitudes , and subtract all diagrams containing up to @xmath16-graviton interaction vertices . by iterating this procedure we can systematically build a gravity lagrangian
. however , to jump - start this process we first need a propagator and three vertex .
consider the graviton propagator .
the standard de donder gauge propagator , @xmath17 { i\over k^2 + i\eps } \ , , \equn\label{dedonderpropagator}\ ] ] is unacceptable since it contracts ` left ' and ` right ' indices ; for example , the last term contracts @xmath18 with @xmath19 .
moreover , it contains explicit dependence on the dimension @xmath13 , which must somehow cancel from the tree - level @xmath3-matrix elements since there is no such dependence in the klt relations or in the gauge theory amplitudes .
( we have chosen @xmath20 to have signature @xmath21 . )
a better propagator is @xcite @xmath22 however , even this propagator is unacceptable because of the second term , which contracts a left index with a right one . to prevent such contractions , we wish to use instead the propagator , @xmath23 in a sense , for each graviton @xmath24 we assign the index @xmath18 to be a ` left ' index and @xmath19 to be a ` right ' index . of course
, since @xmath24 is symmetric it does not matter which index is assigned to the left and which to the right , but once the choice is made we demand that left indices never contract with right ones . at first sight
the gravity propagator ( [ newpropagator ] ) might seem rather peculiar since it appears to violate the fundamental property of the graviton being a symmetric tensor field . nevertheless , it is not difficult to show that one may use the propagator ( [ newpropagator ] ) instead of ( [ symmetricpropagator ] ) with no effect on the @xmath12-graviton @xmath3-matrix elements , provided that all vertices satisfy a _ rigid _ left
right interchange symmetry . by a rigid symmetry
we mean that each @xmath12-point vertex be symmetric under a simultaneous interchange of all left and right lorentz indices , @xmath25 where the @xmath26 are the left indices associated with each graviton and the @xmath27 are the right indices . with this requirement
, one can show that there is no need to symmetrize the graviton propagator in the left and right indices .
consider , for example , the diagram in .
the external legs of the diagrams do not need to be explicitly symmetrized since the external graviton polarization tensors automatically symmetrize the legs .
the interchange of the @xmath18 and @xmath19 indices can be undone by performing a rigid interchange of left and right indices on the left - most vertex .
although the rigid interchange will also flip the indices on the external legs , this can be undone since the indices of each external leg are contracted with a symmetric tensor polarization .
the same type of argument works for general diagrams .
thus , the propagator ( [ newpropagator ] ) can be used instead of the propagator ( [ symmetricpropagator ] ) since the symmetrization of indices is automatically taken care of by the rigidly symmetrized vertices and by the symmetric polarization tensors on the external legs .
1.2 truein -.2 cm given the propagator ( [ newpropagator ] ) we must also choose a three vertex .
any three vertex which agrees on shell with the three - graviton vertex is suitable for our purposes .
( kinematic restrictions prevent a massless particle from decaying into two others , but in terms of formal polarization tensors the vertex does not vanish ; below we shall also obtain the same vertex from the einstein - hilbert action . ) from string theory ( see e.g. ref .
@xcite ) , the on - shell three - graviton vertex can be expressed in terms of products of gauge theory vertices .
this motivates the choice of the three graviton vertex , @xmath28\ , , \cr } \equn\label{gravityymthreevertex}\ ] ] where @xmath29 is the color ordered gervais - neveu @xcite yang - mills three vertex , from which the color factor has been stripped .
our main reasons for using the vertex ( [ gravityymthreevertex ] ) is its simplicity and the fact that it makes the relationship to gauge theory manifest . by construction
, this vertex is symmetric under the rigid interchange of left and right indices .
armed with the propagator ( [ newpropagator ] ) and the three vertex ( [ gravityymthreevertex ] ) we may then use the klt relations to find a gravity lagrangian using qcd gluon amplitudes as input . at the first step of the process one defines a four - vertex by subtracting from the four - point @xmath3-matrix obtained via the klt relations all diagrams containing a kinematic pole , as shown in .
this four vertex automatically has the property that left lorentz indices do not contract with right ones , since the gravity @xmath3-matrix and diagrams containing the three - point vertex ( [ gravityymthreevertex ] ) have this property .
moreover , the vertex defined in this way also has the rigid symmetry under an interchange of left and right indices . 3.8 truein -.2 cm we may then convert this four - vertex to an @xmath30 term in the lagrangian by inverting the usual procedure of obtaining feynman vertices from lagrangians .
of course , there is some ambiguity in this process since one can always add terms which vanish on shell . as long as these terms
do not mix left and right lorentz indices and satisfy a rigid left - right symmetry , which can always be imposed by hand , they are acceptable terms ; differences in the four - vertex then induce differences in the deduced higher - point vertices .
once the four - graviton terms in the lagrangian have been chosen we can then continue the process to obtain a five vertex by subtracting from the five - graviton amplitude all diagrams containing three- and four - point vertices . in principle
, one can continue in this way to an arbitrary number of external legs , allowing one to build a lagrangian for gravity order by order in perturbation theory using qcd gluon amplitudes . by construction
this lagrangian has the property that it generates all tree - level @xmath3-matrix elements and that it never contracts left indices with right ones .
starting from qcd gluon tree amplitudes , we have carried out the construction for up to five points yielding the local gravity lagrangian @xmath31 , where @xmath32\ , , \cr l_4 & = -\kappa^2 \bigl [ { 1\over 32 } h_{\mu\nu,\lambda } h_{\mu\nu,\lambda } h_{\rho\sigma } h_{\rho\sigma } + { 1\over 2 } h_{\mu \nu,\lambda}h_{\mu\rho}h_{\sigma\nu,\rho } h_{\sigma\lambda } + { 1\over 8 } h_{\mu \nu}h_{\mu \lambda,\rho}h_{\sigma \nu,\rho } h_{\sigma \lambda } -{1\over 4 } h_{\mu\nu}h_{\mu\lambda}h_{\rho\sigma,\nu } h_{\rho\sigma,\lambda}\cr & \hskip 1.3 cm -{1\over 4 } h_{\mu\nu } h_{\lambda\rho}h_{\lambda\sigma,\rho } h_{\mu\sigma,\nu } + { 1\over 16 } h_{\mu\nu } h_{\mu\nu}h_{\lambda\rho } h_{\lambda\rho,\sigma\sigma } + { 1\over 24 } h_{\mu\nu } h_{\lambda\nu } h_{\lambda\rho } h_{\mu\rho,\sigma\sigma}\bigr ] \ , , \cr l_5 & = \kappa^3 \bigl [ { 3\over 64 } h_{\mu\nu } h_{\mu\nu } h_{\rho\sigma } h_{\rho\sigma,\kappa\lambda } h_{\kappa\lambda } -{1\over 12 } h_{\mu\nu } h_{\mu\nu,\kappa } h_{\rho\sigma,\lambda } h_{\rho\kappa } h_{\lambda\sigma } -{1\over 48 } h_{\mu\nu,\kappa } h_{\mu\nu } h_{\rho\sigma } h_{\rho\kappa,\lambda } h_{\lambda\sigma } \cr & \hskip 1.3 cm -{1\over 12 } h_{\mu\nu,\kappa } h_{\mu\nu,\lambda } h_{\rho\sigma } h_{\rho\kappa } h_{\lambda\sigma } -{3\over 32 } h_{\mu\nu } h_{\mu\nu } h_{\rho\sigma,\kappa\lambda } h_{\rho\kappa } h_{\lambda\sigma } -{1\over 12 } h_{\mu\nu } h_{\rho\nu,\mu } h_{\rho\sigma } h_{\lambda\sigma,\kappa } h_{\lambda\kappa } \cr & \hskip 1.3 cm -{1\over 6 } h_{\mu\nu } h_{\rho\nu } h_{\rho\sigma } h_{\lambda\sigma,\mu\kappa } h_{\lambda\kappa } + { 1\over 6 } h_{\mu\nu } h_{\rho\nu,\kappa } h_{\rho\sigma,\mu } h_{\lambda\sigma } h_{\lambda\kappa } -{1\over 6 } h_{\mu\nu } h_{\rho\nu,\kappa } h_{\rho\sigma } h_{\lambda\sigma,\mu } h_{\lambda\kappa } \cr & \hskip 1.3 cm + { 1\over 12 } h_{\mu\nu } h_{\rho\nu } h_{\rho\sigma,\mu\kappa } h_{\lambda\sigma } h_{\lambda\kappa } + { 1\over 8 } h_{\mu\nu } h_{\rho\nu,\lambda } h_{\rho\sigma } h_{\mu\sigma,\kappa } h_{\lambda\kappa } + { 1\over 24 } h_{\mu\nu } h_{\rho\nu,\lambda\kappa } h_{\rho\sigma } h_{\mu\sigma } h_{\lambda\kappa } \bigr]\ , , \cr } \equn \label{gaugelagrangian}\ ] ] where @xmath33 . in minkowski space , one of any two contracted indices
should be taken to be a raised index using @xmath34 , but we have suppressed this . in principle
it might have been necessary to introduce auxiliary fields for locality to hold ; indeed , as discussed below , when comparing this lagrangian to the einstein - hilbert gravity lagrangian it is useful to introduce an auxiliary dilaton .
although the lagrangian ( [ gaugelagrangian ] ) is not unique since the terms can be modified by adding or subtracting contributions that vanish on shell and appropriately modifying the higher - point contributions , it is a relatively simple one . more importantly , as we shall see below
, it allows for a relatively straightforward match to the conventional einstein - hilbert lagrangian . by adjusting the four - point terms in the lagrangian
we have found a solution for @xmath35 that contains only six terms , but then the connection to the einstein - hilbert lagrangian is a bit more complicated . for compactness
we have removed the rigid symmetrization between left and right indices ; however , if one uses the propagator ( [ newpropagator ] ) , each vertex must be rigidly symmetrized between left and right indices , e.g. , @xmath36\ , .
\equn\ ] ] although we do not explicitly give the generated feynman rules here , it is quite straightforward to obtain these .
the advantages of using feynman rules generated by this lagrangian instead of the conventional de donder gauge rule are clear : besides the fact that one can use the propagator ( [ newpropagator ] ) instead of the more complicated de donder gauge propagator ( [ dedonderpropagator ] ) the three- , four- and five - point vertices are quite a bit simpler than the corresponding de donder gauge vertices . once the interaction terms have been rigidly symmetrized , when deriving the feynman rules we can in a sense treat @xmath37 as a tensor field with no special index symmetry .
note that the feynman diagrams generated by this lagrangian do not contain explicit factors of @xmath13 . in the conventional de donder
gauge such factors do appear , but somehow cancel from the @xmath3-matrix elements .
the relative simplicity of these feynman rules is related to the preservation of the @xmath3 matrix property that left and right lorentz indices should not contract with each other .
although the klt relations might appear to suggest that field variables exist where gravity can be reformulated as a polynomial theory with no more than four - point interactions as in the gauge theory case , we have been unsuccessful in finding such a lagrangian .
the lagrangian ( [ gaugelagrangian ] ) does contain general coordinate invariance although it is not manifest since the symmetry has been gauge fixed . to make this explicit ,
we now relate the lagrangian ( [ gaugelagrangian ] ) to the usual einstein - hilbert one . in order to do
so we must first eliminate all terms in the einstein - hilbert lagrangian which contract left with right graviton indices , i.e , @xmath38 \ , , \equn\label{badterms}\ ] ] where @xmath39 \equiv h_{\mu_1 \mu_2 } h_{\mu_2 \mu_3 } \cdots h_{\mu_n \mu_1}$ ] . because of the way that these types of terms are tangled in the einstein - hilbert lagrangian , it is not obvious how one can accomplish this .
nevertheless , the existence of the lagrangian ( [ gaugelagrangian ] ) implies that there must be some rearrangement with the desired property . in ref .
@xcite the dilaton was introduced to allow for a field redefinition which removes the graviton trace from the quadratic terms in the lagrangian .
the appearance of the dilaton as an auxiliary field to help rearrange the lagrangian is motivated by string theory which requires the presence of such a field .
following the discussion of ref .
@xcite , we then consider the lagrangian for gravity coupled to a dilaton , @xmath40 ( our conventions are those of weinberg @xcite , except that we have flipped the signature of the metric , @xmath41 ; this then induces an overall sign flip in the lagrangian . ) in de donder gauge , for example , taking @xmath42 , the quadratic part of the lagrangian is @xmath43 the term involving @xmath44 can be eliminated with the simultaneous field redefinitions @xcite , @xmath45 so that the lagrangian reduces to @xmath46 ( this field redefinition is a bit different than the one in ref .
@xcite because we have chosen to normalize the dilaton kinetic term differently so as to slightly simplify the induced interaction terms discussed below . ) for the case of purely graviton external states the dilaton will not contribute to the tree - level @xmath3-matrix because of the selection rule that @xmath47 must be created or annihilated in pairs .
of course , if the dilaton , or any matter fields appearing in the theory , are taken as external states they can propagate in intermediate states . in this case , one would need to specify the precise matter content of the theory before proceeding with the analysis .
nevertheless , the fundamental factorization of the @xmath3-matrix implied by the klt relations would still hold , since it follows from the underlying string theory . here
we wish to generalize the field redefinitions ( [ fieldredef ] ) to all orders in @xmath48 so as to remove all terms of the form ( [ badterms ] ) from the lagrangian .
our generalization for the case of no gauge fixing is @xmath49 where @xmath37 is the graviton field , followed by the change of variables @xmath50 we have verified through @xmath51 that this choice eliminates all terms ( [ badterms ] ) which mix left and right lorentz indices , even before fixing the gauge .
as yet we have not performed any gauge fixing so the action is generally coordinate invariant , even if the choice of field variables obscures this .
one might be concerned that the field redefinition ( [ stringvars ] ) would alter the gravity @xmath3-matrix .
however , the @xmath3-matrix is guaranteed to be invariant under non - linear field redefinitions or under linear ones that do not alter the coupling to the external traceless polarization tensors . indeed , our explicit calculations respect this property , as required .
an important remaining question is how one can choose a gauge fixing so that the terms in the einstein - hilbert lagrangian resemble the terms of the lagrangian ( [ gaugelagrangian ] ) deduced from the qcd amplitudes .
we have found a solution , which is to replace the field redefinition ( [ phiredef ] ) with a non - linear generalization , @xmath52 \ , .
\equn\label{phiredefb}\ ] ] then we add a gauge fixing term to the lagrangian , @xmath53 following the standard procedure , where @xmath54 with these choices we then obtain the desired form of the lagrangian through @xmath55 , which we express in terms of the lagrangian ( [ gaugelagrangian ] ) derived from qcd amplitudes , @xmath56 \,.\cr } \equn\ ] ] for the case of pure graviton external states we may integrate out @xmath47 from the path integral .
equivalently , we may substitute the equation of motion for @xmath47 , @xmath57 into the lagrangian . for the two- and three - graviton terms in the lagrangian , this gives exactly the terms in the lagrangian ( [ gaugelagrangian ] ) .
for the four - point we obtain exactly @xmath58 plus a non - local piece that vanishes for on - shell gravitons , @xmath59 since the four - point lagrangians differ a bit off - shell , at the five - point level , terms that are non - zero on - shell need to be added to @xmath35 , @xmath60 \ , . \equn\ ] ] in a sense we may take @xmath61 as the more fundamental one since it comes directly from the einstein - hilbert lagrangian , while the lagrangian in provides the guidance needed to obtain it .
this shows the connection of the lagrangian obtained from qcd ( [ gaugelagrangian ] ) and the einstein - hilbert lagrangian for a very particular set of field variables and and gauge fixing .
although the two lagrangians differ somewhat even after integrating out the auxiliary dilaton , we have explicitly shown that for up to five gravitons they generate the same on - shell scattering amplitudes .
presumably , similar results can be obtained without introducing the dilaton ; nevertheless , we found it useful for clarifying the required reorganization of the einstein - hilbert lagrangian .
other reorganizations based on introducing other fields such as an anti - symmetric tensor , which is also motivated by string theory , are also possible @xcite .
at loop level there are also ghost contributions that can be obtained via the usual methods .
( the jacobian generated by the field redefinition is trivial , at least in perturbation theory , since it is a point transformation . )
additionally the auxiliary dilaton would propagate in the loops , which would then need to be subtracted ( see e.g. , ref .
@xcite ) . at loop level
the unitarity method advocated in refs .
@xcite is , however , an efficient way to obtain the @xmath3-matrix without the need for feynman rules .
as one simple application , one may obtain the soft factor for gravity amplitudes directly from the soft behavior of qcd amplitudes .
gravity tree amplitudes have the well known behavior @xcite , @xmath62 as the momentum of graviton @xmath12 , which we have taken to carry positive helicity , becomes soft .
using the three - graviton vertex ( [ gravityymthreevertex ] ) which is expressed in terms of the qcd three - gluon vertex , the gravity soft factor can then be expressed in terms of the qcd soft factor : @xmath63 where @xmath64 is the eikonal factor for a positive helicity soft gluon in qcd .
the @xmath65 are spinor inner products and @xmath66 are massless weyl spinors of momentum @xmath67 , labeled with the sign of the helicity ( see e.g. , @xcite ) .
although not manifest , the soft factor ( [ finalsoftgrav ] ) is independent of the choices of null helicity ` reference momenta ' @xmath68 and @xmath69 . by choosing @xmath70 and @xmath71 we recover the form of the soft graviton factor for @xmath72 used in , for example , refs .
@xcite . as a less trivial example
, the explicit factorization of the left and right indices of the interaction vertices would imply that the all - plus helicity graviton current satisfying eqs .
( b.10 ) and ( b.11 ) of ref .
@xcite do actually follow from einstein gravity . in ref .
@xcite the hitch in obtaining these recursion relations from einstein gravity was the different choices of helicity reference momenta on the left and on the right .
( although , not derived directly from einstein gravity these recursion relations were useful for defining ` half - soft ' functions which serve as building blocks for one - loop gravity amplitudes . ) in this case , @xmath73 , which prevented a derivation of the recursion relations from einstein gravity . with the factorization of left and right lorentz indices ,
such contractions simply do not occur .
since the tree - level @xmath3-matrix is generated by solving perturbatively the classical equations of motion , one might suspect that it is possible to relate more general solutions of the classical equations of motion for gravity to gauge theory solutions .
the property that the @xmath3-matrix can be expressed in terms of ` left ' and ` right ' sectors is generic in string theory .
for this reason , it should be possible to extend the discussion in this letter to theories containing , for example , the anti - symmetric tensor or to supergravity theories .
it also seems reasonable that gravity theories in curved spaces can be reformulated to express tree - level amplitudes in terms of gauge theory ones .
moreover , the intimate connection of perturbative gravity amplitudes with gauge theory ones suggests a closer connection of diffeomorphism invariance with non - abelian gauge symmetry than one might have suspected .
we feel that these issues deserve further attention .
s. gubser and i. klebanov , phys .
lett . b413:41 ( 1997 ) [ hep - th/9708005 ] ; + j. maldacena , adv .
theor . math .
phys . 2:231 ( 1998 ) [ hep - th/9711200 ] ; + s. gubser , i. klebanov and a. polyakov , phys . lett . b428:105 ( 1998 ) [ hep - th/9802109 ] ; [ hep - th/9802116 ] ; + e. witten , adv .
phys . 2:253 ( 1998 ) [ hep - th/9802150 ] . n. arkani - hamed , s. dimopoulos and g. dvali , phys .
b429:263 ( 1998 ) [ hep - ph/9803315 ] ; + e.a .
mirabelli , m. perelstein , m.e .
peskin , preprint hep - ph/9811337 ; + t. han , j.d .
lykken , r .- j .
zhang , preprint hep - ph/9811350 ; + j.l .
hewett , preprint hep - ph/9811356 . z. bern , l. dixon , d.c .
dunbar and d.a .
kosower , nucl .
b425:217 ( 1994 ) [ hep - ph/9403226 ] ; nucl .
b435:59 ( 1995 ) [ hep - ph/9409265 ] ; + z. bern and a.g .
morgan , nucl .
b467:479 ( 1996 ) [ hep - ph/9511336 ] ; + z. bern , l. dixon and d.a .
kosower , ann .
nucl . part .
sci . 46:109 ( 1996 ) [ hep - ph/9602280 ] ; nucl .
b513:3 ( 1998 ) [ hep - ph/9708239 ] .
w. siegel , phys .
d47:5453 ( 1993 ) [ hep - th/9302036 ] ; phys .
d48:2826 ( 1993 ) [ hep - th/9305073 ] ; in _ proceedings of strings 1993 _ , eds .
halpern , a. sevrin and g. rivlis ( world scientific , singapore , 1994 ) [ hep - th/9308133 ] .
dewitt , phys .
162:1239 ( 1967 ) ; + m. veltman , in _
les houches 1975 , proceedings , methods in field theory _ , eds .
r. balian and j. zinn - justin ( north - holland , amsterdam , 1976 ) ; + s. sannan , phys .
d34:1749 ( 1986 ) . | we demonstrate that qcd gluon amplitudes can be used to construct a lagrangian for gravity .
this procedure makes use of perturbative ` squaring ' relations between gravity and gauge theory that follow from string theory .
we explicitly carry out the construction for up to five - point interactions and discuss a set of field variables in the einstein - hilbert lagrangian for interpreting the lagrangian obtained from qcd . a spin - off from our analysis is that it can be used to provide simpler tree - level gravity feynman rules than for conventional gauges .
# 1_#1 _ = 0.1 cm = -2.5 cm = -2.0 cm = 23.2truecm = 16.7truecm 30 pt plus 1000pt minus 100pt # 1 # 1= to # 1eq .
( [ # 1 ] ) # 1eq .
( [ # 1 ] ) # 1#2eqs .
( [ # 1 ] ) and ( [ # 2 ] ) # 1#2eqs .
( [ # 1 ] ) and ( [ # 2 ] ) # 1fig .
[ # 1 ] # 1fig . [ # 1 ] # 1#2figs . [ # 1 ] and [ # 2 ] # 1#2figs . [ # 1 ] and [ # 2 ] # 1section [ # 1 ] # 1section [ # 1 ] # 1#2sections [ # 1 ] and [ # 2 ] # 1appendix [ # 1 ] # 1appendix [ # 1 ] # 1#2appendices [ # 1 ] and [ # 2 ] # 1#2appendices [ # 1 ] and [ # 2 ] # 1table [ # 1 ] # 1table [ # 1 ]
# 1 = = = by - by - by by 2 -to # 1.#2#1#2 # 1.#2 # 1.#2.#3 - 1mu#1 | # 2 | # 3 - 1mu # 1.#2.#3 - 1mu#1^+ | # 2 | # 3^+-1mu # 1.#2(#1#2 )
@=11 p # 1#2#3#4(#1#2#3#4 ) # 1#2x_#1 # 2 hep - th/9904026 ucla/99/tep/10 + hutp-99/a017 + april , 1999 + 2.cm * perturbative gravity from qcd amplitudes * 2.cm z. bern@xmath0 0.2 cm _ department of physics , + university of california at los angeles , + los angeles , ca 90095 - 1547 _ .4 cm and .3 cm a. k. grant@xmath1 0.2 cm _ department of physics , + harvard university , + cambridge , ma 02138 _ ' '' '' width 3.6in@xmath2 research supported by the us department of energy under grant de - fg03 - 91er40662
. + @xmath1 research supported by the national science foundation grant phy-9802709 . |
quantum computation is an attractive field on precision control of quantum systems . but
quantum systems are inevitably influenced by the decoherence effect induced from the environment , which stands in the way of physical implementation of quantum computation . in order to overcome this difficulty
, there are many strategies to correct or avoid errors during the implementation of the gate operations .
the geometric phases @xcite and their non - abelian extensions , quantum holonomies @xcite , accompanying in evolutions of quantum systems , unveil important geometric structures in the description of the dynamics of quantum states .
geometric phases depend only on global geometric properties of the evolution paths so that they are largely insensitive to many types of local noise . based on this distinct merit ,
holonomic quantum computation ( hqc ) , first proposed by zanardi and rasetti @xcite , has emerged as a promising strategy to implement universal quantum computation in a robust way @xcite .
it is well known that the geometric phases , either abelian or non - abelian , consist of both adiabatic and non - adiabatic parts . in adiabatic evolution
, it requires that quantum evolutions fulfill the famous adiabatic condition , and thus quantum gates constructed in this way generally have very slow speeds .
this is unacceptable as the time needed for an adiabatic quantum gate maybe on the order of the coherence time of the used quantum two - level systems in typical systems @xcite .
therefore , one needs to consult quantum gates based on non - adiabatic evolution @xcite , which does not require the adiabatic condition so that the gate speeds will only depend on the merits of the employed quantum systems . in this case , the built - in fault - tolerance of the non - adiabatic geometric phases provide a more practical way in implementing quantum computation .
therefore , many renewed efforts have recently been given in this direction both theoretically and experimentally @xcite .
meanwhile , as an another promising way to avoid the effect of decoherence , decoherence - free subspace ( dfs ) can suppress the collective dephasing noise caused by the interaction between quantum systems and their environment @xcite .
therefore , many efforts have been paid to combine the hqc with dfs encoding @xcite . in this way
, one can consolidate the best of the two quantum computation strategies , i.e. , resilient against environment induced collective decoherence of the dfs approach and the operational robustness of the hqc against local noises . here , we propose a non - adiabatic hqc ( nhqc ) scheme with dfs encoding based on nitrogen - vacancy ( nv ) centers coupled commonly to the whispering - gallery mode ( wgm ) of a fused - silica microsphere cavity .
the nv center system is considered as a promising candidate for physical implementation of quantum computation , due to its sufficiently long electronic spin lifetime as well as the possibility of coherent manipulation even at room temperature @xcite .
since the electronic spins of the nv centers can be well initialized and manipulated in an optical way , quantum gates acting on the single - spin state can be obtained with very high efficiently @xcite .
based on the symmetry structure of the system - environment interaction , we propose an encoding method of the logical qubits in the dfs , where any logical qubit state that undergoes a cyclic evolution will end up with a new state in the same subspace , without going out of the subspace .
moreover , based on the numerical simulation under realistic conditions , we show that our nhqc scheme in dfs can realize a universal set of quantum gates with high fidelity .
the setup we consider and the energy level configuration of the nv centers are schematically shown in fig . [ setup ] . in the cavity , the lowest - order wgm , corresponding to the light traveling around the equator of the microsphere , offers exceptional mode properties for reaching strong light - matter coupling . as shown in the fig .
[ setup](a ) , @xmath1 nv centers , from separate diamond nano - crystals , are strongly coupled to the wgm of a microsphere cavity @xcite .
the nv centers we considered can be modeled as three - level systems , as shown in fig .
1(b ) , where the states @xmath2 and @xmath3 are encoded as our qubit states @xmath4 and @xmath5 , respectively .
the state @xmath6 is labeled by @xmath7 and leave aside the metastable @xmath8 state , which has not yet been fully understood @xcite . in our implementation , the optical transition @xmath9 and @xmath10 ( with transition frequencies @xmath11 and @xmath12 ) are coupled by the wgm with frequency @xmath13 and a classical laser filed with frequency @xmath14 @xcite , respectively .
both coupling is far - off resonant from their transition frequencies so that the @xmath15 state can be adiabatically eliminated .
note that there is a degenerate state @xmath16 for the qubit state @xmath17 , these two degenerated states can be selectively addressed by the polarization of the optical field @xcite .
the nv centers are fixed and the distance of two nv centers is much larger than the wavelength of the wgm , so that each driven laser field , with frequency @xmath18 and initial phase @xmath19 , can interact individually with an nv center and the direct coupling among the nv centers can be negligible .
then the interaction of the whole quantum system , in units of @xmath20 , can be written as @xmath21,\end{aligned}\ ] ] where @xmath22 is the creation ( annihilation ) operator of the wgm of the cavity , @xmath23 is the frequency of @xmath24th energy level of the identical nv centers with @xmath25 , @xmath26 ( @xmath27 ) is the coupling strength between @xmath28th nv center and the cavity ( laser ) . in the interaction picture with respect to the free hamiltonian @xmath29 , under the rotating - wave approximation ( rwa ) , the interaction can be written as @xmath30 where we have assumed @xmath31 for simplicity , the effective cavity assisted interaction strength are @xmath32 with @xmath33 , @xmath34 , @xmath35 , and @xmath36 .
when two laser fields are applied to a pair of nv centers ( for example the @xmath37th and the @xmath38th ) , one can obtain the effective hamiltonian , under the condition of @xmath39 , as @xmath40 where @xmath41 . neglecting the level shift terms , which can be compensated by using additional lasers @xcite , the effective hamiltonian between the two nv centers reduces to @xmath42 with this hamiltonian ,
we next show how a universal set of non - adiabatic holonomic gates can be implemented .
now we turn to construct universal single logical qubit operations for the nhqc in dfs .
it is noted that a logical qubit consists of two physical qubits is not sufficient for a dephasing environment @xcite .
we here utilize three physical qubits to encode a logical qubit .
the interaction between three physical qubits ( nv centers ) and the dephasing environment can be described by the interaction hamiltonian , @xmath43 , where @xmath44 is the collective dephasing operator and @xmath45 is an arbitrary environment operator . in this case , there exist a three - dimensional dfs of @xmath46 where the computational basis , i.e. , the logical qubit states , are encoded as @xmath47 , @xmath48 , and @xmath49 is used as a third ancillary state . in order to implement the single logical qubit operation , we apply operation @xmath50 $ ] , which can be obtained from eq .
( [ h ] ) for an operate time of @xmath51 , to the nv centers @xmath52 and @xmath53 by tuning laser beams with the same detuning @xmath54 from their respectively transition and phase difference @xmath55 .
meanwhile , we also apply @xmath56 $ ] on nv centers @xmath53 and @xmath57 by tuning laser beams with the same detuning @xmath58 and identical phase .
then , in the dfs @xmath59 the effective hamiltonian reads @xmath60 where the effective rabi frequency @xmath61 with @xmath62 and the phase @xmath63 can be tuned by the amplitude of the incident lasers .
this hamiltonian , consisting of a wgm and three physical qubits ( three nv centers ) , is in the @xmath0 type with ancillary logical state @xmath64 at the top while the logical qubit states @xmath65 and @xmath66 at the bottom . in the dressed state representation , the two degenerate states of eq .
( [ h1 ] ) are @xmath67 decouples from the bright state @xmath68 and the excited state @xmath64 , while the bright state @xmath68 couples to the excited state @xmath64 with the effective rabi frequency @xmath69 . as a result
, the evolution operator @xmath70 can realize the holonomic gates with following two conditions .
the first one is that controlling the operation time to meet @xmath71 .
this condition will ensure the states evolving in the subspace @xmath72 undergo the cyclic evolution @xmath73 .
furthermore , the relation @xmath74 is always hold , which means the evolution is pure geometric and the parallel - transport condition is natural satisfied .
therefore , we can obtain the non - adiabatic holonomic single qubit gates in the space spanned by \{@xmath65 , @xmath66 } as @xmath75 where @xmath76 and @xmath77 can be chosen by tuning the amplitude and relative phase of laser beams , and thus a set of universal single qubit gates can be realized non - adiabatically . for examples
, we can implement a hadamard gate with @xmath78 and a phase gate as @xmath79 . with initial state @xmath80 ,
the red ( dot ) line represents the fidelity .
( a ) @xmath81 and @xmath82 .
( b ) @xmath83 and @xmath82.,width=491 ] inevitably , the implementation process will suffer from decoherence .
considering the main decoherence effect , the collective relaxation rate @xmath84 , dephasing rate @xmath85 of nv centers and the decay rate @xmath86 of the cavity , we simulate the performance of our scheme under realistic conditions with the lindblad master equation @xcite @xmath87+\frac 1 2 [ \kappa\mathscr{l}(a)+\gamma\mathscr{l}(s^-)+\gamma_{\phi}\mathscr{l}(s^z)],\end{aligned}\ ] ] where @xmath88 is the hamiltonian in the form of eq .
( [ int ] ) for the case of @xmath89 with the compensation of the stark shift , @xmath90 is the density operator , @xmath91 and @xmath92 is the lindblad operator . in our simulation , we have used the following conservative set of experimental parameters .
the nv centers are located near the microcavity surface and the maximum coupling between an nv center and the cavity could be @xmath93 ghz with the mode volume of @xmath94 @xcite . for @xmath95 , @xmath96 mhz with @xmath97 mhz , @xmath98 ghz and @xmath99 ghz to fulfill the condition of @xmath100 .
the qubit relaxation and dephasing rates are estimated to be @xmath101khz @xcite .
the cavity decay rate is @xmath102mhz with @xmath103 @xcite .
assume that the logical qubit is initially prepared in @xmath80 state while the cavity is in the vacuum state , the time - depend state populations and fidelity under the @xmath104 and hadamard gates are depicted in fig .
[ single](a ) and [ single](b ) with the fidelity to be @xmath105 and @xmath106 , respectively .
note that , in the simulation , we use the hamiltonian in eq .
( [ int ] ) , and thus the obtained high fidelity also verifies the validity of the effective hamiltonian in eqs .
( [ eff1 ] ) and ( [ h ] ) .
in addition , we assuming the collective dephasing of all the qubits , the violation of which will introduce very small infidelity to the gate operations , as shown in fig . [ max1](a ) and
[ max1](b ) for the two exemplified gates @xmath107 and @xmath78 respectively , where we have investigated different cases by choosing different @xmath108 in the initial state of @xmath109 . in the simulation
, we have choose different decoherence of the physical qubits as @xmath110 , @xmath111 and @xmath112 , respectively . in unit of @xmath113 with line ( circle ) the identical ( different ) decoherence of qubits under gate operations ( a ) @xmath114 , ( b ) @xmath115.,width=453 ]
the implementing of the holonomic one - qubit logical gates can be scalable to two - qubit scenario straightforwardly . for two logical qubits interacting collectively with the dephasing environment
, there will exist a six - dimensional dfs @xmath116 , @xmath117 , @xmath118 , and @xmath119 .
meanwhile , @xmath120 and @xmath121 are used as ancillary states . , the red line represents the fidelity .
( a ) initial state of qubit @xmath122 .
( b ) initial state of qubit @xmath123.,width=491 ] in order to implement the logical two - qubit operations , we apply the interaction in eq .
( [ h ] ) on the nv centers @xmath57 and @xmath124 , by tuning laser beams with detuning @xmath54 and phase difference @xmath125 , for the operation time of @xmath126 , i.e. , @xmath127 $ ] .
meanwhile , we also apply @xmath128 $ ] on the nv centers @xmath57 and @xmath129 by tuning laser beams with detuning @xmath58 and same phase .
the total effective hamiltonian then reads @xmath130 + \mathrm{h.c.},\end{aligned}\ ] ] where the effective rabi frequency @xmath131 with @xmath62 and the phase @xmath132 can be tuned by the amplitude of the incident laser .
obviously , the effective hamiltonian can be decomposed to two commuting parts as @xmath133 with @xmath134 therefore , @xmath135 = \exp\left[-i\pi h_a\right]\exp\left[-i\pi h_b\right]$ ] under the @xmath113 pulse criterion @xmath136 , which acts nontrivially on the computational subspace @xmath137 and @xmath138 , respectively .
that is , @xmath139 , consisting of a wgm and six nv centers , effectively reduces to two @xmath0-like hamiltonian with @xmath140 ( @xmath141 ) acts on the top state @xmath142 ( @xmath143 ) and the bottom states @xmath144 and@xmath145 ( @xmath146 and @xmath147 ) . analogous to the single - qubit gate case , the holonomic two - qubit logical gate in the subspace @xmath148 can be written as @xmath149 in general , we can realize nontrivial two - qubit holonomic logical gate in dfs by controlling the @xmath150 and @xmath151 separately , that is , adjusting the amplitude and phase of two laser beams , respectively .
for example , acting @xmath152 on logical qubit 1 and 2 and then acting @xmath153 on logical qubit @xmath53 with @xmath154 and @xmath155 , we can realize a cnot gate in view of eqs .
( [ u1 ] ) and ( [ u2 ] ) .
in addition , choosing the experimentally achievable parameters similar as in the single qubit case , numerical simulation of the populations and the fidelity of @xmath156 operation are shown in fig .
[ two](a ) and [ two](b ) with different initial state @xmath122 and @xmath123 , respectively .
the fidelity can reach about @xmath105 and @xmath157 .
in summary , we have put forward a universal set of non - adiabatic holonomic gates in dfs by using nv centers coupled to the wgm of a cavity . by controlling the amplitude and relative phase of the driving lasers , we can realize arbitrary single - qubit and two - qubit operations .
numerical simulation shows that our scheme is stable to deviation of corresponding experimental parameters and insensitive to decoherence such as collective noises and local noises .
ultrahigh quality factor of the cavity and the exceptional spin properties of the nv centers make our scheme a promising candidate in experimental implementation of nhqc in dfs with high gate fidelity and short operation time .
this work was supported by the nfrpc ( no .
2013cb921804 ) , the pcsirt ( no .
irt1243 ) , and the project of anhui xinhua university ( no .
2014zr010 ) . | a cavity qed implementation of the non - adiabatic holonomic quantum computation in decoherence - free subspaces is proposed with nitrogen - vacancy centers coupled commonly to the whispering - gallery mode of a microsphere cavity , where a universal set of quantum gates can be realized on the qubits . in our implementation , with the assistant of the appropriate driving fields , the quantum evolution is insensitive to the cavity field state , which is only virtually excited .
the implemented non - adiabatic holonomies , utilizing optical transitions in the @xmath0 type of three - level configuration of the nitrogen - vacancy centers , can be used to construct a universal set of quantum gates on the encoded logical qubits .
therefore , our scheme opens up the possibility of realizing universal holonomic quantum computation with cavity assisted interaction on solid - state spins characterized by long coherence times . 10 m. v. berry , `` quantum phase factors accompanying adiabatic changes , '' proc .
r. soc .
lond . a * 392 * , 45 - 57 ( 1984 ) .
y. aharonov and j. anandan , `` phase change during a cyclic quantum evolution , '' phys .
rev .
lett . * 58 * , 1593 ( 1987 ) .
f. wilczek and a. zee , `` appearance of gauge structure in simple dynamical systems , '' phys .
rev .
lett . * 52 * , 2111 ( 1984 ) .
j. anandan , `` non - adiabatic non - abelian geometric phase , '' phys .
lett .
a * 133 * , 171175 ( 1988 ) .
p.
zanardi and m. rasetti , `` holonomic quantum computation , '' phys .
lett .
a * 264 * , 9499 ( 1999 ) .
l .- m .
duan , j. i. cirac , and p. zoller , `` geometric manipulation of trapped ions for quantum computation , '' science . * 292 * , 1695 - 1697 ( 2001 ) .
s .- l .
zhu and z. d. wang , `` unconventional geometric quantum computation , '' phys .
rev .
lett .
* 91 * , 187902 ( 2003 ) .
x .- d .
zhang , q. h. zhang , and z. d. wang , `` physical implementation of holonomic quantum computation in decoherence - free subspaces with trapped ions , '' phys .
rev .
a * 74 * , 34302 ( 2006 ) .
j. t. thomas , m. lababidi , and m. tian , `` robustness of single - qubit geometric gate against systematic error , '' phys .
rev .
a * 84 * , 042335 ( 2011 ) .
x .- b .
wang and k. matsumoto , `` nonadiabatic conditional geometric phase shift with nmr , '' phys
. rev .
lett .
* 87 * , 097901 ( 2001 ) .
s .- l .
zhu and z. d. wang , `` implementation of universal quantum gates based on nonadiabatic geometric phases , '' phys .
rev .
lett . * 89 * , 097902 ( 2002 ) .
e. sjqvist , d. m. tong , l. mauritz andersson , b. hessmo , m. johansson , and k. singh , `` non - adiabatic holonomic quantum computation , '' new j. phys . * 14 * , 103035 ( 2012 )
. a. a. abdumalikov , j. m. fink , k. juliusson , m. pechal , s. berger , a. wallraff , and s. filipp , `` experimental realization of non - abelian non - adiabatic geometric gates , '' nature * 496 * , 482485 ( 2013 ) . g. feng , g. xu , and g. long , `` experimental realization of nonadiabatic holonomic quantum computation , '' phys .
. lett . *
110 * , 190501 ( 2013 ) .
l .- m .
duan and g .- c .
guo , `` preserving coherence in quantum computation by pairing quantum bits , '' phys .
rev .
lett . * 79 * , 1953 ( 1997 ) .
p.
zanardi and m. rasetti , `` noiseless quantum codes , '' phys .
rev .
lett . * 79 * , 3306 ( 1997 ) .
d. a. lidar , i. l. chuang , and k. b. whaley , `` decoherence - free subspaces for quantum computation , '' phys . rev
. lett . * 81 * , 2594 ( 1998 ) .
d. kielpinski , v. meyer , m. a. rowe , c. a. sackett , w. m. itano , c. monroe , and d. j. wineland , `` a decoherence - free quantum memory using trapped ions , '' science .
* 291 * , 10131015 ( 2001 ) .
j. e. ollerenshaw , d. a. lidar , and l. e. kay , `` magnetic resonance realization of decoherence - free quantum computation , '' phys .
rev .
lett .
* 91 * , 217904 ( 2003 ) .
m. bourennane , m. eibl , s. gaertner , and c. kurtsiefer , `` decoherence - free quantum information processing with four - photon entangled states , '' phys . rev .
lett . * 92 * , 107901 ( 2004 ) .
l .- a .
wu , p. zanardi , and d. a. lidar , `` holonomic quantum computation in decoherence - free subspaces , '' phys .
rev .
lett . * 95 * , 130501 ( 2005 ) .
g. f. xu , j. zhang , d. m. tong , e. sjqvist , and l. c. kwek , `` nonadiabatic holonomic quantum computation in decoherence - free subspaces , '' phys . rev .
lett . * 109 * , 170501 ( 2012 ) .
z .- t .
liang , y .- x .
du , w. huang , z .- y .
xue , and h. yan , `` nonadiabatic holonomic quantum computation in decoherence - free subspaces with trapped ions , '' phys .
rev .
a * 89 * , 062312 ( 2014 ) .
j. zhang , l .- c .
kwek , e. sjqvist , d. m. tong , and p. zanardi , `` quantum computation in noiseless subsystems with fast non - abelian holonomies , '' phys .
rev . a * 89 * , 042302 ( 2014 ) .
g .- f . xu and g .- l .
long , `` universal nonadiabatic geometric gates in two - qubit decoherence - free subspaces , '' sci
. rep .
* 4 * , 6814 ( 2014 ) .
m. stoneham , `` is a room - temperature , solid - state quantum computer mere fantasy , '' physics . * 2 * , 34 ( 2009 ) .
f. shi , x. rong , n. xu , y. wang , j. wu , b. chong , x. peng , j. kniepert , r .- s .
schoenfeld , w. harneit , m. feng , and j. du , `` room - temperature implementation of the deutsch - jozsa algorithm with a single electronic spin in diamond , '' phys .
rev .
lett . * 105 * , 040504 ( 2010 ) .
f. jelezko , t. gaebel , i. popa , a. gruber , and j.wrachtrup , `` observation of coherent oscillations in a single electron spin , '' phys .
rev .
lett . * 92 * , 076401 ( 2004 ) .
l. jiang , j. s. hodges , j. r. maze , p. maurer , j. m. taylor , d. cory , p. r. hemmer , r. l. walsworth , a. yacoby , a. s. zibrov , and m. d. lukin , `` repetitive readout of a single electronic spin via quantum logic with nuclear spin ancillae , '' science .
* 326 * , 267272 ( 2009 ) .
y .- s .
park , a. k. cook , and h. wang , `` cavity qed with diamond nanocrystals and silica microspheres , '' nano lett .
* 6 * , 20752079 ( 2006 ) .
p. e. barclay , k .-
m . c. fu , c. santori , and r. g. beausoleil , `` chip - based microcavities coupled to nitrogen - vacancy centers in single crystal diamond , '' appl .
. lett . * 95 * , 191115 ( 2009 ) .
n. b. manson , j. p. harrison , and m. j. sellars , `` nitrogen - vacancy center in diamond : model of the electronic structure and associated dynamics , '' phys .
rev .
b * 74 * , 104303 ( 2006 ) . c. santori , p. tamarat , p. neumann , j. wrachtrup , d. fattal , r. g. beausoleil , j. rabeau , p. olivero , a. greentree , s. prawer , f. jelezko , and p. hemmer , `` coherent population trapping of single spins in diamond under optical excitation , '' phys
rev .
lett . * 97 * , 247401 ( 2006 ) .
p. tamarat , t. gaebel , j. r. rabeau , m. khan , a. d. greentree , h. wilson , l. c. l. hollenberg , and s. prawer , `` stark shift control of single optical centers in diamond , '' phys .
rev .
lett . * 97 * , 083002 ( 2006 ) .
v. m. acosta , c. santori , a. faraon , z. huang , k .-
m . c. fu , a. stacey , d. a. simpson , k. ganesan , s. tomljenovic - hanic , a. d. greentree , s. prawer , and r. g. beausoleil , `` dynamic stabilization of the optical resonances of single nitrogen - vacancy centers in diamond , '' phys .
rev .
lett . * 108 * , 206401 ( 2012 ) .
g. lindblad , `` on the generators of quantum dynamical semigroups , '' commun .
math .
phys . * 48 * , 119 - 130 ( 1976 ) .
p. neumann , r. kolesov , v. jacques , j. beck , j. tisler , a. batalov , l. rogers , n. b. manson , g. balasubramanian , f. jelezko , and j. wrachtrup , `` excited - state spectroscopy of single nv defects in diamond using optically detected magnetic resonance , '' new j. phys .
* 11 * , 013017 ( 2009 ) .
s. clark and a. parkins , `` entanglement and entropy engineering of atomic two - qubit states , '' phys .
rev .
lett . * 90 * , 047905 ( 2003 ) .
d. w. vernooy , v. s. ilchenko , h. mabuchi , e. w. streed , and h. j. kimble , `` high - q measurements of fused - silica microspheres in the near infrared , '' opt . lett . * 23 * , 247 - 249 ( 1998 ) . |
very long baseline interferometry ( vlbi ) observations hold the potential to achieve the highest astrometric accuracy in astronomy , provided that the fringe phase observable can be calibrated @xcite .
the development of advanced phase referencing ( pr ) techniques to compensate for the tropospheric propagation errors have led to routinely achieving micro - arcsecond ( @xmath2as ) astrometry at frequencies between ca .
10 and a few tens of ghz using alternating observations of the target and a nearby calibrator , which can be up to a few degrees away @xcite . the increasingly fast tropospheric fluctuations at higher frequencies set an upper threshold for application of pr techniques at @xmath343 ghz ( with but a single case at 86 ghz @xcite ) .
more recently , the development of phase calibration techniques using ( nearly ) simultaneous observations at multiple mm - wavelengths , that is source frequency phase referencing ( sfpr ) @xcite and multi frequency phase referencing ( mfpr ) @xcite , have extended the capability to measure @xmath2as astrometry up to mm - wavelengths .
this capability for accurate astrometry has resulted in a wide applicability to many scientific problems ( * ? ? ? * and references therein ) .
nevertheless the application of these advanced pr techniques to relatively low frequencies @xmath4 ghz are hindered by the contribution of ionospheric propagation effects , increasingly dominant at lower frequencies , which have a different nature to the tropospheric effects .
the unpredictability of the spatial irregularities in the plasma density in the ionosphere introduces differential path variations between the sky directions of the two sources , and propagate into systematic position errors even for small source separations .
in addition , at the lowest frequencies , the temporal variations are also an issue .
these both are responsible for degrading the positional accuracy achieved with this technique and , eventually , prevent the phase connection process and the use of conventional phase referencing . therefore , a new strategy is required to overcome the limitations imposed by the ionospheric propagation medium and to reach the full potential of the instruments working on these spectral regimes , such as the square kilometer array ( ska ) that will have vlbi capability between 0.3 and 14 ghz . in general , observations which involve more than one calibrator have demonstrated advantages for astrometric vlbi at low frequencies @xcite .
the alternative is the unusual configuration when a target and a strong calibrator lie within the field - of - view ( fov ) of the vlbi antennas ( i.e. an ` in - beam ' calibrator ) , and thus can be observed simultaneously @xcite .
a useful variation of this combines the observations of a weak ` in - beam ' calibrator source and nodding to a strong more distant calibrator @xcite .
the observations of the strong calibrator are used to remove the first - order atmospheric effects ; then the observations of the weak ` in - beam ' source , which is observed along with the target source , provide further adjustments of the spatial and temporal fluctuations , with longer coherence times .
the results obtained with this approach are positive , however its widespread application is still limited by sensitivity .
another useful approach is when there are two calibrators aligned with , but on opposite sides of , the target @xcite . during the observations the telescopes alternate every few minutes between the three sources , and in the analysis successive scans on the calibrators
are used for the spatial and temporal interpolation to the enclosed position and scan time of the target source . the rare source configuration required for this approach to work results in limited applicability , and
the calibration time overhead is large . in this paper we present results from the multiview technique which , by deriving 2d phase screens from observations of three or more calibrators , achieves a superior mitigation of atmospheric errors that results in increased precision astrometry , along with wide applicability by relaxing the constraints on the angular separation up to few degrees , and does not require alignment of sources .
the scope of application is for the low frequency regime where the perfomance of pr is degraded due to the spatial structure of the ionospheric dominant errors .
it is a development of the `` cluster cluster '' vlbi technique , which allowed simultaneous observations of a target and multiple calibrators around it by replacing single telescopes by sites with multiple elements @xcite .
the ability of the `` cluster cluster '' technique to address the ionospheric effects has been demonstrated with joint observations between connected interferometer arrays at 1.6 ghz of a target and three calibrator sources @xcite . despite these benefits its use has been limited by the shortage of compatible observing sites , and the complexity in its implementation .
we revisit this technique in the light of the next generation of instruments for low frequency observations that will become operational in the course of the next decade .
these have the multi - beam capability as an `` in - built '' feature , such as the australian ska pathfinder ( askap ) in the near future , and ska in the longer term .
we believe that the implementation of multiview techniques will enhance the performance of vlbi observations , by providing higher precision astrometric measurements of many targets at low frequencies . in this paper
we describe the observations in section [ sec : obs ] ; the basis of the multiview technique in section [ sec : meth ] ; section [ sec : res ] presents a demonstration and quantifies its astrometric capabilities , along with a comparative study with outcomes from conventional pr techniques for a range of target calibrator angular separations , including in - beam phase referencing , using vlba observations at 1.6 ghz ; section [ sec : disc ] are discussions and conclusions .
in an effort to demonstrate the improvements by multiview calibration over conventional techniques , we conducted two epochs of observations with the nrao very long baseline array ( vlba ) separated by one month , on 2015 june 8 ( epoch i , obs . i d : bo047a7 ) and july 7 ( epoch ii , obs .
i d : bo047a4 ) , at 1.6 ghz .
both epochs of observations used identical setups with a duration of @xmath5 hours .
table [ table : sources ] lists the source names and coordinates and fig . [
fig : obs_setup ] shows the distribution in the sky .
.all observed sources : four quasars and one oh - maser line source ( also known as wx psc ) .
columns 1 and 2 are the source names and aliases used through this paper , respectively .
columns 3 and 4 are the right ascension and declination coordinates used at the correlator .
columns 5 and 6 are the corresponding position errors , if available .
column 7 is the peak brightness at 1.6 ghz measured from our observations . for quasars ,
all quantities are from the vlba radio fundamental catalog ( l. petrov , solution rfc_2015b ( unpublished ) available on the web at http://asrtogeo.org/vlbi/solutions/rfc_2015b ) .
for the oh - maser , position comes from the simbad astronomical database ( http://adsabs.harvard.edu/abs/2000a%26as..143....9w ) [ cols="^,^,<,<,<,^,^ " , ] l| c c |c analysis i d . & & @xmath6 ( mas ) + & epoch i & epoch ii & @xmath7 + mv@xmath8 & 0.17 & 0.14 & 0.1 + pr@xmath9 & 0.19 & 0.11 & 2.7 + pr@xmath10 & 0.42 & 0.22 & 3.9 + pr@xmath11 & 0.75 & 0.44 & 11.2 + ( red squares ) , @xmath12 ( green triangles ) and @xmath13 ( blue crosses , epoch ii only ) away , on a subgroup of baselines , in epoch i ( left ) and epoch ii ( right ) .
[ fig : vis],scaledwidth=90.0% ] , pr@xmath10 and pr@xmath11 , for epoch i ( upper row ) and epoch ii ( lower row ) .
the lowest intensity contour in all images are at 3-@xmath14 level of the mv map and doubling thereafter .
the images have been restored with the same beam : @xmath15 mas with pa=@xmath16 .
see table [ tab : errors ] for values of peak brightness , rms noise and astrometric offsets in the images .
[ fig : c4.maps],scaledwidth=80.0% ] , pr@xmath10 and pr@xmath11 analyses ( see table [ tab : analyses ] for description ) , from the observations of quasars at the two epochs , with respect to those in the vlba radio fundamental catalog rfc_2015b .
the size of the plotted symbols corresponds to the estimated thermal noise error in each case ( see table [ table : errors2 ] ) .
the labels describe the analysis i d . and epoch of observations .
_ b ) _ zoom for mv astrometric solutions .
the error bars are the thermal noise errors .
both epochs agree within the error bars .
_ c ) _ solid line shows the corresponding repeatability astrometric errors versus the angular separation between target and calibrator for pr analysis , and for an effective @xmath17 separation for mv . filled and empty symbols show the flux fractional recovery quantity versus angular separation for mv ( diamond ) , pr@xmath9 ( triangle ) , pr@xmath10 ( circle ) , and pr@xmath11 ( square ) analyses , for epoch i ( empty ) and epoch ii ( filled ) . [
fig : astro],scaledwidth=90.0% ] .
the lowest intensity contour is the 3-@xmath14 level and doubling thereafter .
restoring beam is @xmath15 mas with pa=@xmath16 .
[ fig : selfcal.maps],scaledwidth=95.0% ] at epoch ii , with respect to epoch i. the estimated thermal noise errors are 0.5 mas and are not displayed .
the labels describe the analysis i d . and epoch of observations .
also shown is the expected apparent motion of the oh - maser source between both epochs , due to the proper motion and parallax ( see text ) .
_ b ) _ same as _ left _ , after correcting for the apparent motion of oh - maser source due to the proper motion and parallax.[fig : line],scaledwidth=90.0% ]
_ demonstration of multiview high precision astrometry at low frequencies _
+ the ionospheric propagation effects are the main limitation to routinely achieving high precision astrometry at frequencies @xmath18 ghz , using state - of - the - art phase referencing methods developed for higher frequencies .
this is due to the distinct direction dependent signature , which limits high precision measurements to cases when there is a suitable very close calibrator ca .
arcmins away . our combination of multiple calibrators around the target results in a significant reduction of the systematic astrometric errors , from the mitigation of the spatial structure effects , by using a 2-d linear ( spatial ) interpolation to estimate the calibration along the direction of the target source .
+ we have presented an empirical demonstration of the superior mitigation of mv along with a comparative study with phase referencing analysis using a single source at a range of angular separations , including `` in - beam '' phase referencing .
all the analysis have been carried out using the same vlba observations at 1.6 ghz .
we have used the repeatability between two epochs of observations to provide an empirical estimate of the systematic astrometric errors , which are otherwise very difficult to quantify .
we achieve high precision mv astrometry of ca .
@xmath19as in a single epoch of observations of quasars with calibrators at @xmath20 and larger angular separations , effectively reaching the thermal noise limit of the observations .
this corresponds to more than an order of magnitude improvement with respect to the precision achieved using pr with a single calibrator @xmath20 away , which is ca .
@xmath21 mas , due to the residual systematic errors in the analysis .
this underlines the importance of correcting for the spatial structure of the ionospheric residuals .
the comparative improvement can be interpreted as mv compensation being equivalent to that from pr with a single calibrator ca .
10 times closer , in this case ca .
@xmath22 away , assuming a linear dependence between astrometric precision and target calibrator angular separation . also , our results indicate that there is a common limiting factor for accuracy and precision , namely the residual ionospheric propagation errors , and both are improved by a quality calibration as provided by mv . we have also demonstrated the performance in the weak source case using the observations of the oh - maser line source , and compared mv using @xmath20 , @xmath12 and @xmath13 angular separations with ` in beam ' pr using a calibrator @xmath23 away .
the repeatability errors are larger , as expected from lower snr , but interestingly keep the same corresponding relative astrometric signature found in the analysis of the quasars .
that is mv@xmath24 calibration is a factor of two better , with respect to pr@xmath25 using a calibrator five times closer to the target .
therefore we have demonstrated the benefits of using multiple calibrators , in our case with an improvement of more than one order of magnitude in astrometric precision , reaching the thermal limit of the observations , of @xmath26as . in general
we expect mv to be relevant for observations in the frequency regime where the ionospheric effects continue to be the dominant source of errors , that is , in observations up to ca
we conclude that greater improvements are expected from increased sensitivity , and faster duty cycles , with maximum benefits from simultaneous observations and closer source distribution , to minimize the non - linear deviations of the actual ionospheric spatial structure above each antenna from a planar surface .
_ mv in the context of ska and multi - beam instruments _
+ precise astrometric capability is of great importance in the ska era .
it is a ska goal to achieve @xmath27as astrometric accuracy at a single epoch of observations at @xmath281.4 ghz @xcite .
the high sensitivity and long baselines of ska vlbi observations will result in a much reduced thermal noise level and high spatial resolution .
therefore this goal is achievable as long as a sufficiently accurate ionospheric phase calibration strategy is in place . for a single calibrator source and pr techniques ,
the required angular separation to the target would be ca .
1 arcmin @xcite .
this puts a very tight limit on the number of available calibrator sources , even at the ska sensitivities @xcite .
this constraint on the angular separation can be significantly relaxed , by using multiple calibrator sources and mv techniques , as suggested by the demonstration presented in this paper and our previous simulation studies .
additionally , the multi - beam capability of ska will allow for simultaneous observations of all sources and therefore eliminate the errors arising from short term phase fluctuations , which result in a reduction of the coherence losses ( i.e. characterized by ffr quantity ) and the thermal noise errors , while improving the overall performance .
this applies to other instruments with multibeam systems such as askap and westerbork synthesis radio telescope with apertif , among others .
our demonstration includes three calibrators ; note that the more calibrators the better , as this will allow the most accurate reconstruction of the atmospheric effects .
therefore we expect mv can deliver the goal of @xmath29as astrometry for many targets with ska .
this paper used astrometric observations of a group of quasars in the role of targets and calibrators and an oh maser source in the ground state at 1.6 ghz .
scientific applications using such a group of sources applied to studies with the ska of the nearby universe , including the milky way galaxy and the local group of galaxies are described in @xcite . although that case - study report discussed the scientific applications conservatively based on only ` in - beam ' pr astrometry , our results indicate that mv would provide further benefits .
our empirically estimated mv astrometric accuracy of ca .
@xmath30as at 1.6 ghz with vlba observations , using calibrators more than ca .
@xmath20 away is at the state - of - the - art , only comparable with ` in - beam ' phase referencing observations with a calibrator ca .
10s of arcminutes away @xcite .
this improvement is expected to continue to apply at all angular scales .
hence , using three calibrators within the ska - mid antenna beams and mv will result in a further increase by one order of magnitude of the astrometric precision , extrapolating from our comparative study .
in general , allowing for larger angular separations makes it possible to select good calibrator sources , which are fundamental for multi - epoch studies . with the higher probabilities of finding suitable calibrators
the general applicability is also highly increased . in some cases , such as pulsar observations in the galactic plane
, it might even be desirable to use calibrators out of the plane to reduce the effect of scattering
. this would be possible using mv .
high precision astrometry observations of methanol masers at 6.7 ghz holds the prospect to contribute to the successful program for 3-d mapping of our galaxy , as a complement to the precise water maser measurements @xcite .
however the advanced pr strategies used at 22 ghz , in the tropospheric dominated regime , fail to provide high precision astrometry at 6.7 ghz , as this is in the ionospheric dominated regime . therefore mv with fast source switching between sources , or simultaneous observations if possible , provides a strategy for superior calibration of the tropospheric and ionospheric errors resulting in precise astrometry .
+ * acknowledgements * the vlba is operated by national radio astronomy observatory and is a facility of the national science foundation operated under cooperative agreement with associated universities inc .. go and hi have been supported by the jsps bilateral collaboration program and kakenhi programs 25610043 and 16h02167 .
mr , rd , go and hi acknowledge support from dfat grant ajf-124 .
this research has made use of the simbad database , operated at cds , strasbourg , france . , s. , rioja , m. j. , dodson , r. , smirnov , o. , & guirado , j. c. 2010 , in 10th european vlbi network symposium and evn users meeting : vlbi and the new generation of radio arrays , manchester , proceedings of science , vol . | the arrival of the square kilometer array ( ska ) will revitalise all aspects of very long baseline interferometry ( vlbi ) astronomy at the lower frequencies . in the last decade
there have been huge strides towards routinely achieving high precision vlbi astrometry at frequencies dominated by the tropospheric contributions , most notably at 22ghz , using advanced phase referencing techniques . nevertheless to increase the capability for high precision astrometric measurements at low radio frequencies ( @xmath08 ghz ) an effective calibration strategy of the systematic ionospheric propagation effects that is widely applicable is required .
observations at low frequencies are dominated by distinct direction dependent ionospheric propagation errors , which place a very tight limit on the angular separation of a suitable phase referencing calibrator .
the multiview technique holds the key to the compensation of atmospheric spatial - structure errors , by using observations of multiple calibrators and 2-d interpolation . in this paper
we present the first demonstration of the power of multiview using three calibrators , several degrees from the target , along with a comparative study of the astrometric accuracy between multiview and phase - referencing techniques .
multiview calibration provides an order of magnitude improvement in astrometry with respect to conventional phase referencing , achieving @xmath1micro - arcseconds astrometry errors in a single epoch of observations , effectively reaching the thermal noise limit .
multiview will achieve its full potential with the enhanced sensitivity and multibeam capabilities of ska and the pathfinders , which will enable simultaneous observations of the target and calibrators .
our demonstration indicates that the 10 micro - arcseconds goal of astrometry at 1.6ghz using vlbi with ska is feasible using the multiview technique . |
the independence sampler is the incorporation of rejection sampling within an mcmc framework .
the rejection sampler obtains samples from a random variable , @xmath0 , with probability density function @xmath1 by first proposing a candidate value @xmath2 from a random variable , @xmath3 , with probability density function @xmath4 , and secondly accepting @xmath2 as a sample from @xmath0 with probability @xmath5 , where @xmath6 .
otherwise @xmath2 is rejected , see @xcite , page 60 .
the success of the rejection sampler depends upon making a good choice of @xmath4 such that @xmath7 is small and that @xmath4 is straightforward to sample from .
the mcmc independence sampler is the modification of the above where a markov chain @xmath8 is constructed with at iteration @xmath9 , a candidate @xmath2 proposed from @xmath3 and if accepted @xmath10 is set equal to @xmath2 .
otherwise @xmath11 . the rejection sampler , and
consequently , the independence sampler can usually be implemented in a straightforward and efficient manner for low dimensional ( target ) distributions but as the dimension of @xmath0 increases it becomes increasingly more challenging to obtain a good choice of @xmath4 .
therefore the independence sampler is rarely used as an mcmc algorithm in its own right but instead independence sampler moves are often incorporated within metropolis - within - gibbs to effectively update low dimensional subsets of @xmath0 , see @xcite , page 15 .
the main focus for independence samplers has been to choose the proposal density @xmath4 so as to have an acceptance probability as close to 1 as possible . whilst this makes intuitive sense ,
the aim of the current paper is to challenge the idea of aiming for an acceptance probability as close to 1 as possible within the context of using independence samplers for updating augmented data in mcmc algorithms .
specifically , we are interested in the bayesian statistical problem of obtaining samples from the posterior distribution of the parameters @xmath12 of a model given data @xmath13 , @xmath14 in the case where the likelihood , @xmath15 is intractable .
we assume that given augmented data @xmath16 , @xmath17 is tractable and an mcmc algorithm can be constructed to obtain samples from the joint posterior of @xmath12 and @xmath16 , @xmath18 .
then it is natural to construct an mcmc algorithm which alternates between updating the parameters and the augmented data as follows : 1 .
update @xmath12 given @xmath13 and @xmath16 .
_ i.e. _ use @xmath19 .
2 . update @xmath16 given @xmath13 and @xmath12 .
_ i.e. _ use @xmath20 .
our focus is the use of independence samplers to update @xmath16 given @xmath13 and @xmath12 . for updating augmented data
a natural independence sampler often presents itself .
for example , in an epidemic modelling context where @xmath13 denotes the removal times of infected individuals , @xmath12 denotes the infection and infectious period parameters and @xmath16 denotes the infection times of individuals , a natural candidate for the infection time of individual @xmath21 who is removed at time @xmath22 is @xmath23 , where @xmath24 denotes the infectious period distribution , see @xcite , @xcite and section [ ss : homo ] . for non - centered parameterisations , @xcite , we can often denote @xmath25 as a deterministic function @xmath26 with @xmath27 easy to compute , where @xmath28 is a vector of independent and identically distributed uniform random variables , see @xcite and section [ ss : bdm ] . then to update @xmath29 we can propose a new value from @xmath30 .
the dimension of the augmented data , @xmath16 , can be orders of magnitude higher than @xmath12 and @xmath13 , so updating one component of @xmath16 at a time can be prohibitive .
therefore we seek generic guidelines for updating multiple components of @xmath16 at a time and optimising the performance of the resulting independent sampler . specifically , this work formalises findings in @xcite and @xcite in using the independence sampler for data augmentation giving simple guidelines for producing close to optimal independence samplers .
the guidelines obtained are similar to those given in @xcite for the random walk metropolis algorithm and comparisons with the random walk metropolis algorithm are made .
the paper is structured as follows . in section [ s : theo ] , we study the properties of the independence sampler for independent and identically distributed product densities @xmath31 .
this idealised scenario mimics the set up in @xcite where optimal scaling of the random walk metropolis algorithm was first explored and as in @xcite allows us to get a handle on understanding the key factors in optimising the independence sampler .
in particular , we show that the optimal number of components , @xmath32 , of @xmath13 to update , is the @xmath32 which maximises the mean number of components per move . in the case where this optimal @xmath32 is large
this corresponds to a mean acceptance rate of approximately @xmath33 .
thus there is a somewhat surprising link with the optimal scaling of the random walk metropolis algorithm , @xcite with which we make comparison and highlight the benefits of the independence sampler . in section [ s : ex ]
, we explore the optimal performance of the independence sampler for increasingly complex problems . in section [
s : ex : intro ] , we study product gaussian target densities with gaussian and @xmath9-distribution proposals demonstrating the optimal scaling results obtained in section [ s : theo ] . in sections [ ss :
homo ] and [ ss : bdm ] we apply the independence sampler to two epidemic models , the classic homogeneously mixing sir epidemic model , @xcite and @xcite and a birth - death - mutation ( bdm ) model for an emerging , evolving disease , @xcite and @xcite . in section [ ss :
homo ] , we show that for the homogeneously mixing sir epidemic model updating a proportion of the infection times so as to obtain a mean acceptance rate of approximately @xmath33 is optimal .
this demonstrates that as observed with the random walk metropolis algorithm the findings of section [ s : theo ] are informative in designing independence samplers beyond the limited confines of product densities . for the bdm model in section [ ss : bdm ] the findings are somewhat different with a lower optimal mean acceptance rate corresponding to large scale data augmentation
. finally , in section [ s : conc ] , we make some concluding remarks highlighting the possible benefits of the independence sampler over random walk metropolis for large scale data augmentation and the differences seen between the two epidemic models in sections [ ss : homo ] and [ ss : bdm ] .
in this section we consider the theoretical properties of the independence sampler for the special case where @xmath34 , a product of independent and identically distributed univariate densities , @xmath35 .
the main focus is on the asymptotic behaviour as the number of components , @xmath36 mirroring analysis performed in @xcite for the random walk metropolis algorithm .
the aim is to characterise the optimal performance of the independence sampler in terms of the number of components to update and to draw interesting comparisons of similarities and differences with the random walk metropolis algorithm . for the independence sampler we propose to select uniformly at random @xmath32 components @xmath37 from @xmath38 to update . for @xmath39 , @xmath40
is drawn from @xmath3 with probability density function @xmath41 , whilst for @xmath42 , @xmath43 .
therefore the acceptance probability for the proposed move from @xmath44 to @xmath45 is @xmath46 for @xmath47 and @xmath48 , let @xmath49 denote the position of the markov chain after @xmath9 iterations . as in @xcite , we assume that the markov chain is initiated with @xmath50 drawn from @xmath51 and thus for all @xmath52 , @xmath53 .
the independent and identically distributed nature of the stationary and proposal distributions means that as in @xcite it suffices to focus on the behaviour and performance of the independence sampler on the first component only .
specifically , for @xmath52 , letting @xmath54}$ ] we show that for fixed @xmath32 , as @xmath36 , the movement in the first component of @xmath55 converges to a markov jump process with jumps governed by @xmath56 and @xmath4 .
let @xmath57 , then for the independence sampler to be well - behaved we require that @xmath58 , see @xcite and we make this assumption throughout . for a move to occur in the first component we must propose to move the first component and @xmath59 other components from @xmath60 .
let @xmath61 be a random sample from @xmath62 with @xmath63 , where @xmath64 .
define @xmath65 , @xmath66 and @xmath67 in the obvious fashion .
then we define @xmath68 \nonumber \\ & = & { \mathbb{e}}_{\mathbf{y}^{n- } , \mathbf{j}_{k-1 } } \left [ 1 \wedge \frac{\omega ( y)}{\omega ( x_1 ) } \prod_{i=1}^{k-1 } \frac{\omega ( y_{j_i})}{\omega ( x_{j_i } ) } \right],\end{aligned}\ ] ] where @xmath69 .
a useful observation is that the proposed values @xmath70 are independent of @xmath44 .
let @xmath71 $ ] and let @xmath72 we have the following lemma which mirrors @xcite , lemma 2.1 , which states that with sufficiently high probability we can focus upon @xmath73}^n$ ] ( @xmath74 ) contained in @xmath75 .
the proof of lemma [ lemma1 ] is given in appendix [ app : a ] . [ section ] [ lemma1 ] for @xmath76 , @xmath77 we are now in position to state and prove the main result of this section , theorem [ thm ] .
[ theorem]theorem [ thm ] for @xmath78 , let @xmath79 , then @xmath80 where @xmath81 is a markov jump process with infinitesimal generator @xmath82 for any @xmath83 function @xmath84 . * proof .
* we begin by defining the ( discrete time ) generator of @xmath85 , @xmath86 , \end{aligned}\ ] ] where @xmath84 is any @xmath83 function of the first component .
note that if there is no proposed update in the first component then @xmath87 . therefore letting @xmath88
if there is a proposed update of the first component and 0 otherwise , we have that @xmath89 \nonumber \\ & = & n \times \frac{k}{n } \times { \mathbb{e}}\left[\left . \ { h ( \mathbf{y}^n ) - h ( \mathbf{x}^n ) \ } \left\ { 1 \wedge \frac{\pi_n ( \mathbf{y}^n)}{\pi_n ( \mathbf{x}^n ) } \right\ } \right| \chi^n = 1 \right ] \nonumber \\ & = & k { \mathbb{e}}_{y_1 } \left [ ( h ( y_1 ) - h ( x_1 ) ) { \mathbb{e}}_{\mathbf{y}^{n- } , \mathbf{j}_{k-1 } } \left [ 1 \wedge \frac{\omega ( y_1)}{\omega ( x_1 ) } \prod_{j=1}^{k-1 } \frac{\omega ( y_{j_j})}{\omega ( x_{j_j } ) } \right ] \right ] . \nonumber \\ \end{aligned}\ ] ] we compare @xmath90 with the generator @xmath91 defined in for the limiting jump process .
now by , for all @xmath92 and @xmath93 , @xmath94 - h^\ast ( y , x ) \right ) \ , dy \right| \nonumber \\ & = & \left| \int \ { h ( y ) - h ( x_1 ) \ } q(y ) \left ( h(y , x_1 , \mathbf{x}^{n- } ) - h^\ast ( y , x ) \right ) \ , dy \right| \nonumber \\ & \leq & 2 \sup_z | h(z)| \int q(y ) \left ( h(y , \mathbf{x}^n ) - h^\ast ( y , x ) \right ) \ , dy \nonumber \\ & \leq & 2 \sup_z | h(z)| n^{-\frac{1}{8 } } \rightarrow 0 \hspace{0.5 cm } \mbox{as } { n \rightarrow \infty}. \end{aligned}\ ] ] hence , @xmath95 the theorem follows along identical lines to @xcite , theorem 1.1 .
since @xmath83 separates points ( see , @xcite , page 113 ) , the theorem follows from and lemma [ lemma1 ] by corollary 8.7 ( f ) of chapter 4 of @xcite .
@xmath96 we proceed by discussing properties of the limiting jump process .
let @xmath97 where @xmath98 and @xmath99 .
then @xmath100 $ ] denotes the mean acceptance probability , in stationarity , of a proposed move and @xmath101 $ ] denotes the corresponding mean number of components updated .
moreover , @xmath101 $ ] denotes the mean number of jumps , per unit time , of the limiting jump process , and hence , we seek @xmath32 which maximises @xmath101 $ ] .
the distribution of @xmath102 depends largely on the _ closeness _ of the target ( @xmath1 ) and proposal ( @xmath4 ) distributions with @xmath103 if for all @xmath104 , @xmath105 .
let @xmath106 , then @xmath107 where the @xmath108 are independent and identically distributed . note that @xmath109 = -d ( q \| f)$ ] and @xmath110 = d ( f \| q)$ ] , where for two probability density functions @xmath111 and @xmath112
, @xmath113 is the kullback - leibler divergence .
hence , @xmath114 = - \ { d ( q \| f ) + d ( f \| q)\ } = -i , \mbox { say } , \end{aligned}\ ] ] which makes explicit the role played by the _ closeness _ of the two densities . it should be noted that @xmath115 if there exists @xmath104 such that @xmath116 and @xmath117 , in such cases efficient independence sampling may still exist , for example , @xmath118 and @xmath119 for small , positive @xmath120 . for finite @xmath121
, it follows from by the central limit theorem that for large @xmath32 , @xmath122 is approximately gaussian with mean @xmath123 $ ] and variance @xmath124 , say .
now if @xmath121 is small , which will be the case where the central limit theorem is relevant , then @xmath125 .
moreover , if @xmath126 where @xmath127 is small , then it is straightforward to show that @xmath128 and that @xmath129 .
thus for @xmath32 large , with @xmath130 , we have by @xcite , proposition 2.4 , that @xmath131 & \approx & k { \mathbb{e } } [ 1 \wedge \exp ( v_k^\ast ) ] \nonumber \\ & = & k \times \left\{\phi \left ( - \frac{k i}{\sqrt{k j } } \right ) + \exp \left ( - ki + \frac{kj}{2 } \right ) \phi \left ( - \sqrt{kj } + \frac{ki}{\sqrt{k j } } \right ) \right\ } \nonumber \\ & \approx & k \times 2 \phi \left ( - \sqrt{\frac{k i}{2 } } \right),\end{aligned}\ ] ] where the latter approximation follows from setting @xmath132 .
replacing @xmath32 by @xmath133 and @xmath121 by @xmath134 in the right hand side of , we obtain @xmath135 , which is the function maximized in @xcite , corollary 1.2 to maximise the optimal scaling of the random walk metropolis algorithm .
the only difference is the form of @xmath121 which here depends upon the kullback - leibler divergence between the target and proposal distribution , whereas in @xcite @xmath136 $ ] and depends upon the _ smoothness _ of @xmath1 .
most importantly , @xmath137 maximises @xmath138 and therefore @xmath32 should be chosen approximately equal to @xmath139 . thus if @xmath121 is small ( there is close agreement between @xmath1 and @xmath4 ) @xmath32 will be large .
moreover , mirroring @xcite , corollary 1.2 , such a @xmath32 corresponds to a mean acceptance probability of ( approximately ) @xmath140 .
thus it is not necessary to compute @xmath121 but instead suffices to monitor the mean acceptance probability .
this will be shown to be a useful guiding principle in the examples below .
however , it should be noted that scenarios exist , see section [ ss : homo ] below , where the acceptance rate is above ( below ) 0.234 for all @xmath32 , in such cases it is optimal to choose @xmath141 @xmath142 . returning to optimising the independence sampler in the case @xmath118 and @xmath119
, it is straightforward to show that the probability a proposed move is accepted is @xmath143 . optimising the function @xmath144 gives @xmath145 , and hence for small @xmath120 , @xmath146
thus as @xmath147 , the optimal acceptance probability ( @xmath148 ) converges to @xmath149 .
therefore non - trivial asymptotic acceptance probabilities can exist in the case @xmath150 and typically these will be different from 0.234 .
a key question is how does the independence sampler compare to the random walk metropolis algorithm .
provided @xmath58 , theorem [ thm ] holds and we have that the mixing of the independence sampler algorithm is @xmath151 , the same order of mixing as for the random walk metropolis algorithm for continuous ( and sufficiently differentiable ) densities .
the mixing of the random walk metropolis algorithm for discontinuous densities is @xmath152 , @xcite whilst modifications such as metropolis adjusted langevin algorithms ( mala ) and hybrid monte carlo ( hmc ) algorithms mix in @xmath153 and @xmath154 iterations , see @xcite and @xcite , respectively , for sufficiently well behaved ( continuous ) target densities .
thus the independence sampler is competitive with the random walk metropolis algorithm and theorem [ thm ] holds under very weak conditions compared with those imposed for corresponding random walk metropolis algorithms .
the similarity of the right hand side of to @xmath138 might suggest that computing @xmath121 for the two algorithms would assist in comparing there performances with smaller @xmath121 the better .
however , the different nature of the moves , global in the independence sampler and local in the random walk metropolis , means that this is not the case . in simulation studies with @xmath155 , @xmath156 and a range of @xmath157 , the independence sampler , with appropriately chosen @xmath32
was found to outperform the optimal random walk metropolis algorithm ( @xmath158 ) for @xmath159 .
thus the independence sampler is competitive with , and often superior to , random walk metropolis , for continuous target densities so long as a reasonable choice of @xmath4 is made , and is clearly preferable for discontinuous target densities which is often the case in real life bayesian problems , see section [ s : ex ] .
in this section we illustrate how large scale independence sampling can be exploited to construct effective mcmc algorithms .
we start with an independent and identically distributed gaussian product density as the target distribution and consider both gaussian and @xmath9-distribution proposals . specifically , we take @xmath31 , where @xmath160 is a standard gaussian density .
the proposal distributions are symmetric about 0 with gaussian proposals @xmath161 , where @xmath162 and @xmath9-distribution proposals @xmath163 @xmath164 .
we conducted a simulation study using 5 gaussian and 5 @xmath9-distribution proposals with @xmath165 and @xmath166 iterations of the mcmc algorithm starting from the stationary distribution . for each proposal distribution we considered 50 choices of @xmath32 , the exact choices of which depended on @xmath121 and
were chosen to give acceptance rates on the full range 0 to 1 .
for the gaussian proposal it is straightforward to show that @xmath167 .
we considered @xmath168 with corresponding @xmath169 .
a key quantity for comparing the independence sampler for different choices of @xmath170 , and hence @xmath121 , is the normalised efficiency .
we define the normalised efficiency for @xmath32 as the mean number of components updated ( @xmath171 acceptance rate ) when proposing to update @xmath32 components divided through by the maximum mean number of components updated for @xmath172 . correspondingly the normalised theoretical efficiency is given by @xmath173 from applying the central limit theorem approximation obtained in section [ s : theo ] . the plots in figure [ fig.gaus ]
show that in all cases the optimal acceptance rate is close to 0.234 with very similar behaviour for the normalised efficiency varying with acceptance rate , even for @xmath174 with @xmath175 .
similar results are obtained in section in @xcite , section 6 for the optimal performance of the random walk metropolis algorithm .
as @xmath176 , @xmath177 and the agreement between the observed normalised efficiency normalised theoretical efficiency becomes very close .
[ h ] for the @xmath9-distribution , @xmath150 for @xmath178 , otherwise @xmath179 - { \mathbb{e } } [ \log ( 1 + y_\nu^2/\nu ) ] \right\},\ ] ] where @xmath155 and @xmath180 .
it is not possible to obtain a closed form analytical expression for @xmath121 but it is straightforward to estimate using monte carlo integration .
we consider @xmath181 with corresponding @xmath182 .
the plots in figure [ fig.t ] show that the optimal acceptance rate is higher than 0.234 for a @xmath9-distribution proposal with an optimal acceptance rate of 0.383 corresponding to @xmath183 for a @xmath184 proposal .
note that this is close to @xmath185 , the optimal acceptance rate of the uniform distributions example given in section [ s : theo ] .
it is worth noting that choosing @xmath32 to obtain an acceptance rate of approximately 0.234 is in general a good approach as only a small loss in efficiency is observed .
as @xmath186 increases the optimal acceptance rate converges towards 0.234 and the normalised efficiency tends towards the theoretical normalised efficiency given by the central limit theorem approximation .
this is further demonstrated in figure [ fig.t2 ] by plotting normalised efficiency against normalised theoretical efficiency .
note that @xmath187 and @xmath188 do not feature on this plot as @xmath189 .
[ h ] in this section we show how the importance sampler can be applied to temporally observed , homogeneously mixing sir epidemic models , @xcite .
we assume that there is a population of size @xmath190 with the disease introduced into the population via a single introductory case .
( the extension to multiple introductory cases is trivial . )
we assume that the disease follows an @xmath191 epidemic model , where initially all individuals , except the introductory case , are susceptible . on becoming infectious ,
an individual is infectious for a given period of time , distributed according to a gamma random variable @xmath192 .
( alternative infectious period distributions can easily be considered . )
whilst infectious , an individual @xmath21 , say , makes infectious contacts at the points of a homogeneous poisson point process with rate @xmath193 with the individual contacted chosen uniformly at random from the entire population .
infectious contacts with susceptible individuals result in the immediate infection of the individual and the start of their infectious period .
infectious contacts with infectives have no effect on the recipient .
suppose that @xmath194 individuals are infected during the course of the epidemic and we are analysing the completed epidemic data .
for each individual , @xmath21 say , infected during the course of the epidemic there will be an infection time , @xmath195 and a removal ( recovery ) time , @xmath196 , which mark the start and end of the infectious period , respectively .
we follow @xcite , @xcite and @xcite in assuming that the removal times , @xmath197 are observed , whilst the infection times @xmath198 are unobserved .
furthermore , we assume that the removal times are ordered such that @xmath199 .
the key interest is in the posterior distribution of @xmath200 and to obtain samples from this distribution imputation of @xmath201 is required .
we use the mcmc algorithm proposed in @xcite , section 3 with the modification that the number of components to be updated is fixed to @xmath202 . as with @xcite ,
the mcmc algorithm is applied to the extensively studied abakaliki smallpox outbreak , @xcite , @xcite , where @xmath203 and @xmath204 .
we considered various fixed values of @xmath205 with optimal @xmath206 and 30 , respectively , based upon the maximised mean number of components updated over 100000 iterations , see figure [ fig.hme_accept_kaccept_k_m030 ] . for @xmath205 ,
the corresponding values of @xmath32 which had acceptance rates closest to @xmath207 were @xmath208 and 29 , respectively . thus choosing @xmath32 so that the acceptance rate is close to @xmath207 is effective in obtaining a close to optimal algorithm . in @xcite , the situation where @xmath209 is assumed to be unknown is also considered with the posterior mean of @xmath209 being @xmath210 . for unknown @xmath209 ,
the acceptance rate is above @xmath207 for all @xmath32 and thus @xmath211 performs optimally .
we can go further in illustrating the usefulness of the theoretical results derived in section [ s : theo ] for choosing @xmath32 . in figure
[ fig.hme_kaccept_accept_m030 ] , we plot the normalised efficiency for @xmath212 , since for @xmath213 , the acceptance rate is always above @xmath207 . also on the plot ( in red )
is the normalised theoretical curve @xmath214 given by against acceptance rate @xmath215 . in a similar fashion to section [ s : ex : intro ] this illustrates that the asymptotic results which are valid as the number of components updated tend to @xmath216 are applicable for small @xmath32 .
[ h ] [ h ] a simulation study was conducted to study the general applicability of the results obtained above for the abakaliki data .
data sets were simulated with @xmath217 , @xmath218 and @xmath219 with @xmath220 chosen to give a mean infectious period of 10 and @xmath193 to give the mean size of a major epidemic outbreak to be 10 . for each @xmath209 , the optimal @xmath32 increases with @xmath190 and vice versa . throughout choosing @xmath32 with acceptance rate closest to @xmath207 produced close to optimal performance .
plots of the normalised efficiency against the acceptance rate showed increasing agreement with the asymptotic theoretical curve as @xmath190 increases . in this section
we consider a birth - death - mutation ( bdm ) model which is applicable to the early stages of a mutating disease .
the model has previously been used by @xcite to analyse data from a tuberculosis outbreak in san francisco in the early 1990s reported in @xcite .
we explore and seek to optimise the performance of the forward simulation mcmc algorithm introduced by @xcite .
note that all the other analyses reported above used abc algorithms . the data consist of the genotypes of 473 bacteria samples sampled from individuals infected with tuberculosis in san francisco during an observational period in 1991 - 92 .
the data are clustered by genotype and summarised in table [ tab : tub ] .
let @xmath221 denote the total number of tuberculosis cases at time @xmath9 .
the data are assumed to be a random sample taken at time @xmath222 , where @xmath223 evolving from @xmath224 .
.[tab : tub ] observed cluster size distribution of tuberculosis bacteria genotype data , @xcite . [ cols="<,>,>,>,>,>,>,>,>,>,>",options="header " , ] the bdm model is a markov process defined as follows .
individuals are classified by ( geno)type .
each individual born into the process has an exponentially distributed lifetime ( infectious period ) with mean @xmath225 .
whilst alive individuals give birth ( infects ) and mutates at the points of independent homogeneous poisson point processes with rates @xmath209 and @xmath226 , respectively .
each individual born inherits the ( geno)type of their parent and all mutations result in the creation of a new , previously unseen ( geno)type ( infinite allele model , @xcite ) .
we reparameterise the model by setting @xmath227 , @xmath228 and @xmath229 , where @xmath230 is the rate at which events occur for an individual , @xmath231 is the probability that the event is a birth ( infection ) and @xmath232 is the probability that the event is a death ( recovery ) .
since the stopping time @xmath222 at which the population is observed only depends upon the number of individuals alive in the population , there is no information in the data about @xmath230 .
thus , without loss of generality , we assume @xmath233 making inference about @xmath234 given the genotype data @xmath13 . in order to construct a tractable likelihood it is necessary to generate the state of the population at time @xmath222 , @xmath235 .
this can be done using a non - centered parameterisation @xcite where the augmented data @xmath236 consist of realisations of @xmath30 with @xmath237 combine with @xmath234 to generate the underlying state of the bdm model at time @xmath222 and @xmath238 is used to estimate the probability of observing @xmath13 .
details of the construction are given in @xcite , section 4 .
the time consuming step of the mcmc algorithm for the bdm model is the simulation of the state of the process using @xmath237 and @xmath234 . in @xcite ,
@xmath234 are updated using random walk metropolis keeping @xmath237 fixed and @xmath237 are updated using an independence sampler , draws from @xmath30 , keeping @xmath234 fixed .
we thus focus on the independence sampler for updating @xmath237 .
note that @xmath238 is updated by a separate independence sampler but this is very fast to implement ( no need to simulate the bdm process ) , and so we do nt comment on this step .
the dimensions of @xmath239 and @xmath240 are the same but vary from iteration to iteration , typically being around 30000 .
to circumvent issues with this @xcite used random vectors of a fixed length @xmath241 with only those elements needed to simulate the process used . in this paper
we also used a fixed length vector updating @xmath32 out of @xmath242 components in @xmath239 and @xmath240 noting that in each simulation not all ( updated ) components will be used . in @xcite , @xmath239 and @xmath240
are broken down into blocks of 50 components with 1 component in each block proposed to be updated .
this amounts to proposing to update @xmath243 values in each iteration of which typically around 600 are used in the simulation . in this paper
we propose to update @xmath32 components each of @xmath239 and @xmath240 , @xmath244 and @xmath245 , where @xmath246 ( @xmath247 ) is a uniformly random sample without replacement from @xmath248 , for the sake of consistency with the updating strategy throughout this paper .
in addition to using different values for @xmath32 , we also examine the performance of the algorithm using @xmath249 and the original @xmath250 , which are all found to be empirically sufficient .
we ran the mcmc algorithm for @xmath251 iterations with the first @xmath252 iterations discarded as burn - in .
the acceptance rate is plotted against @xmath32 for all three values of @xmath242 on the left of figure [ fig.bdm_accept_kaccept_k ] , which is analogous to figure [ fig.hme_accept_kaccept_k_m030 ] , with the mean number of components updated on the right .
the results shown in figures [ fig.bdm_accept_kaccept_k ] demonstrate an interesting departure from those found earlier in the paper with an optimal acceptance rate of @xmath207 .
the mean number of components updated increases with @xmath32 even as the acceptance rate drops below @xmath253 .
however , for both parameters @xmath231 and @xmath232 , the effective sample size levels off at around 3000 for all @xmath254 , which suggests that seeking to optimise the mean number of components updated does not tell the full story in this case .
in this paper we have demonstrated the potential benefits , both theoretical and practical , of the independence sampler over the random walk metropolis algorithm . in particular , we have shown that simple choices of proposal distributions can be used to construct effective independence samplers and that similar considerations to the tuning of the random walk metropolis algorithm are required .
there are a number of points to consider in the wider application of the results derived in section [ s : theo ] and applied in section [ s : ex ] .
firstly , we have not considered the computational time required to update @xmath32 components . in the homogeneously mixing epidemic model ( section [ ss : homo ] ) , and in particular , the bdm model ( section [ ss : bdm ] ) the time taken per iteration was essentially independent of @xmath32 .
however , it is possible for the homogeneously mixing epidemic model by careful updating of the calculation of the likelihood for the time taken per iteration to be smaller for smaller @xmath32 . in such cases the optimal acceptance rate will be larger than @xmath207 and if the time per iteration is proportional to @xmath32 it will be optimal to update a single component at a time . secondly , the theoretical results of section [ s : theo ] for independent and identically distributed product densities are shown to give clear guidance for optimising the independence sampler for the homogeneously mixing epidemic model but not for the bdm model .
the reason for this difference is not immediately obvious but is likely to depend on the relationship of the observed data to the augmented data . for the homogeneously mixing epidemics the local behaviour of @xmath201 is important , for example ensuring @xmath201 is consistent with an epidemic outbreak , whereas for the bdm model it is global properties of @xmath255 , the total numbers of births , deaths and mutations which are most important . for the random walk metropolis algorithm
optimal scaling results differ depending upon whether the acceptance probability depends on local behaviour ( discontinuous product densities , @xcite ) or global behaviour ( continuous product densities , @xcite , elliptically symmetric densities @xcite ) of the proposed moves .
this research was supported by the engineering and physical sciences research council under grant ep / j008443/1 .
we would like to thank an anonymous referee for their careful reading of the paper and suggestions for improving presentation of the findings .
bailey , n.t.j .
( 1975 ) _ the mathematical theory of infectious diseases and its applications .
second edition .
_ griffin , london .
beskos , a. , pillai , n. , roberts , g. , sanz - serna , j - m . and
stuart , a. ( 2013 ) optimal tuning of the hybrid monte carlo algorithm .
_ bernoulli _ , * 19 * , 15011534 .
dellaportas , p. and roberts , g.o .
( 2013 ) an introduction to mcmc .
_ spatial statistics and computational methods _ ( j. mller , eds . ) springer , new york , 143 .
del moral , p. , doucet , a. and jasra , a. ( 2012 ) an adaptive sequential monte carlo method for approximate bayesian compuation .
* 22 * , 10091020 .
ethier , s.n . and kurtz , t.g .
( 1986 ) _ markov processes , characterization and convergence . _
wiley , new york .
fearnhead , p. and prangle , d. ( 2012 ) constructing summary statistics for approximate bayesian computation : semi - automatic approximate bayesian computation ( with discussion ) .
_ j. r. stat .
methodol . _ * 74 * , 419474 kimura , m. and crow , j. ( 1964 ) .
the number of alleles that can be maintained in a finite population .
_ genetics _ * 49 * , 725-738 .
mckinley , t.j . ,
ross , j.v . ,
deardon , r. and cook , a.r .
( 2014 ) simulation - based bayesian inference for epidemic models .
statist . data anal . _ * 71 * , 434447 .
neal , p. and huang , c.l.t .
( 2015 ) forward simulation mcmc with applications to stochastic epidemic models .
j. stat . _
* 42 * , 378396 .
neal , p.j . and roberts , g.o .
( 2005 ) a case study in non - centering for data augmentation : stochastic epidemics .
comput . _ * 15 * , 315327 .
neal , p.j . and roberts , g.o .
( 2006 ) optimal scaling for partially updating mcmc algorithms .
* 16 * , 475515 .
neal , p.j . , roberts , g.o . and yuen , w.k .
( 2012 ) optimal scaling of random walk metropolis algorithms with discontinuous target densities .
* 22 * , 18801927 oneill , p.d . and becker , n.g .
inference for an epidemic when susceptibility varies .
_ biostatistics _ * 2 * , 99108 .
oneill , p.d . and roberts , g.o .
bayesian inference for partially observed stochastic epidemics .
ser . a _ * 162 * , 121129 .
papaspoliopoulos , o. , roberts , g.o . and skld , m. ( 2003 ) non - centered parameterisations for hierarchical models and data augmentation
. _ bayesian statistics 7 _ ( j.m .
bernardo , m.j .
bayarri , j.o .
berger , a.p .
dawid , d. heckerman , a.f.m .
smith and m. west , eds . ) oxford university press , 307326 .
ripley , b. ( 1987 ) _ stochastic simulation .
_ wiley , new york .
roberts , g.o . ,
gelman , a. and gilks , w.r .
( 1997 ) weak convergence and optimal scaling of random walk metropolis algorithms . _
_ , * 7 * , 110120
. roberts , g. o. and rosenthal , j. s. ( 1998 ) optimal scaling of discrete approximations to langevin diffusions .
_ j. r. stat .
methodol . _ * 60 * 255268 .
sherlock , c. and roberts , g.o .
( 2009 ) optimal scaling of the random walk metropolis on elliptically symmetric unimodal targets .
_ bernoulli _ * 15 * , 774798 .
sisson , s. a. , fan , y. and tanaka , m. m. ( 2007 ) sequential monte carlo without likelihoods .
usa _ , * 104 * , 17601765 .
small , p. m. , hopewell , p. c. , singh , s. p. , paz , a. , parsonnet , j. , ruston , d. c. , schecter , g. f. , daley , c. l. , and schoolnik , g. k. ( 1994 ) the epidemiology of tuberculosis in san francisco .
a population - based study using conventional and molecular methods . _ new england journal of medicine _ * 330 * 1703 - 1709 .
tanaka , m. m. , francis , a. r. , luciani , f. and sisson , s. a. ( 2006 ) using approximate bayesian computation to estimate tuberculosis transmission parameters from genotype data .
_ genetics _ * 173 * , 15111520 .
tiernay , l. ( 1994 ) markov chains for exploring posterior distributions .
statist . _
* 22 * , 17011728 .
xiang , f. and neal , p. ( 2014 ) efficient mcmc for temporal epidemics via parameter reduction .
statist . data anal . _ * 80 * , 240250 .
since @xmath256 , for all @xmath257 , @xmath258 , since @xmath259 is the stationary distribution of @xmath74 .
therefore , we have that @xmath260 let @xmath266 . then letting @xmath267 , \end{aligned}\ ] ] we note that for all @xmath268 , @xmath269 , where @xmath270 denotes equality in distribution .
hence for all @xmath271 , @xmath272 = h^\ast ( y , x_1)$ ] . therefore given that @xmath273 it follows that @xmath274)^4 ] \nonumber \\ & = & \binom{n-1}{k-1}^{-4 } \sum_{\mathbf{i}_1 \in \mathcal{i}_n } \sum_{\mathbf{i}_2 \in \mathcal{i}_n } \sum_{\mathbf{i}_3 \in \mathcal{i}_n } \sum_{\mathbf{i}_4 \in \mathcal{i}_n } { \mathbb{e}}\left [ \prod_{j=1}^4 ( \hat{h}_{\mathbf{i}_j } ( y , x_1 , \mathbf{x}_0^{n- } ) - { \mathbb{e}}[\hat{h}_{\mathbf{i}_j } ( y , x_1 , \mathbf{x}_0^{n- } ) ] ) \right ] .
\nonumber \\\end{aligned}\ ] ] note that if @xmath268 have no elements in common then @xmath275 and @xmath276 are independent
. therefore @xmath277 ) ] $ ] is only non - zero if and only if for @xmath278 , @xmath279 has at least an element in common with one the other indices .
moreover , @xmath280 ) ] | \leq 1 $ ] .
the number of combinations of @xmath281 such that @xmath282 and @xmath283 have at least one element in common is @xmath284 which is bounded above by @xmath285 for all sufficiently large @xmath242 .
similarly , the number of combinations of @xmath286 such that @xmath283 , @xmath287 and @xmath288 all have at least one element in common with @xmath282 is @xmath289 which is bounded above by @xmath290 for all sufficiently large @xmath242 .
now @xmath277 ) ] $ ] is only non - zero if either @xmath286 can be grouped into two pairs such that both pairs have at least one element in common or if three of the components all have at least one element in common with the fourth .
( note that there is overlap between these two classifications . ) thus using and , it is straightforward to combine with to show that @xmath291)^4 ] \nonumber \\ & \leq & \binom{n-1}{k-1}^{-4 } \left\ { 3 \left(\frac{n^{2k-3}}{\{(k-2)!\}^2 } \right)^2 + 4 \frac{(k-1)^2 n^{4k-7}}{\ { ( k-2)!\}^4 } \right\ } \nonumber \\ & \leq & \frac{(k-1)^4}{(n - k)^{4k-4 } } \left\ { 3 n^{4k-6 } + 4 ( k-1)^2 n^{4k-7 } \right\}. \end{aligned}\ ] ] since the bound obtained in , holds for all @xmath292 , it follows from and that @xmath293 the lemma immediately follows by combining and . | the independence sampler is one of the most commonly used mcmc algorithms usually as a component of a metropolis - within - gibbs algorithm .
the common focus for the independence sampler is on the choice of proposal distribution to obtain an as high as possible acceptance rate . in this paper
we have a somewhat different focus concentrating on the use of the independence sampler for updating augmented data in a bayesian framework where a natural proposal distribution for the independence sampler exists .
thus we concentrate on the proportion of the augmented data to update to optimise the independence sampler .
generic guidelines for optimising the independence sampler are obtained for independent and identically distributed product densities mirroring findings for the random walk metropolis algorithm .
the generic guidelines are shown to be informative beyond the narrow confines of idealised product densities in two epidemic examples . _
correspondence address : _ department of mathematics and statistics , fylde college , lancaster university , lancaster , la1 4yf , uk _ running title : _
optimal scaling of the independence sampler * keywords : * augmented data ; birth - death - mutation model ; markov jump process ; mcmc ; sir epidemic model . |
we consider the problem of goodness of fit test for the model of ergodic diffusion process when this process under the null hypothesis belongs to a given parametric family .
we study the cramer - von mises type statistics in two different cases .
the first one is based on local time estimator and the second one is based on empirical distribution function estimator .
we show that the cramer - von mises type statistics converge in both cases to some limits which do not depend on the unknown parameter , so the test is asymptotically parameter free ( apf ) .
let us remind the similar statement of the problem in the well known case of the observations of independent identically distributed random variables @xmath0 .
suppose that the distribution of @xmath1 under hypothesis is @xmath2 , where @xmath3 is some unknown parameter . then the cramer - von mises type test is @xmath4 ^ 2{\rm d}f\left(x- \hat \vartheta _ n\right)\ ] ] where the statistic @xmath5 under hypothesis converges in distribution to a random variable @xmath6 which does not depend on @xmath3 .
therefore the threshold @xmath7 can calculated as solution of the equation @xmath8 the details concerning this result can be found in darling @xcite . for more general problems
see the works of kac , kiefer & wolfowitz @xcite , durbin @xcite or martynov @xcite , @xcite . a similar problem exists for the continuous time stochastic processes , which are widely used as mathematic models in many fields .
the goodness of fit tests ( gof ) are studied by many authors . for example
kutoyants @xcite discusses some possibilities of the construction of such tests . in particular , he considers the kolmogorov - smirnov statistics and the cramer - von mises statistics based on the continuous observation .
note that the kolmogorov - smirnov statistics for ergodic diffusion process was studied in fournie @xcite and in fournie and kutoyants @xcite .
however , due to the structure of the covariance of the limit process , the kolmogorov - smirnov statistics is not asymptotically distribution free in diffusion process models .
more recently kutoyants @xcite has proposed a modification of the kolmogorov - smirnov statistics for diffusion models that became asymptotically distribution free .
see also dachian and kutoyants @xcite where they propose some gof tests for diffusion and inhomogeneous poisson processes with simple basic hypothesis .
it was shown that these tests are asymptotically distribution free . in the case of ornstein - uhlenbeck process kutoyants
showed that the cramer - von mizes type tests are asymptotically parameter free @xcite .
another test was studied by negri and nishiyama @xcite .
suppose that we observe an ergodic diffusion process , solution to the following stochastic differential equation @xmath9 we want to test the following null hypothesis @xmath10 where @xmath11 is some known function and the shift parameter @xmath3 is unknown .
we suppose that @xmath12 .
let us introduce the family @xmath13 the alternative is defined as @xmath14 where @xmath15 \right\ } $ ] .
we suppose that the trend coefficients @xmath16 of the observed diffusion process under both hypotheses satisfy the conditions : @xmath17 . _
the function @xmath18 is locally bounded and for some @xmath19 , _
@xmath20 and + @xmath21 . _
the function @xmath18 satisfies _
@xmath22 remind that under the condition @xmath17 , the equation has a unique weak solution ( see @xcite ) .
moreover under the condition @xmath21 , the diffusion process is recurrent and its invariant density @xmath23 under hypothesis @xmath24 can be given explicitly ( see @xcite , theorem 1.16 ) : @xmath25 denote by @xmath26 a random variable ( r.v . ) having this density and the corresponding mathematic expectation by @xmath27 . to simplify the notations , for the case @xmath28
, we denote the density function as @xmath29 , and the corresponding distribution function as @xmath30 ; correspondingly the r.v .
is @xmath31 , and the mathematical expectation is @xmath32 .
denote @xmath33 as the class of functions having polynomial majorants i.e. @xmath34 with some @xmath35 .
let @xmath36 the derivative of @xmath37 w.r.t .
@xmath38 .
let us fix some @xmath39 , and denote by @xmath40 the class of tests @xmath41 of asymptotic size @xmath42 , i.e. @xmath43 our object is to construct this kind of tests .
+ to verify the hypothesis @xmath44 , we propose two tests .
the first one is based on the local time estimator ( lte ) @xmath45 of the invariant density , which can be written as @xmath46 the unknown parameter is estimated via the maximum likelihood estimator ( mle ) @xmath47 , which is defined as the solution of the equation @xmath48 where @xmath49 is the log - likelihood ratio @xmath50 we give the following regularity conditions @xmath51 to have the consistency and the asymptotical normality of the mle : + * condition @xmath52 . *
_ @xmath53 .
the function @xmath54 is continuously differentiable , the derivative @xmath55 and is uniformly continuous in the following sense : _
@xmath56 _ @xmath57 .
the fisher information _ @xmath58 _ moreover , for any @xmath59 _ @xmath60 denote the statistic based on the lte as follows @xmath61 we will prove that under hypothesis @xmath62 , it converges in distribution to @xmath63 with @xmath64 , @xmath65 , where @xmath66 and @xmath67 are independent wiener processes .
the cramer - von mises type test is defined as @xmath68 where @xmath69 is the @xmath70 quantile of the distribution of @xmath71 , that is the solution of the following equation @xmath72 the main result for the cramer von mises test based on local time estimator is the following : [ mainresult1 ] let the conditions @xmath73 , @xmath74 and @xmath75 be fulfilled , then the test @xmath76 belongs to @xmath40 .
the theorem is proved in section [ sectionlte ] .
note that neither @xmath71 nor @xmath69 depends on the unknown parameter .
this allows us to conclude that the test is apf .
+ the second test is based on the same mle and the empirical distribution function ( edf ) : @xmath77 the corresponding statistic is @xmath78 which converges in distribution to @xmath79 thus we propose the cramer - von mises type test @xmath80 where @xmath81 is the solution of the equation @xmath82 the main result for the cramer von mises test based on empirical distribution function estimator is the following : [ mainresult2 ] under conditions @xmath73 , @xmath74 and @xmath75 , the test @xmath83 belongs to @xmath40 .
the theorem is proved in section [ sectionedf ] .
in this section , we study the test @xmath76 , where @xmath84 under the basic hypothesis @xmath62 , the density of the invariant law can be presented as follows : @xmath85 note that the distribution function of the process satisfies @xmath86 in addition , for any integrable function @xmath87 , @xmath88 note that the fisher information in our case does not depend on the unknown parameter @xmath89 : @xmath90 where @xmath91 is the true value of the unknown parameter . from the condition @xmath74
, it follows that there exist some constants @xmath92 and @xmath93 such that for all @xmath94 , @xmath95 it can be shown that for @xmath96 , @xmath97 + similar result can be deduced for @xmath98 , so we have @xmath99 let the conditions @xmath21 and @xmath51 be fulfilled , then the mle @xmath47 is consistent , i.e. , for any @xmath59 , @xmath100 it is asymptotically normal @xmath101 and the moments converge i.e. , for @xmath35 @xmath102 where @xmath103 .
the proof can be found in @xcite , theorem 2.8 .
we can define @xmath104 and denoted @xmath105 , the asymptotical normality can be written as @xmath106 + we define @xmath107 . in @xcite theorem 4.11 , we can find the following representation @xmath108 let us put @xmath109 then @xmath110 can be written as @xmath111 we can state [ intlim]let the condition @xmath74 be fulfilled , then @xmath112 * proof .
* applying the estimate , for @xmath96 , @xmath113 further , @xmath114 moreover @xmath115 and finally @xmath116 then we have @xmath117 similar estimate can be obtained for @xmath98 , therefore the result holds for @xmath94 .
we obtain finally @xmath118 this result yields directly the conditions @xmath119 of theorem 4.11 in @xcite : @xmath120 and @xmath121 so we can deduce the convergence and the asymptotical normality of @xmath110 .
in fact under the condition @xmath21 , the lte @xmath122 is consistent and asymptotically normal , that is @xmath123 where @xmath124 , and @xmath125 moreover @xmath126 we can define @xmath127 the distribution of @xmath128 is @xmath129 , and we have the following convergence @xmath130 for @xmath131 and @xmath110 , we need more than and convergence .
[ lemconv ] let conditions @xmath74 and @xmath52 be fulfilled , then @xmath132 is asymptotically normal : @xmath133 for any @xmath134
. * proof . *
the first integral in converges to zero , so we only need to verify the convergence for the part of it integral .
let us denote for simplicity @xmath135 it is sufficient to verify that for any @xmath136 , @xmath137 remember that @xmath131 can be defined as follows , @xmath138 where @xmath139 here @xmath140 and @xmath141 .
it was proved in @xcite , theorem 2.8 that @xmath142 converges in distribution to @xmath143 , where @xmath144 where @xmath145 is a r.v . with normal distribution @xmath146 , which can be written as @xmath147 therefore @xmath148 take @xmath149 .
we have to verify that the joint finite - dimensional distribution of @xmath150 @xmath151 converges to the finite - dimensional distribution of @xmath152 @xmath153 note that the only stochastic term in @xmath154 is @xmath155 , so is equivalent to @xmath156 take @xmath157 , and put @xmath158 we have @xmath159 the law of large number gives us @xmath160 moreover , the central limit theorem for stochastic integral gives us @xmath161 in addition @xmath162 is a zero mean normal r.v . with variance @xmath163
furthermore @xmath164 and @xmath165 @xmath166 we find that @xmath167 this is as to say @xmath168 thus follows from this last convergence in distribution , and so the lemma is proved .
+ [ lemintconv ] let conditions @xmath74 and @xmath52 be fulfilled , then @xmath169 * proof .
* denote @xmath170 and @xmath171 , we will prove the following properties \i ) for @xmath172 $ ] and @xmath173 , @xmath174 \ii ) @xmath175 , such that @xmath176 from i ) it follows the convergence in every bounded set @xmath177 $ ] : @xmath178 the result in i ) along with ii ) gives us the result .
+ first we prove i ) .
we have @xmath179 @xmath180 for the first part , let us recall the following result , given in @xcite , page 119 : for any @xmath181 , chosen @xmath182 sufficiently large , we have @xmath183 now , denoted @xmath184 the distribution of @xmath185 , we have @xmath186\\ \label{u_tbounded } & & \quad\leq1-[1-f_t(1)]+p\int_1^\infty u^{p-1}\frac{c_n}{u^{n / p}}{\textrm{d}}u\leq c.\end{aligned}\ ] ] remember that under condition @xmath187 , @xmath188 and @xmath189 are sufficiently smooth .
so , for @xmath190 $ ] we can write @xmath191 so we have @xmath192 for the second part , we can write @xmath193 suppose that @xmath194 , @xmath195 similar result holds for @xmath196 .
then we obtain @xmath197 thus we have @xmath198 now we prove ii ) .
as in lemma [ intlim ] , we can deduce that @xmath199 so for @xmath200 , @xmath201 note that @xmath202 and along with we get @xmath203 for any @xmath204 , take @xmath205 , then we have .
+ + * proof of theorem [ mainresult1 ] . * + we can write @xmath206 see that @xmath207 and that @xmath208 , @xmath209 , the smoothness of @xmath54 gives us the convergence @xmath210 + applying lemma [ intlim ] and lemma [ lemintconv ] we get @xmath211 we see that the limit of the statistic @xmath71 does not depend on @xmath91 , and the test @xmath212 with @xmath69 defined by @xmath213 belongs to @xmath40 .
+ the same procedure can be applied with other estimators of the unknown parameter and of the invariant density , provided that they are consistent and asymptotically normal .
for example , we can take the minimum distance estimator ( mde ) @xmath214 for @xmath91 : @xmath215 and the kernel estimators @xmath216 as estimator for the invariant density @xmath217 under some regularity conditions , the mde @xmath218 is asymptotically normal ( see @xcite or @xcite ) : @xmath219 also if we do not present explicitly @xmath220 here , it can be verified that @xmath221 does not depend on @xmath89 .
the kernel estimator @xmath216 has the same asymptotic properties of the lte ( see @xcite )
. then we can construct the statistic @xmath222 which converges to @xmath223 that does not depend on the unknown parameter . so that the test @xmath224 with @xmath225 the solution of the equation @xmath226 belongs to @xmath40 . in this section
, we study the gof test @xmath227 defined by the statistic @xmath228 where @xmath229 is the empirical distribution function : @xmath230 denote @xmath231 and @xmath232 in @xcite theorem 4.6 , the following equality is presented : @xmath233 then @xmath234 using we have , for @xmath96 , @xmath235 and @xmath236 for @xmath98 we have @xmath237 and we can write @xmath238 these inequalities allow us to deduce the following bounds @xmath239 and
@xmath240 moreover @xmath241 hence we get the asymptotic normality of @xmath242 : @xmath243 as in lemma [ lemconv ] and lemma [ lemintconv ] , if conditions @xmath52 and @xmath74 hold , we can show the convergence of the vector @xmath244 : @xmath245 and the convergence of the integral : @xmath246 we obtain finally @xmath247 ^ 2{\textrm{d}}x\\ & = & \int_{-\infty}^\infty\big[\eta_t^f(x)-\hat u_tf(x-\tilde{\vartheta}_t)\big]^2{\textrm{d}}x\\ & = & \int_{-\infty}^\infty\big[\eta_t^f(x ) -\hat u_tf(x-{\vartheta}_0)\big]^2{\textrm{d}}x+o(1)\\ & \longrightarrow&\int_{-\infty}^\infty\big[\eta^f(x-{\vartheta}_0 ) -\hat u f(x-{\vartheta}_0)\big]^2{\textrm{d}}x\\ & = & \int_{-\infty}^\infty\left(\eta^f(y)-\hat u f(y)\right)^2{\textrm{d}}y=\delta.\end{aligned}\ ] ] so that the limit of the statistic @xmath248 does not depend on @xmath91 , and the test @xmath249 with @xmath81 the solution of @xmath250 belongs to @xmath40 . + * remark .
* it can be shown that in the case of kolmogorov - smirnov tests @xmath251 where @xmath252 the limit distributions of these statistics ( under hypothesis ) do not depend on @xmath253 .
the proofs can be done following the same lines as in kutoyants @xcite and negri @xcite respectively .
in this section we discuss the consistency of the proposed tests .
we study the tests statistics under the alternative hypothesis that is defined as @xmath254 where @xmath15 \right\ } $ ] .
under this hypothesis we have : let all drift coefficients under alternative satisfy the conditions @xmath255 and @xmath75 , then for any @xmath256 we have @xmath257 and @xmath258 * proof .
* remember that under hypothesis @xmath259 , the mle @xmath47 converges to the point which minimize the distance @xmath260 where @xmath261 is the random variable of invariant density @xmath262 ( see @xcite , proposition 2.36 ) : @xmath263 in addition , denoted with @xmath264 the norm in @xmath265 , we have @xmath266 we can deduce @xmath267 moreover @xmath268 and finally we have the result for @xmath269 : @xmath270 a similar result can be obtained for @xmath271 .
we consider the ornstein - uhlenbeck process .
remind that the tests for o - u process were studied in @xcite as well .
suppose that the observed process under the null hypothesis is we simulate @xmath283 trajectories of @xmath71 ( resp .
@xmath284 ) and calculate the empirical @xmath70 quantiles of @xmath71 ( resp .
@xmath284 ) .
we obtain the simulated density for @xmath71 and @xmath284 that are showed in graphic [ density ] .
the values of the thresholds @xmath69 for different @xmath42 are showed in graphic [ d_vep ] .
dachian s. and kutoyants y.a . (
2007 ) . on the goodness - of - fit tests for some continuous time processes .
statistical models and methods for biomedical and technical systems , f.vonta et al .
( eds ) , birkhauser , boston , 395 - 413 . | a problem of goodness - of - fit test for ergodic diffusion processes is presented . in the null hypothesis
the drift of the diffusion is supposed to be in a parametric form with unknown shift parameter .
two cramer - von mises type test statistics are studied . the first one is based on local time estimator of the invariant density
, the second one is based on the empirical distribution function .
the unknown parameter is estimated via the maximum likelihood estimator .
it is shown that both the limit distributions of the two test statistics do not depend on the unknown parameter , so the distributions of the tests are asymptotically parameter free .
some considerations on the consistency of the proposed tests and some simulation studies are also given .
* keywords : * ergodic diffusion process , goodness - of - fit test , cramer - von mises type test . |
observations of type ia supernovae@xcite , along with the usual assumption that the universe is homogeneous , indicate that the expansion of the universe is accelerating . this acceleration necessitates the presence of an exotic form of matter called dark energy , which can most easily be accounted for by a cosmological constant .
however , because the scale of the cosmological constant is so small ( @xmath0 in planck units ) it is natural to look for alternative explanations for the data .
one such alternative explanation is to account for the supernova data using a comsmological model that is inhomogeneous.@xcite the simplest such models are the lematre - tolman - bondi(ltb ) spacetimes.@xcite these are spherically symmetric spacetimes with pressureless fluid ( dust ) as the matter .
models of this form can be constructed that account for the supernova data as well as the standard homogeneous `` concordance '' cosmology does.@xcite thus to test the difference between concordance cosmology and inhomogeneous cosmology , a different test is needed .
such a test is provided by the kinematic sunyaev - zeldovich ( ksz ) effect.@xcite this effect comes about when cosmic microwave background ( cmb ) photons scatter off the gas in a galaxy cluster .
the spectrum of the scattered photons depends on the velocity of the galaxy cluster with respect to the microwave background , and thus an analysis of the cmb can be used to find this velocity .
thus , each ltb model must be compared not only to the supernova data but also to the ksz data , and only those models that match both sets of data can be considered viable .
analyses of the ksz effect for particular ltb models have been performed@xcite as well as a general analysis for slightly inhomogeneous models.@xcite unfortunately , each ltb model is different , and so each must be seperately subjected to the ksz test ( as well as the supernova test ) to see if that model is viable . in this paper we will present a general method for finding the velocity of galaxy clusters in ltb models
. we will then apply this method to the models treated in@xcite .
section 2 presents the general method of calculating these velocities .
section 3 applies this method to the models of@xcite .
conclusions are presented in section 4 .
the metric in the ltb spacetimes takes the form @xmath1 here the area radius @xmath2 is not one of the coordinates but is instead a function of the coordinates @xmath3 and @xmath4 .
the function @xmath5 is a function of @xmath4 .
an overdot denotes derivative with respect to @xmath3 while a prime denotes derivative with respect to @xmath4 .
the ltb spacetime has a prefered timelike vector @xmath6 which is the four - velocity of the dust that makes up the matter in the spacetime .
however , an ltb spacetime with a cosmic microwave background has another prefered vector , which we will call @xmath7 , the four - velocity of the cmb .
since the spacetime is spherically symmetric , there is also a preferred unit spacelike vector @xmath8 such that @xmath9 is an outgoing null vector and @xmath10 is an ingoing null vector .
expressed in terms of @xmath7 and @xmath8 , the ltb four - velocity @xmath11 takes the form @xmath12 where @xmath13 is the velocity of the ltb observer with respect to the cmb and where @xmath14 .
now let @xmath15 and @xmath16 be respectively ingoing and outgoing cmb photons each with the same frequency @xmath17 as measured in the cmb rest frame .
then it follows that these vectors take the form @xmath18 now define the quantity @xmath19 by @xmath20 then it follows that @xmath21 we restrict attention to times when the cmb has decoupled from the matter , so that @xmath22 are null geodesics . thus , to find the velocity of the galaxies with respect to the cmb
, we need to find the behavior of null geodesics in ltb spacetimes . to find the behavior of null geodesics ,
we first recall some facts about the ltb spacetimes . from the einstein field equation
, it follows that @xmath23 where the function @xmath24 depends only on @xmath4 .
the density is given by @xmath25 it is helpful to introduce the quantities @xmath26 and @xmath27 by @xmath28 then equation ( [ rdot ] ) becomes @xmath29 whose solution is @xmath30 here @xmath31 is a function of @xmath4 whose meaning is the time at which the shell of dust with coordinate @xmath4 shrinks to zero radius . to have a genuine big bang singularity , we chose @xmath32 .
we use the coordinate freedom to change @xmath4 to any function of @xmath4 to set @xmath27 to the constant @xmath33 . with this choice , the particular ltb model
is then determined by the choice of the function @xmath34 .
in particular , when @xmath35 the ltb solution becomes the standard dust frw cosmology with @xmath36 . for any future directed null geodesic @xmath37 define the quantity @xmath38 .
then it follows that @xmath39 where @xmath40 where @xmath41 and the sign of @xmath42 depends on whether @xmath37 is outgoing or ingoing .
note from the definition of @xmath43 that @xmath44 where @xmath45 is the @xmath43 corresponding to @xmath16 and correspondingly for @xmath46 .
it follows from the form of the ltb metric that @xmath47 then using the geodesic equation we find @xmath48 from which it follows that @xmath49 + \epsilon { \sqrt { 1+f } } { a ' } ) \label{main}\ ] ] we will use equation ( [ main ] ) to find the velocity @xmath13 of the dust relative to the cmb as follows : note that since we have chosen @xmath32 it follows that at early times the spacetimes becomes frw .
note also that in frw spacetimes @xmath50 and therefore @xmath43 is constant . for cmb photons with the average energy this constant , which we will call @xmath51 is the same for outgoing and ingoing photons .
integrating equation ( [ main ] ) back from the point at which we want to calculate @xmath13 to the big bang yields @xmath52 and since we do this for both outgoing and ingoing geodesics , we obtain both @xmath53 and @xmath54 which in turn allows us to calculate @xmath55 and thus the velocity @xmath13
. we will numerically integrate equation ( [ main ] ) .
however , in order to do that we will need to be able to evaluate all the quantities on the right hand side of that equation at all points in the integration . as we will show , all that is needed is to know @xmath2 and @xmath56 and the rest of the quantities follow .
however , @xmath2 and @xmath56 also change as the integration procedes , so we need to know how they change . from equation([elldef ] ) we find @xmath57 note that an ingoing geodesic that goes through @xmath58 becomes an outgoing geodesic , where this change is implemented as a change in the sign of @xmath42 .
we must evolve the system of equations ( [ main]-[evolvea ] ) . to do this
, we must be able to evaluate all quantities on the right hand sides of these equations from just @xmath2 and @xmath56 , that is , we must be able to find @xmath59 and @xmath60 .
given @xmath2 and @xmath56 we obtain @xmath61 and thus we find all specified functions of @xmath4 such as @xmath62 and @xmath5 . from equation ( [ adotsquared ] )
we obtain @xmath63 which then yields @xmath64 since @xmath65 . differentiating equation ( [ timediff ] ) with respect to @xmath4 and solving for @xmath66 yields @xmath67 where @xmath68\ ] ] with @xmath69 .
this then yields @xmath70 through @xmath71 .
finally , differentiating equation ( [ adot ] ) with respect to @xmath4 yields @xmath72 which in turn yields @xmath60 through @xmath73
. what remains is then simply to find the initial values of @xmath2 and @xmath56 for each point in our past light cone , along with the coresponding values of the redshift @xmath74 .
as shown in @xcite this can be done by integrating equations ( [ evolver]-[evolvea ] ) from our current position , along with the evolution equation for the redshift @xmath75 for this integration , the initial value of @xmath2 is zero , while the initial value of @xmath56 is given by @xmath76 where @xmath77 is the ratio of the density in matter to the critical density .
in the models of @xcite a choice of @xmath77 is made , and the function @xmath62 takes the form @xmath78 where the constant @xmath79 is chosen for best fit with the supernova data . in particular , the two models considered are ( i ) @xmath80 , which gives rise to @xmath81 , and ( ii ) @xmath82 , which gives rise to @xmath83 .
each of these models fits the supernova data about as well as the standard concordance cosmology . because @xmath84 at large @xmath4 , these models are asymptotically @xmath85 frw cosmologies with an underdensity in a region near the center
note that the large value of @xmath79 means that in these models we live in an underdensity of fairly small size .
using the method of the previous section , we calculate the velocity of the dust with respect to the cmb for each point on our past light cone .
the results are shown in figures ( 1 ) and ( 2 ) . here
figure ( 1 ) is for the @xmath86 model and figure ( 2 ) is for the @xmath87 model .
also plotted on the figures are the results of observations of the ksz effect for galaxy clusters.@xcite plot of velocity ( in units of km / s ) _ vs _ redshift for the @xmath88 ltb model ] plot of velocity ( in units of km / s ) _ vs _ redshift for the @xmath89 ltb model ] the observations clearly rule out the @xmath89 model .
in addition , the @xmath88 model does not provide a good fit to the observations , so this model may also be ruled out .
any successful ltb model must pass both the supernova test and the ksz test .
together these tests provide a powerful constraint that rules out the models of @xcite .
however , the models of @xcite were chosen on the basis of simplicity and fitting the supernova test ; so it is not surprising that one additional test suffices to rule them out .
it is therefore possible that reasonable ltb models could be made that have fair agreement with both the supernova data and the ksz data .
however , as ksz measurements improve in accuracy , this test will become more stringent and might eventually rule out all ltb models .
i would like to thank nick kaiser and juan garcia - bellido for helpful discussions .
this work was supported by nsf grants phy-0456655 and phy-0855532 to oakland university . | lematre - tolman - bondi inhomogeneous spacetimes can be used as a cosmological model to account for the type ia supernova data . however , such models also give rise to large velocities of galaxy clusters with respect to the cosmic microwave background .
these velocities can be measured using the kinematic sunyaev - zeldovich effect .
this paper presents a calculation of galaxy cluster velocities as a function of redshift for such a model . |
millisecond radio pulsars ( msrps ) are believed to descend from the neutron stars in low mass x - ray binaries ( lmxbs ) that have been spun up by acquiring angular momentum through accretion of matter from their companions @xcite .
khz quasiperiodic oscillations and millisecond pulse periods of neutron stars in several lmxbs support this theory ( @xcite and references therein ) and the most compelling proof came from the msrp j1023 + 0038 which has lost its accretion disc within the past decade @xcite . in spite of continuing efforts to understand the evolution of lmxbs @xcite ,
sometimes discoveries of msrps with peculiar characteristics impose problems and compel us to think of alternative scenarios .
two such systems are binary msrps psr j1903 + 0327 which has an eccentric orbit @xcite and psr j1614 - 2230 which is highly recycled with comparatively larger companion mass @xcite ; and special evolutionary channels have been proposed for these objects @xcite .
but all these evolutionary models usually ignore the effects of binding energies of neutron stars which should be taken care of for a more realistic understanding of the binary evolution as neutron stars posses significant amounts of binding energies . in the present work ,
we show that the change in the total binding energy in accretion driven evolution is significant and should be incorporated in realistic models of binary evolutions .
the gravitational mass of a neutron star is defined as @xmath0 where @xmath1 is the mass density at a radial distance @xmath2 from the center and @xmath3 is the radius of the neutron star .
this is the measurable entity and being used in studies of neutron star mass distributions @xcite .
but in the case of relativistic objects like neutron stars , one should think about the proper volume element @xmath4 which becomes @xmath5^{-1/2 } \
, dr$ ] in schwarzschild geometry .
so the proper mass of the neutron star becomes @xmath6^{-1/2 } \rho(r ) \ , dr$ ] and the total number of particles ( baryons ) becomes @xmath7^{-1/2 } n(r ) \ , dr$ ] where @xmath8 is the particle number density at @xmath2 .
this makes one define the baryonic mass as @xmath9^{-1/2 } n(r ) \ , dr$ ] where @xmath10 is the mass of a baryon .
the binding energy of the neutron star is @xmath11 where @xmath12 is the velocity of light and the gravitational binding energy of the neutron star is @xmath13 . as the only difference in the expression of @xmath14 from that of @xmath15 is the appearance of the metric element @xmath16^{-1/2}$ ] inside the integral , so @xmath17 is actually the manifestation of the space - time curvature for objects as compact as neutron stars .
solving tollman - oppenheimer - volkov ( tov ) equations of hydrostatic equilibrium , one can obtain the values of all these entities with a suitable choice of the equation of state ( eos ) .
it is noteworthy to mention here that although tov equations have been derived for non - rotating stars in schwarzschild geometry , these are good enough even for the millisecond pulsars .
the definition of the proper volume @xmath4 is different for a rotating - star where the metric is different , but we have found with the freely available rapidly rotating neutron star " code by nikolaos stergioulas that for a fixed value of @xmath18 , @xmath19 usually decreases by an amount less than 1% if we increase the angular velocity @xmath20 from zero to 3000 @xmath21 .
so in the present work , we keep ourself confined within schwarzschild geometry .
one important question that arises here is what is the value of @xmath10 ? following @xcite , we set @xmath10 equal to one atomic mass unit
_ u _ " ( based on @xmath22 scale , 931.49402 mev ) which is appropriate as pre - collapse evolved stars contain large amounts of @xmath22 or nuclei of similar binding energy per nucleon .
the mass of @xmath23 is 1.007825 _ u _ and the mass of @xmath24 is 4.002602 _ u _ , so choosing @xmath25 _ u _ is valid even for the present work although we are mainly concerned about the change in @xmath26 due to the accretion of hydrogen or helium rich matter on the neutron star from its binary companion ( usually in the red - giant phase ) . among numerous nuclear esos
, we choose a standard eos apr @xcite which agrees with the possible @xmath27 regions for three neutron stars determined from x - ray observations of thermonuclear bursts @xcite as well as gives maximum value of @xmath15 to be greater than the mass of the most massive neutron star discovered till date - the millisecond pulsar psr j1614 - 2230 having @xmath28 @xcite .
but there is a possibility of discovering even heavier neutron stars in future - the black widow " pulsar b1957 + 20 may be one such example which has the lower limit of @xmath15 as @xmath29 with the best fit value as @xmath30 @xcite .
further study is needed to get a precise mass measurement for this neutron star .
but at the present situation , one should not exclude very stiff esos in any study involving esos . as an example of such a stiff eos , we use ms0 @xcite although it disagrees with the findings of @xcite . as an eos much softer that apr would not give @xmath15 as high as @xmath31
, we do not use any such eos .
it is worthwhile to mention here that the statistical method of @xcite favors esos which are even softer than apr at high densities ( see their figure 8) .
but their results fail to give @xmath15 as high as @xmath32 and there is enough scope to improve their method ( see their discussions ) . on the other hand
, the study of the cooling phase of the x - ray burster 4u 1724 - 307 by @xcite favors an eos as stiff as ms0 . in the top panel of fig .
[ fig : mrbn ] , we show mass - radius relations of neutron stars as obtained by solving tov equations with these two esos .
it is clear that the difference between three types of masses increases with the increase of @xmath18 leading to the increase in @xmath17 and @xmath33 with @xmath15 ( shown in the middle and bottom panels of the same figure ) .
this happens because with the increase of @xmath15 , neutron stars become more compact and the general relativistic effects more dominant . for a fixed value of @xmath15 , the values of @xmath17 and @xmath33 are higher for a softer eos than those for a stiffer eos due to the higher compactness of the neutron star in the first case
this fact had been observed earlier by @xcite . while @xmath17 is useful to understand general relativistic effects , it is @xmath33 which is relevant in case of the accretion driven evolution which we discuss next .
we study the role of the binding energy in the evolution of millisecond pulsars which are the results of the mass accretion onto the initial slow neutron stars from their companions .
let us assume a millisecond pulsar with gravitational mass @xmath34 and binding energy @xmath35 initially was a slow neutron star with the gravitational mass @xmath36 and binding energy @xmath37 before the accretion process started . here
@xmath38 implying @xmath39 ( bottom panel of fig .
[ fig : mrbn ] ) .
the increase in the gravitational mass is @xmath40 and the increase in the binding energy is @xmath41 and the total amount of mass accreted is given by @xmath42 .
unfortunately , people usually assume @xmath43 ; one of the main aims of the present paper is to point out this error . in fig .
[ fig : macc ] we plot @xmath36 along x - axis , @xmath34 along y - axis and @xmath44 as the colour palette .
the increase of @xmath15 of a neutron star with the amount of accreted matter can be studied along a straight line parallel to the y - axis .
the x - coordinate of the line will give the initial value of the @xmath15 ( @xmath36 ) and the y - coordinate at any point in the line will give the final value of @xmath15 ( @xmath34 ) for which the amount of mass accreted can be obtained from the colour at that point .
but @xmath36 is not the observable entity , it is @xmath34 which can be measured . by noting the colours at different points along a straight line parallel to the x - axis , one can get the amount of accreted mass , where the y - coordinate of the straight line should be set equal to the measured value of @xmath34 and the x - coordinate of any point along the line will correspond to the assumed value of @xmath36 . let us substantiate this for the case of the millisecond pulsar psr j1614 - 2230 which has @xmath45 @xcite , spin period 3.15 ms and a companion of mass @xmath46 . table [ tb : macc ]
shows the values of @xmath44 for different values of @xmath47 setting @xmath48 .
we find that @xmath44 ( @xmath49 ) is quite high for @xmath50 which is difficult to explain with present evolutionary models .
there are two possibilities to explain the fact , first the neutron star was probably born massive ( @xmath51 ) or the accretion process is much more efficient .
further study is needed to be sure of possible origin of psr j1614 - 2230 .
moreover , we see from table [ tb : macc ] that for fixed values of @xmath52 and @xmath47 , different esos give different values for @xmath44 . for
a softer eos , fixed values of @xmath52 and @xmath47 give a larger value of @xmath44 than that obtained for a stiffer eos because of the larger compactness of the neutron star in the first case .
thus proper understanding of binary evolution requires the true knowledge of dense matter eos .
similarly for any other msrp , @xmath44 can be estimated with proper choices of @xmath47 and the eos . .@xmath44
as obtained using esos apr and ms0 for different values of @xmath47 when @xmath48 which gives @xmath53 for eos apr and @xmath54 for eos ms0 .
[ cols="^,^,^,^,^,^ " , ] [ tb : macc ]
the extra contribution of the binding energy to the estimate of total accreted mass @xmath44 makes the recycling process more efficient as the conservation of angular momentum gives @xmath55 . here
the angular momentum of the accreted material is given by @xmath56 @xcite , @xmath57 is the alfven radius and @xmath58 is the keplerian velocity at @xmath57 ; @xmath59 , @xmath60 where @xmath61 , @xmath62 , @xmath63 are the final angular momentum , moment of inertia and angular velocity and @xmath64 , @xmath65 , @xmath66 are the initial angular momentum , moment of inertia and angular velocity .
moment of inertia of a neutron star increases almost linearly with the increase of @xmath18 except very close to the maximum mass @xcite .
so if one perform the calculation considering @xmath67 instead of @xmath43 with the same choice of all other parameters , @xmath63 will be higher in the earlier case .
the effect of binding energy will be important in any study for @xmath68 correlation as recently done by @xcite without considering this effect .
we find that neutron stars posses significant amount of binding energies which increase with the increase of the gravitational mass of the neutron stars .
so when a neutron star gains mass by accreting matter from its companion , its binding energy also increases and the amount of accretion needed becomes greater than the difference in the gravitational masses before and after the accretion process .
indeed it is the difference between the baryonic masses before and after the accretion process .
this fact makes it more difficult to explain the evolution of massive msrps like psr j1614 - 2230 .
it seems that probably psr j1614 - 2230 was born massive ( @xmath51 ) .
if this is indeed the case , then it is quite possible of discovering more neutron stars with such high gravitational masses - even the neutron stars in the double neutron star systems may have such high gravitational masses whereas in the present sample of double neutron stars , the gravitational masses are usually much smaller - with a mean of @xmath69 @xcite .
we also find that for the same amount of increase in the gravitational mass of a neutron star , different equations of state predict different values for the increase in the total binding energy .
this fact reveals the importance of knowing the true nature of the matter inside a neutron star for better understanding of the evolution of the neutron star through mass accretion from its companion .
we use two sample esos in this work but the correctness of these are not firmly established .
measurement of the moment of inertia from the faster component of the double pulsar system psr j0737 - 3039 @xcite or detection and analysis of low - frequency ( @xmath70 hz ) gravitational waves @xcite might help us to constrain dense matter esos in near future .
the author thanks sushan konar for discussions and john miller for a valuable referee report . 99 alpar , m. a. , cheng , a. f. , ruderman , m. a. , & shaham , j. , 1982 , nature , 300 , 728 archibald , a. m. , stairs , i. h. , ransom , s. m. , 2009 , science , 324 , 1411 akmal , a. , pandharipande , v. r. , & ravenhall , d. g. , 1998 , phrvc , 58 , 1804 arnett , w. d. , & bowers , r. l. , 1977 , apjs , 33 , 415 bagchi , m. , 2010 , newa , 15 , 126 belczynski , k. , kalogera , v. , rasio , f. a. , et al . , 2008 , apjs , 174 , 223 bhattacharya , d. , & van den heuvel , e. p. j. , 1991 , phys . rep . , 203 , 1 champion , d. j. ; ransom , s. m. , lazarus , p. , et al . , 2008 , science , 320 , 1309 demorest , p. b. , pennucci , t. , ransom , s. m. , roberts , m. s. e. , & hessels , j. w. t. , 2010 , nature , 467 , 1081 freire , p. c. c. , bassa , c. g. , wex , n. , et al . , 2010 , arxiv:1011.5809 iben , i. , tutukov , a. v. , & yungelson , l. r. , 1995 , apj , 100 , 233 konar , s. , bhattacharya , d. , 1999 , mnras , 303 , 588 kiziltan , b. , kottas , a. , & thorsett , s. e. , 2010 , arxiv:1011.4291 liu , x. w. , li , x. d. , 2009 , apj , 692 , 723 lin j. , rappaport , s. , podsiadlowski , ph .
, et al . , 2010 , arxiv:1012.1877 lattimer , j. m. , & schutz , b. f. , 2005 , apj , 629 , 979 mller , h. , serot , b. d. , 1996 , nupha , 606 , 508 nelson , l. a. , dubeau , e. , & mccannell , k. a. , 2004 , apj , 616 , 1124 zel , f. , baym , g. , & gver , t. , 2010 , arxiv:1002.3153 zel , f. , psaltis , d. , ransom , s. , demorest , p. , & alford , m , 2010 , apj , 724 , l199 pfahl , e. , rapppaport , s. , & podsiadlowski , ph . , 2003 , apj , 597 , 1036 podsiadlowski , ph . , dewi , j. d. m. , lesaffre , p. , et al . , 2005 , mnras , 361 , 1243 podsiadlowski , p. , rappaport , s. , pfahl , e. d. , 2002 , apj , 565 , 1107 pylyser , e. , & savonije , g. j. , 1988 , a&a , 191 , 57 rappaport , s. , verbunt , f. , & joss , p. c. , 1983 , apj , 275 , 713 radhakrishnan v. , srinivasan g. , 1982 , current science , 51 , 1096 schwab , j. , podsiadlowski , ph . , & rappaort , s. , 2010 , apj , 719 , 722 suleimanov v. , poutanen , j. , revnivtsev , m. , werner , k. , 2010 , arxiv:1004.4871 steiner , a. w. , lattimer , j. m. , & brown , e. f. , 2010 , apj , 719 , 722 thorsett , s. e. , & chakraborty , d. , 1999 , apj , 512 , 288 webbink , r. f. , rappaport , s. , & savonije , g. j. , 1983 , 270 , 678 wijnands , r. , van der klis , m. , 1998 , nature , 394 , 344 zhang , c. m. , zhao , y. h. , yin , h. x. , et al . , 2010 , arxiv:1010.5429 van kerkwijk , m. h. , breton , r. p. , & kulkarni , s. r. , 2010 , to appear in apj , arxiv:1009.5427 | millisecond pulsars are believed to descend from low mass x - ray binaries .
observable parameters of binary millisecond pulsars _
e.g. _ mass of the pulsar , mass of the companion , spin period of the pulsar , orbital period , orbital eccentricity etc are used to probe the past accretion history of the millisecond pulsars .
but unfortunately in these studies , binding energy of the neutron star is not considered commonly .
we show that the effect of the binding energy is significant in the estimation of the amount of mass accretion and thus should be incorporated in models for binary evolutions .
moreover , different equations of state for dense matter give different values for the accreted mass for the same amount of increase in the gravitational mass of the neutron star implying the need of constraining dense matter equations of state even to understand the spin - up procedure properly .
stars : neutron pulsars : general equation of state |
in this note we present an explicit construction of a relative packing .
the subject of symplectic packing was introduced first in the seminal work of gromov @xcite .
gromov showed that looking at symplectic embeddings of a standard ball into a symplectic manifold , one may obtain an upper bound on the radius of a ball which is stronger than the obstruction coming from the volume .
the first theorem in this direction is a non - squeezing theorem from @xcite .
this result has led to the definition of the gromov capacity , which plays an important role in modern symplectic geometry .
later the subject of symplectic packing was treated by biran , karshon , mcduff , polterovich , schlenk , traynor and others , see @xcite .
new obstructions for symplectic packings of various domains were found . on the other hand
, attempts were made to find explicit constructions of certain symplectic embeddings , in order to show that the obstructions , which were found , are tight ( see , e.g. , @xcite ) .
this note is devoted to proving a new result in this direction .
recently , biran and cornea @xcite found new obstructions on the relative symplectic packing in a number of situations , which are stronger than those in the case of a usual packing . in this note
we consider one specific example from @xcite , and show that the corresponding obstruction is sharp .
let us define first the notion of a relative symplectic packing .
consider a symplectic manifold @xmath1 and a closed lagrangian submanifold @xmath2 .
take a standard open ball @xmath3 of radius @xmath4 , where @xmath5 is endowed with coordinates @xmath6 , and set @xmath7 a relative packing of @xmath8 into @xmath9 is by definition a symplectic embedding @xmath10 such that @xmath11 . in this note
we treat the situation where @xmath12 and @xmath13 is the standard clifford torus in @xmath14 . here , @xmath15 is normalized so that @xmath16 . in @xcite
it was shown that given a relative packing @xmath17 there is an upper bound for the radius of this ball : @xmath18 .
our main result is the following .
[ t : pack example ] for every @xmath19 there exists a relative packing @xmath20 let us mention that in @xcite , the authors also consider relative packings into @xmath21 by more than one ball , and obtain obstructions on their radii .
the case of three balls was treated , and is strongly connected to the properties of the quantum cup - product in floer homology .
the hypothetical upper bound in this case was found , under the assumption of existence of pseudo - holomorphic discs with certain properties . it still remains to show the existence of such discs , and in the case that it will be proved , one can try to find the example which proves the tightness of this upper bound .
the rest of the paper is devoted to proving theorem [ t : pack example ] . in section [ s : constrover ]
we show how to reduce theorem [ t : pack example ] to a two - dimensional problem .
section [ s : proofs ] contains the proof of theorem [ t : pack example ] .
it is well - known that the open symplectic manifold @xmath22 is symplectomorphic to the unit ball @xmath23 . identifying @xmath24
, we have a natural action of the torus @xmath25 on @xmath26 .
the moment map of this action is given by @xmath27 the action of @xmath28 , restricted to the complement @xmath29 of the union of the two complex axes is free , and therefore , by a standard procedure , we obtain that @xmath29 is symplectomorphic to @xmath30 . here
we use the notation @xmath31 look now at @xmath32 and its subset @xmath33 , where @xmath34 the above symplectomorphism between @xmath29 and @xmath30 induces a symplectic embedding @xmath35 from now on we will consider @xmath36 as @xmath37 @xmath38 and the symplectic form is @xmath39 the clifford torus lies entirely in @xmath40 , and its pre - image under the map @xmath41 equals to @xmath42 fix @xmath19 and consider @xmath43 the construction is based on finding a certain area - preserving map @xmath44 given such an @xmath45 , we define @xmath46 as @xmath47 our aim is to find @xmath45 such that the image of @xmath48 under the map @xmath49 will be contained in @xmath50 , and also @xmath51
given @xmath19 , we describe a construction of a relative symplectic embedding @xmath52 as it was shown in section [ s : constrover ] , it is enough to find a symplectic embedding @xmath53 such that its image will be contained in the domain @xmath54 take @xmath66 , and consider the corresponding map @xmath45 .
then for any @xmath67 , such that @xmath68 , define @xmath47 we claim that the resulting map @xmath69 satisfies the desired properties .
first of all , it is clear that the map @xmath49 is a symplectic embedding , and the pre - image of @xmath70 equals @xmath71 .
let us show that , moreover , the image of @xmath49 lies in the domain @xmath72 the presented relative packing construction can be naturally generalized to the corresponding construction of a relative packing of a ball @xmath78 into @xmath79 , for any @xmath80 and radius @xmath81 .
this confirms the sharpness of the upper bound for the radius , obtained by biran and cornea @xcite , for an arbitrary dimension .
p. biran , _ from symplectic packing to algebraic geometry and back _ , proceedings of the 3rd european congress of mathematics ( barcelona 2000 ) , vol .
ii , 507 - 524 , progr .
, * 202 * , birkhauser , basel , 2001 . | in this paper we present an explicit construction of a relative symplectic packing . this confirms the sharpness of the upper bound for the relative packing of a ball into the pair @xmath0 of the standard complex projective plane and the clifford torus , obtained by biran and cornea . |
the study of collective motion in systems consisting of a large number of agents , such as bird flocks , fish schools , suspensions of active swimmers ( bacteria , sperm cells ) , etc has triggered an intense literature in the recent years .
we refer to @xcite for recent reviews on the subject .
many of such studies rely on a particle model or individual based model ( ibm ) that describes the motion of each individual separately ( see e.g in @xcite ) . in this work
, we aim to describe dense suspensions of elongated self - propelled particles in a fluid , such as sperm . in such dense suspensions ,
steric repulsion is an essential ingredient of the dynamics .
a large part of the literature is concerned with dilute suspensions @xcite . in these approaches , the stokes equation for the fluid
is coupled to the orientational distribution function of the self - propelled particles .
however , these approaches are of `` mean - field type '' i.e. assume that particle interactions are mediated by the fluid through some kinds of averages . these approaches do not deal easily with short - range interactions such as steric repulsion or interactions mediated by lubrication forces .
additionally , these models assume a rather simple geometry of the swimmers , which are reduced to a force dipole , while the true geometry and motion of an actual swimmer , like a sperm cell , is considerably more complex . in a recent work @xcite ,
peruani et al showed that , for dense systems of elongated self - propelled particles , steric interaction results in alignment . relying on this work , and owing to the fact that the description of swimmer interactions from first physical principles is by far too complex
, we choose to replace the fluid - mediated interaction by a simple alignment interaction of vicsek type @xcite .
in the vicsek model , the agents move with constant speed and attempt to align with their neighbors up to some noise .
many aspects of the vicsek model have been studied , such as phase transitions @xcite , numerical simulations @xcite , derivation of macroscopic models @xcite .
the alignment interaction acting alone may trigger the formation of high particle concentrations .
however , in dense suspensions , volume exclusion prevent such high densities to occur .
when distances between particles become too small , repulsive forces are generated by the fluid or by the direct reaction of the bodies one to each other .
these forces contribute to repel the particles and to prevent further contacts . to model this behavior
, we must add a repulsive force to the vicsek alignment model .
inspired by @xcite we consider the possibility that the particle orientations ( i.e the directions of the self - propulsion force ) and the particle velocities may be different .
indeed , steric interaction may push the particles in a direction different from that of their self - propulsion force .
we consider an overdamped regime in which the velocity is proportional to the force through a mobility coefficient .
the overdamped limit is justified by the fact that the background fluid is viscous and thus the forces due to friction are very large compared to those due to motion .
indeed , for micro size particles , the reynolds number is very small ( @xmath1 ) and thus the effect of inertia can be neglected .
finally , differently from @xcite we consider an additional term describing the relaxation of the particle orientation towards the direction of the particle velocity .
we also take into account a brownian noise in the orientation dynamics of the particles .
this noise may take into account the fluid turbulence for instance .
therefore , the particle dynamics results from an interplay between relaxation towards the mean orientation of the surrounding particles , relaxation towards the direction of the velocity vector and brownian noise . from now on we refer to the above described model as the vicsek model with repulsion . starting from the above described microscopic dynamical system we successively derive mean - field equations and hydrodynamic equations .
mean field equations are valid when the number of particles is large and describe the evolution of the one - particle distribution , i.e. the probability for a particle to have a given orientation and position at a given instant of time .
expressing that the spatio - temporal scales of interest are large compared to the agents scales leads to a singular perturbation problem in the kinetic equation . taking the hydrodynamic limit , ( i.e. the limit of the singular perturbation parameter to zero ) leads to the hydrodynamic model .
hydrodynamic models are particularly well - suited to systems consisting of a large number of agents and to the observation of the system s large scale structures .
indeed , the computational cost of ibm increases dramatically with the number of agents , while that of hydrodynamic models is independent of it . with ibm ,
it is also sometimes quite cumbersome to access observables such as order parameters , while these quantities are usually directly encoded into the hydrodynamic equations .
the derivation of hydrodynamic models has been intensely studied by many authors .
many of these models are based on phenomenological considerations @xcite or derived from moment approaches and ad - hoc closure relations @xcite .
the first mathematical derivation of a hydrodynamic system for the vicsek model has been proposed in @xcite .
we refer to this model to as the self - organized hydrodynamic ( soh ) model .
one of the main contributions of @xcite is the concept of `` generalized collision invariants '' ( gci ) which permits the derivation of macroscopic equations for a particle system in spite of its lack of momentum conservation .
the soh model has been further refined in @xcite .
performing the hydrodynamic limit in the kinetic equations associated to the vicsek model with repulsion leads to the so - called `` self - organized hydrodynamics with repulsion '' ( sohr ) system .
the sohr model consists of a continuity equation for the density @xmath2 and an evolution equation for the average orientation @xmath3 where @xmath4 indicates the spatial dimension .
more precisely , the model reads @xmath5 where @xmath6 the coefficients @xmath7 , @xmath8 , @xmath9 , @xmath10 , @xmath11 , @xmath12 , @xmath13 , @xmath14 are associated to the microscopic dynamics and will be defined later on .
the symbol @xmath15 stands for the projection matrix @xmath16 of @xmath17 on the hyperplane @xmath18 .
the sohr model is similar to the soh model obtained in @xcite , but with several additional terms which are consequences of the repulsive force at the particle level .
the repulsive force intensity is characterized by the parameter @xmath19 . in the case
@xmath20 , the sohr system is reduced to the soh one .
we first briefly describe the original soh model . inserting ( [ macro4_intro ] ) , ( [ macro5_intro ] ) with @xmath20 into ( [ macro1_intro ] ) , ( [ macro2_intro ] )
leads to @xmath21 together with ( [ macro3_intro ] ) .
this model shares similarities with the isothermal compressible navier - stokes ( ns ) equations .
both models consist of a non linear hyperbolic part supplemented by a diffusion term .
( [ macro6_intro ] ) expresses conservation of mass , while eq .
( [ macro7_intro ] ) is an equation for the mean orientation of the particles .
it is not conservative , contrary to the corresponding momentum conservation equation in ns .
the two equations are supplemented by the geometric constraint ( [ macro3_intro ] ) .
this constraint is satisfied at all times , as soon as it is satisfied initially .
the preservation of this property is guaranteed by the projection operator @xmath22 .
a second important difference between the soh model and ns equations is that the convection velocities for the density and the orientation , @xmath23 and @xmath24 respectively are different while for ns they are equal . that @xmath25 is a consequence of the lack of galilean invariance of the model ( there is a preferred frame , which is that of the fluid ) .
the main consequence is that the propagation of sound waves is anisotropic for this type of fluids @xcite .
the first main difference between the soh and the sohr system is the presence of the terms @xmath26 in the expressions of the velocities @xmath27 and @xmath28 . inserting this term in the density eq .
( [ macro1_intro ] ) results in a diffusion - like term @xmath29 which avoids the formation of high particle concentrations .
this term shows similarities with the non - linear diffusion term in porous media models .
similarly , inserting the term @xmath26 in the orientation eq .
( [ macro2_intro ] ) results in a convection term in the direction of the gradient of the density .
its effect is to force particles to change direction and move towards regions of lower concentration .
the second main difference is the replacement of the linear ( with respect to @xmath2 ) pressure term @xmath30 by a nonlinear pressure @xmath31 in the orientation eq .
( [ macro2_intro ] ) .
the nonlinear part of the pressure enhances the effects of the repulsion forces when concentrations become high . to further establish the validity of the sohr model ( [ macro1_intro])-([macro5_intro ] )
, we perform numerical simulations and compare them to those of the underlying ibm . to numerically solve the sohr model ,
we adapt the relaxation method of @xcite . in this method ,
the unit norm constraint ( [ macro3_intro ] ) is abandonned and replaced by a fully conservative hyperbolic model in which @xmath32 is supposed to be in @xmath17 .
however , at the end of each time step of this conservative model , the vector @xmath32 is normalized .
motsch and navoret showed that the relaxation method provides numerical solutions of the soh model which are consistent with those of the particle model .
the resolution of the conservative model can take advantage of the huge literature on the numerical resolution of hyperbolic conservation laws ( here specifically , we use @xcite ) .
we adapt the technique of @xcite to include the diffusion fluxes . using these approximations ,
we numerically demonstrate the good convergence of the scheme for smooth initial data and the consistency of the solutions with those of the particle vicsek model with repulsion .
the outline of the paper is as follows . in section 2 ,
we introduce the particle model , its mean field limit , the scaling and the hydrodynamic limit . in section 3 we present the numerical discretization of the sohr model , while in section 4 we present several numerical tests for the macroscopic model and a comparison between the microscopic and microscopic models .
section 5 is devoted to draw a conclusion .
some technical proofs will be given in the appendices .
we consider a system of @xmath33-particles each of which is described by its position @xmath34 , its velocity @xmath35 and its direction @xmath36 where @xmath37 , @xmath4 is the spatial dimension and @xmath38 denotes the unit sphere .
the particle ensemble satisfies the following stochastic differential equations @xmath39 eq .
( [ overdamped1 ] ) simply expresses the spatial motion of a particle of velocity @xmath40 .
eq ( [ overdamped2 ] ) shows that the velocity @xmath40 is composed of two components : a self - propulsion velocity of constant magnitude @xmath9 in direction @xmath41 and a repulsive force proportional to the gradient of a potential @xmath42 with mobility coefficient @xmath10 .
equation ( [ overdamped3 ] ) describes the time evolution of the orientation .
the first term models the relaxation of the particle orientation towards the average orientation @xmath43 of its neighbors with rate @xmath44 .
the second term models the relaxation of the particle orientation towards the direction of the particle velocity @xmath40 with rate @xmath13 .
finally , the last term describes standard independent white noises @xmath45 of intensity @xmath46 .
the symbol @xmath47 reminds that the equation has to be understood in the stratonovich sense . under this condition and
thanks to the presence of @xmath48 , the orthogonal projection onto the plane orthogonal to @xmath49 ( i.e @xmath50 where @xmath51 denotes the tensor product of two vectors and @xmath52 is the identity matrix ) , the orientation @xmath53 remains on the unit sphere .
we assume that @xmath9 , @xmath10 , @xmath44 @xmath13 , @xmath54 are strictly positive constants .
the repulsive potential @xmath42 is the resultant of binary interactions mediated by the binary interaction potential @xmath55 .
it is given by : @xmath56 where the binary repulsion potential @xmath57 only depends on the distance .
we suppose that @xmath58 is smooth ( in particular implying that @xmath59 where the prime denotes the derivative with respect to @xmath60 ) .
we also suppose that @xmath61 in particular implying that @xmath62 as @xmath63 . the quantity @xmath64 denotes the typical range of @xmath65 and the fact that @xmath66 ensures that @xmath65 is repulsive . in the numerical test section , we will propose precise expressions for this potential force .
the mean orientation @xmath67 is defined by @xmath68 it is constructed as the normalization of the vector @xmath69 which sums up all orientation vectors @xmath70 of all the particles which belong to the range of the `` influence kernel '' @xmath71 .
the quantity @xmath72 is the typical range of the influence kernel @xmath73 , which is supposed to depend only on the distance .
it measures how the mean orientation at the origin is influenced by particles at position @xmath74 .
here , we assume that @xmath75 is smooth at the origin and compactly supported .
for instance , if @xmath76 is the indicator function of the ball of radius @xmath77 , the quantity @xmath78 computes the mean direction of the particles which lie in the sphere of radius @xmath79 centered at @xmath74 at time @xmath80 .
\(i ) in the absence of repulsive force ( i.e. @xmath81 ) , the system reduces to the time continuous version of the vicsek model proposed in @xcite .
+ ( ii ) the model presented is the so called overdamped limit of the model consisting of ( [ overdamped1 ] ) and ( [ overdamped3 ] ) and where ( [ overdamped2 ] ) is replaced by : @xmath82 with @xmath83 taking the limit @xmath84 in ( [ nonoverdamped4 ] ) , we obtain ( [ overdamped2 ] ) .
as already mentioned in the introduction , for microscopic swimmers , this limit is justified by the very small reynolds number and the very small inertia of the particles .
we now introduce the mean field kinetic equation which describes the time evolution of the particle system in the large @xmath33 limit .
the unknown here is the one particle distribution function @xmath85 which depends on the position @xmath86 , orientation @xmath87 and time @xmath80 .
the evolution of @xmath88 is governed by the following system @xmath89 where the repulsive potential and the average orientation are given by @xmath90 equation ( [ eqmeanfield1 ] ) is a fokker - planck type equation .
the second term at the left - hand side of ( [ eqmeanfield1 ] ) describes particle transport in physical space with velocity @xmath91 and is the kinetic counterpart of eq .
( [ overdamped1 ] ) .
the third , fourth and fifth terms describe transport in orientation space and are the kinetic counterpart of eq .
( [ overdamped3 ] ) .
the alignment interaction is expressed by the third term , while the relaxation force towards the velocity @xmath92 is expressed by the fourth term .
the fifth term represents the diffusion due to the brownian noise in orientation space .
the projection @xmath48 insures that the force terms are normal to @xmath49 .
the symbols @xmath93 and @xmath94 respectively stand for the divergence of tangent vector fields to @xmath38 and the laplace - beltrami operator on @xmath38 .
( [ eqmeanfield2 ] ) is the direct counterpart of ( [ overdamped2 ] ) .
( [ mfeq1 ] ) is the continuous counterpart of eq .
( [ eq : pot ] ) .
indeed , letting @xmath88 be the empirical measure @xmath95 in ( [ mfeq1 ] ) ( where @xmath96 is the dirac delta at @xmath97 ) leads to ( [ eq : pot ] ) .
similarly , eqs .
( [ mfeq2 ] ) , ( [ mfeq3 ] ) are the continuous counterparts of ( [ eq : meanorient ] ) ( by the same kind of argument ) .
the rigorous convergence of the particle system to the above fokker - planck equation ( [ eqmeanfield1 ] ) is an open problem .
we recall however that , the derivation of the kinetic equation for the vicsek model without repulsion has been done in @xcite in a slightly modified context . in order to highlight the role of the various terms ,
we first write the system in dimensionless form . we chose @xmath98 as unit of time and choose @xmath99 as units of space , distribution function and potential .
we introduce the dimensionless variables : @xmath100 and the dimensionless parameters @xmath101 in the new set of variables @xmath102 , eq .
( [ eqmeanfield2 ] ) becomes ( dropping the tildes and the @xmath103 for simplicity ) : @xmath104 while @xmath88 , @xmath105 , @xmath106 , @xmath107are still given by ( [ eqmeanfield2 ] ) , ( [ mfeq1 ] ) , ( [ mfeq2 ] ) , ( [ mfeq3 ] ) ( now written in the new variables ) .
we now define the regime we are interested in .
we assume that the ranges @xmath79 and @xmath64 of the interaction kernels @xmath76 and @xmath108 are both small but with @xmath79 much larger than @xmath64 . more specifically
, we assume the existence of a small parameter @xmath109 such that : @xmath110 we also assume that the diffusion coefficient @xmath54 and the relaxation rate to the mean orientation @xmath44 are large and of the same orders of magnitude ( i.e. @xmath111 ) , while the relaxation to the velocity @xmath13 stays of order @xmath77 , i.e. @xmath112 with these new notations , dropping all hats , the distribution function @xmath113 ( where the superscript @xmath114 now higlights the dependence of @xmath88 upon the small parameter @xmath114 ) satisfies the following fokker - plank equation @xmath115 where the repulsive potential and the average orientation are now given by @xmath116 now , by taylor expansion and the fact that the kernels @xmath76 , @xmath65 only depend on @xmath60 , we obtain ( provided that @xmath76 is normalized to @xmath77 i.e. @xmath117 ) : @xmath118 where the coefficients @xmath119 are given by @xmath120 for example , if @xmath76 is the indicator function of the ball of radius @xmath121 then @xmath122 where @xmath123 is the volume of the sphere @xmath38 . in the cases @xmath124 and @xmath125 , we respectively get @xmath126 and @xmath127 the local density @xmath128 the local current density @xmath129 and local average orientation @xmath130 are defined by @xmath131 more details about this taylor expansion are given in appendix [ sec : taylorexpasion ] .
let us observe that this scaling , first proposed in @xcite is different from the one used in @xcite and results in the appearance of the viscosity term at the right - hand side of eq .
( [ macro2_intro ] ) .
finally , if we neglect the terms of order @xmath132 and we define the so - called collision operator @xmath133 by @xmath134 the rescaled system ( [ eqmeanfield5 ] ) , ( [ eqmeanfield6 ] ) can be rewritten as follows @xmath135 the aim is now to derive a hydrodynamic model by taking the limit @xmath136 of system ( [ lastscaling1 ] ) , ( [ lastscaling2 ] ) where the local density @xmath128 the local current @xmath129 and the local average orientation @xmath130 are defined by ( [ f1 ] ) , ( [ f2 ] ) .
we first introduce the von mises - fisher ( vmf ) probability distribution @xmath137 of orientation @xmath138 defined for @xmath139 by : @xmath140 an important parameter will be the flux of the vmf distribution , i.e. @xmath141 . by obvious symmetry consideration , we have @xmath142 where the quantity @xmath143 does not depend on @xmath144 , is such that @xmath145 and is given by @xmath146 when @xmath12 is small , @xmath147 is close to a dirac delta @xmath148 and represents a distribution of perfectly aligned particles in the direction of @xmath32 .
when @xmath12 is large , @xmath147 is close to a uniform distribution on the sphere and represents a distribution of almost totally disordered orientations .
the function @xmath149 $ ] is strictly decreasing with @xmath150 , @xmath151 .
therefore , @xmath152 represents an order parameter , which corresponds to perfect disorder when it is close to @xmath153 and perfect alignment order when it is close to @xmath77 .
we have following theorem : [ thm : sm ] let @xmath154 be the solution of ( [ lastscaling1 ] ) , ( [ lastscaling2 ] ) .
assume that there exists @xmath88 such that @xmath155 pointwise as well as all its derivatives .
then , there exist @xmath156 and @xmath157 such that @xmath158 moreover , the functions @xmath159 satisfy the following equations@xmath160 where @xmath161 and the coefficients @xmath162 will be defined in formulas ( [ c1 ] ) , ( [ c2 ] ) below .
going back to unscaled variables , we find the model ( [ macro1_intro])-([macro5_intro ] ) presented in the introduction .
* proof : * the proof of this theorem is divided into three steps : ( i ) determination of the equilibrium states ; ( ii ) determination of the generalized collision invariants ; ( iii ) hydrodynamic limit .
we give a sketch of the proof for each step .
* step ( i ) : determination of the equilibrium states * we define the equilibria as the elements of the null space of @xmath163 , considered as an operator acting on functions of @xmath164 only . [ def : small_equi ] the set @xmath165 of equilibria of @xmath163 is defined by @xmath166 we have the following : the set @xmath165 is given by @xmath167 [ lem : smallkin_equi ] for a proof of this lemma , see @xcite .
the proof relies on writing the collision operator as @xmath168 * step ( ii ) : generalized collision invariants ( gci ) .
* we begin with the definition of a collision invariant .
a collision invariant ( ci ) is a function @xmath169 such that for all functions @xmath170 with sufficient regularity we have @xmath171 we denote by @xmath172 the set of ci .
the set @xmath172 is a vector space .
[ def : ci ] as seen in @xcite , the space of ci is one dimensional and spanned by the constants .
physically , this corresponds to conservation of mass during particle interactions .
since energy and momentum are not conserved , we can not hope for more physical conservations .
thus the set of ci is not large enough to allow us to derive the evolution of the macroscopic quantities @xmath2 and @xmath173 to overcome this difficulty , a weaker concept of collision invariant , the so - called `` generalized collisional invariant '' ( gci ) has been introduced in @xcite . to introduce this concept ,
we first define the operator @xmath174 , which , for a given @xmath138 , is given by @xmath175 we note that @xmath176 and that for a given @xmath138 , the operator @xmath177 is a linear operator .
then we have the let @xmath138 be given .
a generalized collision invariant ( gci ) associated to @xmath32 is a function @xmath178 which satisfies : @xmath179 we denote by @xmath180 the set of gci associated to @xmath32 .
[ def : gci ] the following lemma characterizes the set of generalized collision invariants .
there exists a positive function @xmath181 : @xmath182 \to { \mathbb r}$ ] such that @xmath183 the function @xmath181 is such that @xmath184 and @xmath185 is the unique solution in the space @xmath28 defined by @xmath186 ( denoting by @xmath187 the sobolev space of functions which are square integrable as well as their derivative and vanish at the boundary ) of the problem @xmath188 the set @xmath180 is a @xmath4-dimensional vector space . for a proof
we refer to @xcite for @xmath189 and to @xcite for general @xmath190 .
we denote by @xmath191 the vector gci @xmath192 we note that , thanks to ( [ eq : calq_def ] ) and ( [ eq : sm_def_gci ] ) , we have @xmath193 * step ( iii ) : hydrodynamic limit @xmath194 * in the limit @xmath136 , we assume that ( [ eq : converg ] ) holds .
then , thanks to ( [ lastscaling1 ] ) , we have @xmath195 . in view of lemma [ lem : smallkin_equi ] , this implies that @xmath88 has the form ( [ f0 ] ) .
we now need to determine the equations satisfied by @xmath2 and @xmath32 .
for this purpose , we divide eq .
( [ lastscaling1 ] ) by @xmath196 and integrate it with respect to @xmath49 . writing ( [ lastscaling1 ] ) as @xmath197 where @xmath198 we observe that the integral of @xmath199 and @xmath200over @xmath49 is zero since it is in divergence form and the integral of the right - hand side of ( [ eq : kin_reduced ] ) is zero since @xmath77 is a ci .
the integral of @xmath201 gives : @xmath202 we take the limit @xmath136 and use ( [ eq : converg ] ) to get eq .
( [ macro1 ] ) with @xmath203 using ( [ lastscaling2 ] ) , we get @xmath204 . with ( [ c1 ] ) , this leads to the first equation ( [ macro21 ] ) .
multiplying ( [ eq : kin_reduced ] ) by @xmath205 , integrating with respect to @xmath49 and using ( [ eq : intgci ] ) , we get @xmath206 and taking the limit @xmath207 we get @xmath208 this equation describes the evolution of the mean direction @xmath173 the computations which lead to ( [ macro2 ] ) are proved in the appendix [ subsec : small_equi ] .
the coefficient @xmath8 in ( [ macro2 ] ) is defined by @xmath209 where for any function @xmath210 we denote @xmath211 by @xmath212 the sohr model ( [ macro1 ] ) , ( [ macro2 ] ) can be rewritten as follows @xmath213 where the vectors @xmath214 and the function @xmath215 are defined by @xmath216 and where the primes denote derivatives with respect to @xmath2 .
this writing displays this system in the form of coupled nonlinear advection - diffusion equations .
in this section , we develop the numerical approximation of the system ( [ macro1])-([macro22 ] ) in the two dimensional case . as mentioned above , this system is not conservative because of the geometric constraint @xmath217 weak solutions of non - conservative systems are not unique because jump relations across discontinuities are not uniquely defined .
this indeterminacy can not be waived by means of an entropy inequality , by contrast to the case of conservative systems . in @xcite
the authors address this problem for the soh model .
they show that the model is a zero - relaxation limit of a conservative system where the velocity @xmath32 is non - constrained ( i.e. belongs to @xmath17 ) . additionally , they show that the numerical solutions build from the relaxation system are consistent with those of the underlying particle model , while other numerical solutions built directly from the soh model are not . here
we extend this idea to the sohr model .
more precisely , we introduce the following relaxation model ( in dimension @xmath218 ) : @xmath219 the left - hand sides form a conservative system
. we get the following proposition : [ prop1 ] the relaxation model ( [ relax1])-([relax23 ] ) converges to the sohr model ( [ macro1])-([macro22 ] ) as @xmath220 goes to zero .
the proof of proposition [ prop1 ] is given in appendix [ numerical : convergence ] .
this allows us to use well - established numerical techniques for solving the conservative system ( i.e. the left - hand side of ( [ relax1 ] ) , ( [ relax2 ] ) ) .
the scheme we propose relies on a time splitting of step @xmath221 between the conservative part @xmath222 and the relaxation part @xmath223 system ( [ split1]-[split2 ] ) can be rewritten in the following form ( we omit the superscript @xmath224 for simplicity ) @xmath225 where @xmath226 @xmath227 we consider now the following numerical scheme where we denoted @xmath228 the approximation of @xmath163 at time @xmath229 and position @xmath230 : @xmath231 where the numerical flux @xmath232 is given by @xmath233 with @xmath234 and the analogous discretization holds for @xmath235 . in the above formula ,
@xmath236 is a polynomial of matrices of degree @xmath237 calculated with the eigenvalues of the jacobian matrices @xmath238 at an intermediate state depending on @xmath239 and @xmath240 as detailed in @xcite . to ensure stability of the scheme
, the time step @xmath221 satisfies a courant - friedrichs - lewy ( cfl ) condition computed as the minimum of the cfl conditions required for the hyperbolic and diffusive parts of the system .
once the approximate solution of the conservative system is computed , equations ( [ relax3 ] ) and ( [ relax4 ] ) can be solved explicitly . in the limit @xmath241
they give @xmath242 where @xmath243 is the numerical solution of system ( [ split1]-[split2 ] ) .
this ends one step of the numerical scheme for the system ( [ relax1]-[relax2 ] ) .
the goal of this section is to present some numerical solutions of the system ( [ macro1])-([macro22 ] ) which validate the numerical scheme proposed in the previous section .
we will first perform a convergence test .
we then successively compare the solutions obtained with the sohr model with those computed by numerically solving the individual based model ( [ overdamped1 ] ) in regimes in which the two models should provide similar results .
we will finally perform some comparisons between the soh and the sohr system to highlight the difference between the two models .
we will compare the sohr model with another way to incorporate repulsion in the soh model , the so - called dlmp model of @xcite .
for all the tests , we use the model in uscaled variables as described in the introduction ( see ( [ macro1_intro])-([macro5_intro ] ) .
the potential kernel @xmath65 is chosen as @xmath244 which gives @xmath245 , while for @xmath76 , by assumption normalized to @xmath121 we choose the following form @xmath246 this leads to @xmath247 the other parameters , which are fixed for all simulations if not differently stated , are : @xmath248 which , in dimension @xmath218 , lead to ( after numerically computing the gci and the associated integrals ) : @xmath249 in the visualization of the results , we will use the angle @xmath250 of the vector @xmath32 relative to the @xmath74-axis , i.e. @xmath251 .
the first test is targeted at the validation of the proposed numerical scheme . for this purpose
, we investigate the convergence when the space step @xmath252 tends to @xmath253 , refining the grid and checking how the error behaves asymptotically .
the initial mesh size is @xmath254 while the time step is @xmath255 we repeat the computation for @xmath256 @xmath257 @xmath258 .
the convergence rate is estimated through the measure of the @xmath259 norm of the error for the vectors @xmath260 by using for each grid the next finer grid as reference solution .
the initial data is @xmath261 where @xmath262 the boundary conditions are fixed in time on the four sides of the square : @xmath263 .
the error curves for the density and for @xmath264 are plotted in figure [ fig : scheme ] as a function of the space step in log - log scale at time @xmath265 the slope of the error curves are compared to a straight line of slope 1 . from the figure , we observe the convergence of the scheme with accuracy close to @xmath266 -error for the density @xmath2 and the flux direction @xmath267 as a function of @xmath268 in log - log scale .
a straight line of slope @xmath77 is plotted for reference .
this figure shows that the scheme is numerically of order @xmath77 .
, width=415,height=302 ] in this subsection , we validate the sohr model by comparing it to the vicsek model with repulsion on two different test cases .
we investigate the convergence of the microscopic ibm to the macroscopic sohr model when the scaling parameter @xmath196 tend to zero .
the scaled ibm is written : @xmath269 the solution of the individual based model ( [ overdamped1]-[overdamped3 ] ) is computed by averaging different realizations in order to reduce the statistical errors .
the coefficient of the ibm are fixed to @xmath270 for the repulsive range , @xmath271 for the alignment interaction range , while @xmath272 particles are used for each simulation .
the details of the particles simulation can be found in @xcite for classical particle approaches or in @xcite for a direct application to the soh model .
[ [ riemann - problem ] ] riemann problem : + + + + + + + + + + + + + + + + the convergence of the two models is measured on a riemann problem with the following initial data @xmath273 and with periodic boundary condition in @xmath74 and @xmath274 .
the parameters of the sohr model are : @xmath275 . in figure
[ fig : macro_micro ] we report the relative @xmath259 norm of the error for the macroscopic quantities @xmath276 between the sohr model and the particle model with respect to the number of averages for different values of @xmath196 : @xmath277 ( x - mark ) , @xmath278 ( plus ) , @xmath279 ( circle ) , @xmath280 ( square ) at time @xmath265 this figure shows , as expected , that the distance between the two solutions goes to zero when @xmath196 goes to zero . in figure
[ fig : solution_rm ] we report the density @xmath2 and the flux direction @xmath250 for the same riemann problem along the @xmath74-axis for @xmath280 at time @xmath281 , the solution being constant in the @xmath274-direction .
again we clearly observe that the two models provide very close solutions , the small differences being due to the different numerical schemes employed for their discretizations .
[ [ taylor - green - vortex - problem ] ] taylor - green vortex problem : + + + + + + + + + + + + + + + + + + + + + + + + + + + + in this third test case , we compare the numerical solutions provided by the two models in a more complex case .
the initial data are @xmath282 where the vector @xmath283 is given by @xmath284 with periodic boundary conditions in both directions .
the numerical parameters for the sorh model are : @xmath285 @xmath286 , while for the particle simulations we choose : @xmath272 particles , @xmath287 @xmath288 @xmath289 in figure [ fig : dens_macro_micro_t06_gt ] and [ fig : velo_macro_micro_t06_gt ] , we report the density @xmath2 and the flux direction @xmath144 at time @xmath290s . in both figures , the left picture is for the ibm and the right one for the sohr model .
again , we find a very good agreement between the two models in spite of the quite complex structure of the solution . in this part , we show the difference between the soh system ( [ macro6_intro ] ) , ( [ macro7_intro ] ) and the sohr one for different values of the repulsive force @xmath291 we recall that the sohr model reduces to the soh one in the case in which the repulsive force is set equal to zero . to this aim , we rescale the repulsive force @xmath11 by @xmath292 and then we let @xmath293 vary .
the repulsive potential @xmath65 is still given by ( [ phi ] ) , so that @xmath294 .
the other numerical parameters are chosen as follows : @xmath295 @xmath296 @xmath297 @xmath298 @xmath299 @xmath300 @xmath301 the initial data are those of the vortex problem ( [ vortex ] ) except that we start with four vortices instead of only one .
periodic boundary conditions in both directions are used .
figure ( [ fig : solution_f5 ] ) displays the solutions for the sohr system for the density ( left ) and for the flux direction ( right ) at @xmath302 with @xmath303 figure ( [ fig : solution_f005 ] ) displays the solutions for @xmath304 the results are almost undistinguishable to those of the soh model ( @xmath305 ) and are not shown for this reason .
these figures show that when the repulsive force is large enough , the sohr model can prevent the formation of high concentrations .
by contrast , when this force is small , the sohr model becomes closer to the soh one and high concentrations become possible .
in this final part , we want to compare the sohr system to the hydrodynamic model proposed by degond , liu , motsch and panferov in @xcite ( referred to as dlmp model ) .
this model is derived , in a similar fashion as the sohr model , starting from a system of self - propelled particles which obey to alignment and repulsion .
the main difference is that in the dlmp model , the particle velocity is exactly equal to the self propulsion velocity but the particles adjust their orientation to respond to repulsion as well as alignment .
the resulting model is of soh type and is therefore written ( [ macro6_intro ] ) , ( [ macro7_intro ] ) , but with an increased coefficient in front of the pressure term @xmath306 , this coefficient being equal to @xmath307 .
the initial conditions and numerical parameters are the same as in previous test in figure [ fig : solution_dlmp_f5 ] , we report the density @xmath2 ( left ) and the flux direction @xmath144 ( right ) for @xmath308 for the dlmp model . comparing figures ( [ fig : solution_f5 ] ) with figure [ fig : solution_dlmp_f5 ] , we observe that the solutions of the sohr and of the dlmp model are different .
the homogenization of the density seems more efficient with the sohr model than with the dlmp model .
this can be attributed to the effect of the nonlinear diffusion terms that are included in the sohr model but not in the dlmp model .
therefore , the way repulsion is included in the models may significantly affect the qualitative behavior of the solution . in practical situations ,
when the exact nature of the interactions is unknown , some care must be taken to choose the right repulsion mechanism .
in this paper , we have derived a hydrodynamic model for a system of self - propelled particles which interact through both alignment and repulsion . in the underlying particle model
, the actual particle velocity may be different from the self - propulsion velocity as a result of repulsion interactions with the neighbors .
particles update the orientation of their self - propulsion seeking to locally align with their neighbors as in vicsek alignment dynamics .
the corresponding hydrodynamic model is similar to the self - organized hydrodynamic ( soh ) system derived from the vicsek particle model but it contains several additional terms arising from repulsion .
these new terms consist principally of gradients of linear or nonlinear functions of the density including a non - linear diffusion similar to porous medium diffusion .
this new self - organized hydrodynamic system with repulsion ( sohr ) has been numerically validated by comparisons with the particle model .
it appears more efficient to prevent high density concentrations than other approaches based on simply enhancing the pressure force in the soh model . in future work , this model will be used to explore self - organized motion in collective dynamics . to this effect
, it will be calibrated on data based on biological experiments , such as recordings of collective sperm - cell motion .
99 m. aldana , v. dossetti , c. huepe , v. m. kenkre , h. larralde , phase transitions in systems of self - propelled agents and related network models , phys .
lett . , 98 ( 2007 ) , 095702 .
i. aoki , a simulation study on the schooling mechanism in fish , bulletin of the japan society of scientific fisheries , 48 ( 1982 ) 1081 - 1088 .
a. baskaran , m. c. marchetti , nonequilibrium statistical mechanics of self - propelled hard rods , j. stat .
mech . theory exp . , (
2010 ) p04019 .
e. bertin , m. droz and g. grgoire , hydrodynamic equations for self - propelled particles : microscopic derivation and stability analysis , j. phys .
theor . , 42 ( 2009 ) 445001 .
f. bolley , j. a. caizo , j. a. carrillo , mean - field limit for the stochastic vicsek model , appl .
lett . , 25 ( 2011 ) 339 - 343 .
h. chat , f. ginelli , g. grgoire , f. raynaud , collective motion of self - propelled particles interacting without cohesion , phys .
e 77 ( 2008 ) 046113 .
chuang , m. r. dorsogna , d. marthaler , a. l. bertozzi and l. s. chayes , state transitions and the continuum limit for a 2d interacting , self - propelled particle system , physica d , 232 ( 2007 ) 33 - 47 .
i. d. couzin , j. krause , r. james , g. d. ruxton and n. r. franks , collective memory and spatial sorting in animal groups , j. theor .
biol . , 218 ( 2002 ) , 1 - 11 .
f. cucker , s. smale , emergent behavior in flocks , ieee transactions on automatic control , 52 ( 2007 ) 852 - 862 .
p. degond , a. frouvelle , j - g .
liu , macroscopic limits and phase transition in a system of self - propelled particles , j. nonlinear sci .
23 ( 2013 ) , 427 - 456 .
p. degond , a. frouvelle , j - g .
liu , phase transitions , hysteresis , and hyperbolicity for self - organized alignment dynamics , preprint .
arxiv:1304.2929 .
p. degond , j - g .
liu , s. motsch , v. panferov , hydrodynamic models of self - organized dynamics : derivation and existence theory , methods appl .
, 20 ( 2013 ) 089 - 114 .
p. degond , s. motsch , continuum limit of self - driven particles with orientation interaction , math .
models methods appl .
, 18 suppl .
( 2008 ) 1193 - 1215 .
p. degond , p. f. peyrard , g. russo , ph .
villedieu , polynomial upwind schemes for hyperbolic systems , c. r. acad .
paris , ser i , 328 ( 1999 ) 479 - 483 h.fehske , r. schneider and a. weisse , computational many - particle physics , springer verlag , 2007 .
a. frouvelle , a continuum model for alignment of self - propelled particles with anisotropy and density - dependent parameters , math .
, 22 ( 2012 ) 1250011 ( 40 p. ) .
g. grgoire , h. chat , onset of collective and cohesive motion , phys .
lett . , 92 ( 2004 ) 025702 .
s. henkes , y. fily , m. c. marchetti , active jamming : self - propelled soft particles at high density , phys .
e 84 ( 2011 ) 040301 .
j. p hernandez - ortiz , p. t underhill , m. d. graham , dynamics of confined suspensions of swimming particles , j. phys .
: condens .
matter 21 ( 2009 ) 204107 .
r. w hockney and j. w eastwood , computer simulation using particles .
institute of physics publishing , 1988 .
d. l .koch , g. subramanian , collective hydrodynamics of swimming microorganisms : living fluids , annu .
fluid mech . , 43 ( 2011 ) 637 - 59 .
a. mogilner , l. edelstein - keshet , l. bent and a. spiros , mutual interactions , potentials , and individual distance in a social aggregation , j. math .
biol . , 47 ( 2003 ) 353 - 389 .
s. motsch , l. navoret , numerical simulations of a non - conservative hyperbolic system with geometric constraints describing swarming behavior , multiscale model .
9 ( 2011 ) 1253 - 1275 .
s. motsch , e. tadmor , a new model for self - organized dynamics and its flocking behavior , j. stat .
phys . , 144 ( 2011 ) 923 - 947 .
t. j. pedley , n. a. hill , j. o. kessler , the growth of bioconvection patterns in a uniform suspension of gyrotactic micro - organisms , j. fluid .
mech . , 195 ( 1988 ) 223 - 237 .
f. peruani , a. deutsch , m. br , nonequilibrium clustering of self - propelled rods , phys .
e 74 ( 2006 ) 030904(r ) . v. i. ratushnaya , d. bedeaux , v. l. kulinskii , a. v. zvelindovsky , collective behavior of self propelling particles with kinematic constraints : the relations between the discrete and the continuous description , phys . a 381 ( 2007 ) 39 - 46 .
d. saintillan , m.j .
shelley , instabilities , pattern formation and mixing in active suspensions , phys.fluids 20 ( 2008 ) 123304 .
b.szab , g.j szllsi , b. gnci , zs .
jurnyi , d. selmeczi , and t. vicsek phase transition in the collective migration of tissue cells : experiment and model , phys . rev .
lett , 74 ( 2006 ) 061908 .
j. toner , y. tu and s. ramaswamy , hydrodynamics and phases of flocks , annals of physics , 318 ( 2005 ) 170 - 244 t. vicsek , a. czirk , e. ben - jacob , i. cohen , o. shochet , novel type of phase transition in a system of self - driven particles , phys .
, 75 ( 1995 ) 1226 - 1229 .
t. vicsek , a. zafeiris , collective motion , phys . rep .
, 517 ( 2012 ) 71 - 140 .
f. g. woodhouse , r. e. goldstein , spontaneous circulation of confined active suspensions , phys .
109 ( 2012 ) 168105 .
by introducing the change of variable @xmath309 and using taylor expansion , we get @xmath310 where @xmath14 is given by ( [ k0 ] ) and @xmath311 is the hessian matrix of @xmath88 with respect to the variable @xmath74 . here
, we have used that the @xmath312 and @xmath313 terms vanish after integration in @xmath314 by oddness with respect to @xmath314 . by the same computation for the kernel @xmath65
, we have @xmath315 with @xmath316
we prove that ( [ vitesse ] ) leads to ( [ macro2 ] ) .
thanks to ( [ eq : psi ] ) , eq .
( [ vitesse ] ) can be written : @xmath317 where @xmath318 , @xmath319 are given by ( [ eq : t123 ] ) .
now , @xmath320 can be written : @xmath321 we recall that the first two terms of @xmath322 at the right hand side of ( [ t1 ] ) and the corresponding terms in @xmath323 have been computed in @xcite .
the computation for the third term of @xmath324 is easy and we get : @xmath325 where the coefficients are given by @xmath326 now observe that for a constant vector @xmath327 , we have @xmath328 thus , using ( [ eq : t123 ] ) , ( [ eq : nablaoma ] ) and the chain rule , we get for @xmath329 @xmath330 finally , we obtain : @xmath331 where @xmath332 the terms @xmath333 and @xmath334 have been computed in @xcite . in particular , it is easy to see that we get them from the formulae for @xmath335 and @xmath336 by changing @xmath337 into @xmath338
. therefore , we get : @xmath339 where @xmath340 inserting the expressions of @xmath341 and @xmath334 into ( [ eq2 ] ) we get ( [ macro2 ] ) .
we follow the lines of the proof of proposition 3.1 of @xcite .
assume that @xmath342 and @xmath343 as @xmath224 tends to zero .
then , set @xmath344 multiplying equation ( [ relax2 ] ) by @xmath224 and then taking the limit @xmath345 yields @xmath346 it follows that @xmath347 since the vector @xmath348 is parallel to @xmath349 we have @xmath350 which implies that @xmath351 therefore , letting @xmath352 we obtain @xmath353 where @xmath354 is a real number , @xmath355 @xmath356 and @xmath357 . by taking the scalar product of ( [ relax33 ] ) with @xmath358 , we get @xmath359 inserting this expression of @xmath354 into ( [ relax33 ] ) we find the equation for the evolution of the average direction ( [ macro2 ] ) and thus the sohr model ( [ macro1])-([macro22 ] ) . | we study a system of self - propelled particles which interact with their neighbors via alignment and repulsion .
the particle velocities result from self - propulsion and repulsion by close neighbors .
the direction of self - propulsion is continuously aligned to that of the neighbors , up to some noise .
a continuum model is derived starting from a mean - field kinetic description of the particle system .
it leads to a set of non conservative hydrodynamic equations .
we provide a numerical validation of the continuum model by comparison with the particle model .
we also provide comparisons with other self - propelled particle models with alignment and repulsion
. 1- department of mathematics , imperial college london , + london sw7 2az , united kingdom .
+ email:[email protected]
+ @xmath0 + 2-department of mathematics , university of ferrara , 44100 ferrara , italy + email : [email protected] + @xmath0 + 3-universit de toulouse ; ups , insa , ut1 , utm ; + institut de mathmatiques de toulouse ; + f-31062 toulouse , france .
+ @xmath0 + 4-cnrs ; institut de mathmatiques de toulouse umr 5219 ; + f-31062 toulouse , france . +
email : [email protected] + @xmath0 + 5-national university of singapore , department of mathematics , + lower kent ridge road , singapore 119076 . +
email : [email protected] + * acknowledgements : * pd is on leave from cnrs ,
institut de mathmatiques de toulouse , france , and gd is on leave from universit paul sabatier , institut de mathmatiques de toulouse , france , where part of this research was conducted .
tbnm wishes to thank the hospitality of the department of mathematics , imperial college london for its hospitality .
nw wishes to thank the universit paul sabatier ,
institut de mathmatiques de toulouse , france , for its hospitality .
this work has been supported by the french agence nationale pour la recherche ( anr) in the frame of the contract motimo ( anr-11-monu-009 - 01 ) .
* key words : * fokker - planck equation , macroscopic limit , von mises - fisher distribution , generalized collision invariants , non - conservative equations , self - organized hydrodynamics , self - propelled particles , alignment , repulsion . *
ams subject classification : * 35q80 , 35l60 , 82c22 , 82c70 , 92d50 0.4 cm |
the issue of charge symmetry violation ( csv ) is fundamental to our understanding of hadronic interactions , and many experimental and theoretical studies have addressed this issue ( see review @xcite ) . in the framework of qcd , csv arises from the mass difference between the @xmath11 and @xmath12 quarks .
the other principal cause for csv comes from the electromagnetic interaction .
weinberg @xcite pointed out that the effective chiral @xmath13 lagrangian , coming from qcd , contains a term which violates charge symmetry ( see also , a recent review by meissner @xcite ) .
thus , not only are there kinematic reasons for csv due to the mass differences within baryon multiplets , but direct csv effects should exist as well .
recently , gashi _ et al . _
@xcite analyzed low - energy @xmath13 scattering data , and found some indications for direct csv effects in the strong - interaction sector . another way to study
csv is through the pion - nuclear interaction in the lightest nuclei , particularly via isomirror elastic scattering . for the deuteron case ,
the @xmath14 cross section is compared with that for @xmath15 in masterson _
_ @xcite and baru _ et al . _
@xcite , but only small differences are found . for pion elastic scattering from @xmath2 and @xmath3 , one can consider three mirror ratios formed from the differential cross sections : @xmath16 where @xmath6 is refereed to as the superratio " @xcite .
the @xmath17 and @xmath18 scattering cross sections are isomirror ones , as are @xmath19 and @xmath20 .
if charge symmetry were conserved , all three ratios would be equal to unity .
of course , the coulomb interaction is not charge symmetric and has to be taken into account .
the experimental study of these ratios has been concentrated on large - angle scattering because the coulomb interaction , which intrinsically violates charge symmetry , makes a significant contribution in the forward - scattering region . in a series of lampf experiments @xcite , the ratios ( [ 1 ] )
were measured in the range of the @xmath21 @xmath13 resonance .
significant deviation ( several standard deviations ) from unity ( up to @xmath2220% ) was observed for @xmath5 and @xmath6 in the angular range outside of the coulomb cone , @xmath23 .
in addition , strong angular dependence of both @xmath5 and @xmath6 was observed in the angular interval between @xmath24 and @xmath25 .
these csv effects are much more pronounced for @xmath2 and @xmath3 than for @xmath26 , where the observed asymmetry @xmath27 elastic scattering have been evaluated in terms of the asymmetry @xmath27 : @xmath28 if we define @xmath29 by analogy with the ratios ( [ 1 ] ) , we get @xmath30 and for small @xmath31 we have @xmath32 .
thus , @xmath33 . ] is not nearly as large : it is only about 2% ( see , for example , ref .
in addition , inelastic - scattering data for the ratios ( [ 1 ] ) at excitation energy below 20 mev have been reported in ref .
@xcite , and elastic differential cross sections at back angles in ref .
@xcite . thus , csv effects manifest themselves clearly in the three - nucleon system , even if they are very small in the deuteron case . for this reason ,
the main goal of our theoretical analysis is the elucidation of the mechanisms that enhance csv in the three - nucleon system . in other words
, can the observed enhancement be due only to the well - known mass differences of the hadron multiplets together with the coulomb interaction ?
previous studies have indicated that the main reason for csv in the case of the deuteron is the mass difference of the charge @xmath34-isobar states .
the influence of this effect on the scattering amplitude of the @xmath35 elastic scattering was discussed in the papers by masterson _
_ @xcite and baru _ et al . _
@xcite for single and single - with - double scattering , respectively . naturally , as the number of multiple - scattering diagrams increases with an increasing number of nucleons , one expects the effect of the @xmath10-mass splitting to be more prominent for @xmath2 and @xmath3 than for @xmath26 .
in addition to the interaction between charged pions and nuclei due to the external coulomb force , there is an internal coulomb interaction due to the difference in the wave functions ( wfs ) of @xmath2 and @xmath3 .
( note that in terms of the strong interaction , there is no difference between the wfs of @xmath2 and @xmath3 . )
one difference in these wfs arises from the additional coulomb repulsion between the two protons in @xmath3 , which is not present in @xmath2 . in ref .
@xcite , the difference in the structure of @xmath2 and @xmath3 has been described by a first order optical potential in the pion - nuclear interaction in which the @xmath2 and @xmath3 charge and magnetic form factors extracted from experimental data on elastic electron scattering from @xmath2 and @xmath3 were used .
a more detailed analysis of pion-@xmath36 elastic and inelastic scattering was reported in ref .
@xcite , where an optical single - scattering pion - nuclear potential , calculated at a microscopic level with a realistic three - nucleon wave function , was used .
no reasonable description of the csv effects was achieved when the calculations were performed with isospin - symmetric wave functions for both nuclei .
a different approach was used in ref .
@xcite , where the difference in the wfs of @xmath2 and @xmath3 was taken into account explicitly . in all of these approaches @xcite , the difference in the structure of the three - nucleon system results in a sharp enhancement of the ratios @xmath5 and @xmath6 in the angular range @xmath37 .
one of the reasons for this behavior of the angular distribution is that the elastic differential cross section in the single - scattering approach contains a significant contribution at small scattering angles and has a minimum at @xmath38 which is related to the dip in the @xmath13 non - spin - flip amplitude .
therefore , even a small contribution to the interaction that violates charge symmetry ( _ e.g. _ , the coulomb interaction ) can produce an enhanced effect .
the influence of the @xmath10-mass splitting on the observed csv effect might also result in a large effect in this angular range .
moreover , to understand the angular distribution in detail , one must look beyond the single - scattering approach to the @xmath13 interaction .
we therefore examine the contributions of both single- and double-@xmath13 scattering to the pion - nuclear scattering amplitude following techniques developed in ref .
we then take csv effects into account to obtain expressions for the ratios ( [ 1 ] ) , which then are compared with the experimental data @xcite .
an analysis of the experimental status for the ratios ( [ 1 ] ) at energies spanning the @xmath8 resonance is given in section [ sec : expt ] . in section [ sec : theor1 ] , we explain how the basic ingredients of the scattering amplitude and the constraints of single and double scattering are combined for @xmath39 and @xmath40 elastic scattering . in section [ sec : theor2 ] , we derive expressions for the scattering amplitudes , taking into account all three effects responsible for csv . in section [ sec : theor3 ] , we discuss the influence of these factors on the ratios ( [ 1 ] ) to show the effect of the individual csv factors , and we compare the results of our calculations with the data . in section [ sec : add ] , we discuss some related issues associated with csv effects in other nuclei over broader ranges in energy and scattering angle . in section [ sec : conc ] , we summarize our findings .
csv in pion - nucleus scattering was first claimed to have been observed in the difference of total @xmath41 cross sections measured at psi @xcite , and ascribed to the @xmath10-mass splitting .
essentially , the total cross section is mainly determined by the forward scattering amplitude , which at small angles can be approximated well by single scattering . in this approach ,
the different charge states of the @xmath10 are excited in @xmath0 and @xmath1 scattering on the deuteron , and the result is the small observed csv effect .
this has been discussed widely ( see _ e.g. _ , the book by ericson and weise @xcite ) .
the situation with the observation of csv effects in the @xmath41 differential cross sections is less clear .
the first systematic study of the csv effect in the differential @xmath41 cross sections was done at lampf @xcite .
there have been several subsequent measurements at lampf and triumf for both @xmath14 and @xmath15 ( see ref .
@xcite for details ) .
the experimental data weakly suggest a small effect in the asymmetry @xmath27 for the deuteron .
for example , in ref .
@xcite , an asymmetry @xmath42 at 143 mev near @xmath25 in the cm frame is reported .
however , smith _ et al . _
@xcite , in an independent measurement done at triumf , reported asymmetries of @xmath431.5 @xmath44 0.6% at back angles and various energies .
thus , the magnitude of csv is at most 1@xmath432% , the sign is uncertain , and the experimental uncertainties are only slightly less than the @xmath27 values themselves . at about the same time , however , measurements of the ratios ( [ 1 ] ) for @xmath2 and @xmath3 at lampf obtained significantly larger effects . the first evidence for a sizable csv in the differential @xmath45 cross sections ) .
] below and at the @xmath10 resonance was seen for the range of cm scattering angles between @xmath46 and @xmath47 @xcite .
the effect seems to peak near @xmath48 in the cm frame ( _ e.g. _ , @xmath49 at 180 mev @xcite ) .
the experiment was repeated with better statistics and systematics for approximately the same range of scattering angles at energies spanning the @xmath10 @xcite and beyond @xcite .
these measurements also were extended to backward angles from @xmath50 to @xmath51 @xcite and are reported in the previous paper @xcite .
the experimental data for all three ratios ( [ 1 ] ) for incident pion energies between 142 and 256 mev are shown in fig .
the agreement between the four data sets , on the whole , is very good .
the bump observed at @xmath52 , corresponding to the minimum in the non - spin - flip amplitude , is obvious in the ratios @xmath5 and @xmath6 for 142 and 180 mev ( below and on the @xmath10 resonance ) .
thus , csv effects in @xmath39 and @xmath40 scattering are large and statistically significant .
the main goal of the present work is to provide a theoretical basis for these large effects in the three - nucleon system .
we formulate the pion - nuclear amplitude in the range of the @xmath10 resonance as a combination of a single- and double - scattering of pions from the nucleons in the nucleus . for the @xmath54 nuclei ,
the appropriate diagrams are shown in fig .
[ f2 ] . the elementary @xmath13 amplitude @xmath55 is taken as the @xmath56 partial wave , as if @xmath13 scattering takes place entirely through the @xmath10 resonance : @xmath57 where @xmath58 , \nonumber\end{aligned}\ ] ] and @xmath59 and @xmath60 are the spin and isospin projection operators for the @xmath13 system for total spin @xmath61 and isospin @xmath61 : @xmath62 ~{\rm and}~ \hat t = \frac{1}{3 } ( 2 + \vec t \cdot\vec{\tau } ) .
\label{3}\end{aligned}\ ] ] here , @xmath63 and @xmath64 are the isospin operators of the pion and nucleon , respectively , @xmath65 and @xmath66 are pauli matrices , @xmath67 and @xmath68 are the unit vectors in the direction of the incoming and outcoming pions in cm frame , respectively , and @xmath69 is the pion momentum in the cm frame .
everywhere below , we use the following notation : @xmath70 . \label{4}\end{aligned}\ ] ] the wf of the @xmath2 and @xmath3 can be written as @xmath71 where @xmath72 and @xmath73 correspond to the spin and isospin parts of the wf , respectively : @xmath74 where @xmath75 and @xmath76 are the spinor and isospinor of the nucleus , respectively , and @xmath77 , @xmath78 , and @xmath79 ( @xmath80 , @xmath81 , @xmath82 ) are the spinors ( isospinors ) of the nucleons of the nucleus ( @xmath83 are cyclic ) . equations ( [ 6 ] ) represent states of pairs of nucleons ( @xmath84 ) such that their total spin @xmath85 and their isospin @xmath86 .
the representation ( [ 5 ] ) for the wfs of @xmath2 and @xmath3 can be generalized for more complicated wave configurations of nuclei , as is discussed in ref .
@xcite .
the coordinate part of the wf is taken into account in a symmetric form corresponding to a simple @xmath87-shell model . in the following calculations , we use two different forms of the radial wf : + i ) the simple gaussian form : @xmath88 , \label{7}\end{aligned}\ ] ] where @xmath89 and @xmath90 , taken from kamalov _ et al .
the slope @xmath89 was chosen in ref .
@xcite by a best fit to the experimental data for the @xmath2 charge form factor below momentum transfer @xmath91 @xcite .
this form of the wf fails to reproduce the minimum of the charge form factor at @xmath92 ; however , the elastic @xmath93 differential cross sections are reproduced well at @xmath7 = 100 and 200 mev , and at backward scattering angles , the results of the calculations with wf ( [ 7 ] ) tend toward the suppression seen in the experimental cross sections @xcite .
+ ii ) the two - component gaussian parametrization : @xmath94 , \label{8}\end{aligned}\ ] ] where @xmath95 , @xmath96 , @xmath97 , @xmath98 , and n is a normalization constant given in appendix [ sec : cff ] .
this wf was successfully used by foursat _
@xcite for the description of the differential cross sections for the reaction @xmath99 at 770 mev for a wide range of scattering angles .
the wf ( [ 8 ] ) reproduces the minimum of the @xmath3 charge form factor at @xmath100 , but shows an enhancement at momentum transfer @xmath101 . + the diagram in fig .
[ f2]a corresponds to the single - scattering approximation for the elastic pion - nuclear scattering amplitude . to calculate this amplitude , we need to compute the matrix element for the operator ( [ 2 ] ) between initial- and final - state wave functions .
we neglect both the fermi motion of the nucleon inside the nucleus and the off - shell corrections to the @xmath13 amplitudes in expression ( [ 2 ] ) .
( we discuss the accuracy of both of these approximations in sec .
[ sec : adc ] . ) taking nuclear wfs in the form ( [ 5 ] ) , we exclude spin and isospin variables of the nucleons . as in ref .
@xcite , the single - scattering amplitude @xmath102 is : @xmath103 where @xmath104 is a nuclear form factor defined by eq .
( [ a4 ] ) and @xmath105 is the three - momentum transfer .
the operator @xmath106 acts on the nuclear spin and isospin variables and is expressed as @xmath107 , \label{10}\end{aligned}\ ] ] where @xmath108 and @xmath109 have been defined in eq .
( [ 4 ] ) . calculating the matrix element from the operator @xmath110 on isotopic variables , we get @xmath111 in terms of @xmath112 , the expression for the differential cross section with unpolarized particles
yields : @xmath113 where @xmath114 where @xmath115 . in additional to a lower form factor @xmath116 , the angular dependence of @xmath117 is determined by the factors of eq .
( [ 13 ] ) . expressions ( [ 12 ] ) and ( [ 13 ] ) show that @xmath118 is suppressed at @xmath119 ( @xmath120 ) , where only the spin - flip @xmath121 amplitude contributes .
thus , for the @xmath122 case , there is a more significant spin - flip suppression than for the deuteron case , where ( see , for example , ref .
@xcite ) @xmath123 and the minimum in the cross section is much weaker .
we note that this kind of suppression of the spin - flip @xmath13 amplitude in the single - scattering term for the @xmath36 case was pointed out in ref .
@xcite , following ref .
@xcite . in figs .
[ f3 ] @xmath43 [ f6 ] , the single - scattering contributions to the differential cross sections for incident pion kinetic energies @xmath7 = 142 , 180 , 220 , and 256 mev , for both versions of the radial part of the wfs ( [ 7 ] ) and ( [ 8 ] ) , are shown by the dashed curves .
let us consider the double - scattering amplitude @xmath124 , as in ref .
@xcite , corresponding to the diagram shown in fig .
( calculations of double - spin - flip amplitudes were performed recently @xcite . ) in the same approximation as was used while calculating the single - scattering term , @xmath125 where the wf in the momentum space operator @xmath126 is defined by eq .
( [ a2 ] ) .
the momenta @xmath127 , @xmath128 , @xmath129 , and @xmath130 relate to the momenta @xmath131 and @xmath132 , shown in fig .
[ f2]b , via @xmath133 , @xmath134 , @xmath135 , and @xmath136 . in eq .
( [ 14 ] ) , @xmath137 is the green s function for the intermediate state , which , neglecting kinetic energy of the intermediate nucleons , has the form @xmath138 where @xmath139 . by analogy with eq .
( [ 4 ] ) , we introduce @xmath140 and @xmath141 , where @xmath142 , for the first and second nucleons : @xmath143 where @xmath144 $ ] , @xmath145 $ ] , and @xmath146 .
the operator @xmath147 then can be written as @xmath148 the operator @xmath149 depends explicitly on the spin and isospin variables of the nuclei and the pion .
@xmath149 also depends on the vector @xmath150 , which is an integrative variable on the right - hand side of eq .
( [ 14 ] ) .
we extract @xmath150 by introducing the operators @xmath151 : @xmath152 thus , the operator @xmath124 can be expressed in the form @xmath153 where the tensor @xmath154 is expressed in the form @xmath155 such that @xmath156 and @xmath157 . here , the quantities @xmath158 and @xmath159 are complex functions which depend on the momentum of the incoming pion @xmath160 and the momentum transfer @xmath161 ( or on the scattering angle @xmath162 ) .
they also depend on the wfs of the nuclei , which are given in appendix [ sec : tensor ] . using eq .
( [ 20 ] ) for the tensors @xmath154 , the amplitude @xmath163 becomes @xmath164 ) .
\label{21}\end{aligned}\ ] ] we divide the contributions to @xmath165 and @xmath166 into double elastic scattering @xmath167 ( we will define this contribution via index @xmath168 " ) and double charge exchange @xmath169 ( we will denote this contribution via index @xmath170 " ) in the functions @xmath165 and @xmath166 : @xmath171 then , we can present expressions for @xmath165 and @xmath166 in terms of the integrals @xmath158 and @xmath159 : @xmath172,~~ b_2^{ee } = \frac{4}{3}f_{p_{33}}^2 ( j_1 + 2 j_2 ) , \phantom{xxxxxxxxxxxxxxxxxx}\cr a_2^{ee } = \frac{1}{3}f_{p_{33}}^2 [ ( 23 + 17 z ) j_1 + 28 z j_2],~~b_2^{cc } = - \frac{2}{9}f_{p_{33}}^2 ( 2 j_1 + 3 j_2)~{\rm for}~ { \rm { \pi ^+}{^3he},~{\pi ^-}{^3h } } , \cr a_2^{ee } = \frac{1}{9}f_{p_{33}}^2 [ ( 29 + 27 z ) j_1 + 52 z j_2],~~b_2^{cc } = - \frac{2}{9}f_{p_{33}}^2 ( 2 j_1 + 5 j_2)~{\rm for}~ { \rm { \pi ^-}{^3he},~{\pi ^+}{^3h}}. \label{23}\end{aligned}\ ] ] the resulting double - scattering contributions to the differential cross sections are shown by the dotted curves in figs . [ f3 ] @xmath43 [ f6 ] .
although , the @xmath13 @xmath173-wave amplitude is dominant for the multiple scattering of pions on the nucleons of nucleus in the energy range under consideration , the contribution of small nonresonant waves still can play an important role near @xmath174 ( the sharp minimum for single scattering ) .
thus , we limit ourselves to single scattering for nonresonant waves . in this limit ,
the @xmath13 amplitude ( [ 2 ] ) becomes @xmath175 where @xmath176 \nonumber\end{aligned}\ ] ] and @xmath177 and @xmath178 are the spin and isospin projection operators for the state @xmath179 of the @xmath13 system . limiting ourselves to @xmath87- and @xmath173-waves , the nonresonant @xmath13 @xmath180 phases can be taken from a recent gw @xmath13 partial - wave analysis @xcite .
therefore , we take into account two @xmath87- and four @xmath173-waves in our calculations .
the projection operators are @xmath181 where @xmath182 $ ] .
we note that the procedure of taking into account nonresonant waves in the single - scattering approach is analogous to the resonant contribution taken into account and discussed in sec .
[ sec : ssa ] .
the final expressions for the non - spin - flip and spin - flip nonresonant amplitudes are @xmath183 f(\delta)\ , ~{\rm and}~ \cr b_{non } = \frac{1}{3}\,(2 f_{p_{13 } } - f_{p _ { 31 } } - 2 f_{p_{11}})\ , f(\delta)\ , \phantom{x } ~{\rm for}~ \pi^+\,^3{\rm he}\ , ~{\rm and}~ \pi^-\,^3{\rm h } , ~{\rm and}~\cr a_{non } = \left [ \frac{4}{3 } f_{s_{11 } } + \frac{5}{3 } f_{s_{31 } } + \left(\frac{4}{3}f_{p_{11 } } + \frac{5}{3}f_{p_{31 } } + \frac{8}{3}f_{p_{13 } } \right ) z \right ] f(\delta ) \phantom{x } ~{\rm and}~\cr b_{non } = - f_{p_{31}}\ , f(\delta)\ , \phantom{xxxxxxxxxxxxxxxxxxx } ~{\rm for}~ \pi^-\,^3{\rm he}\ , ~{\rm and}~
\pi^+\,^3{\rm h}. \label{26}\end{aligned}\ ] ] the nonresonant contribution for the total amplitude @xmath184 ( see eq . ( [ 27 ] ) below ) is expressed by the substitutions @xmath185 the nonresonant amplitudes are taken into account in the calculations of the ratios ( [ 1 ] ) presented in sec .
[ sec : theor3 ] .
the expression for the sum of the single- and double - scattering amplitudes can be expressed in a form similar to eq .
( [ 21 ] ) : @xmath186 ) , \label{27}\end{aligned}\ ] ] where the functions @xmath187 and @xmath188 represent the contributions of single and double scattering , _
e.g. _ , @xmath189 and @xmath190 .
the amplitudes @xmath191 and @xmath192 are determined by eqs .
( [ 9 ] ) @xmath43 ( [ 11 ] ) and ( [ 26 ] ) ; @xmath165 and @xmath166 by eqs . ( [ 22 ] ) @xmath43 ( [ 23 ] ) . in terms of the functions @xmath187 and @xmath188 , the differential cross section in unpolarized case has the form @xmath193 the combined single- and double - scattering contributions , with interference taken into account , are depicted by the solid curves in figs .
[ f3 ] @xmath43 [ f6 ] .
it can be seen that the model approach we use qualitatively agrees with the data . for forward scattering , @xmath194 ,
wf ( [ 7 ] ) reproduces the cross sections systematically better than wf ( [ 8 ] ) . at larger scattering angles
, the results for wf ( [ 7 ] ) ( wf ( [ 8 ] ) ) lie below ( above ) the experimental data , and the cross - section minimum is shifted to a smaller angle ( @xmath195 ) , in agreement with the experimental data . the key point to be made , however , is that the gentle maximum at @xmath196 arises from the interference between single and double scattering .
it is also seen that the moderation of the rise in the cross sections for backward scattering reflects the contribution of double scattering , which is most pronounced near @xmath197 .
we point out that the description of hadron - nuclear scattering in the backward hemisphere is a very complicated multibody problem that requires detailed information about the wave function of the nucleus and the reaction mechanism .
for example , the 5-component wf used in ref .
@xcite to describe the @xmath93 elastic cross sections at @xmath7 = 200 mev is superior to wf ( [ 7 ] ) .
however , the goal of our study is the description and understanding of csv effects , and these effects do not depend strongly on the details of the wf at small distances . for this reason
, we prefer to use the simpler @xmath87-shell versions of the nuclear wfs ( [ 7 ] ) and ( [ 8 ] ) .
there are three principal sources of the violation of charge symmetry for @xmath198 scattering from light nuclei in the @xmath10 region : 1 .
the coulomb interaction between the charged pions and the nuclei @xmath43 the external coulomb effect , 2 .
the mass splitting of the different charge states of the @xmath10-isobar , and 3 .
the difference between the wfs of @xmath2 and @xmath3 due to the additional coulomb repulsion between the two protons in the @xmath3 nucleus @xmath43 the internal coulomb effect .
we now discuss how we take these effects into account in our calculation of the elastic scattering of charged pions from the @xmath54 nuclei . as was shown in sec .
[ sec : expt ] , experimental data for the ratios ( [ 1 ] ) @xcite were taken outside the coulomb cone , @xmath199 . in this angular range ,
the coulomb amplitude is a smooth function of the scattering angle @xmath162 . here , we take into account the coulomb interaction in a nonrelativistic approach , neglecting the interaction between the photon and the magnetic moment of the nucleus .
thus , the coulomb amplitude of the pion - nucleus interaction in terms of the coulomb phase may be written as @xmath200~f_{\pi } ( \vec{\delta})~f(\vec{\delta } ) , \label{29}\end{aligned}\ ] ] where @xmath201 , @xmath202 and @xmath203 are the charges of the pion and the nucleus , respectively , @xmath204 is the pion cm energy , @xmath205 is the mass of the nucleus , f(@xmath206 ) is the form factor of the nucleus defined by the eq .
( [ a4 ] ) , and @xmath207 is the pion charge form factor that is used in the standard parametrization of ref . @xcite . in calculating the ratios ( [ 1 ] )
, we use the amplitude @xmath208 of eq .
( [ 29 ] ) in combination with the non - spin - flip amplitude @xmath187 of the strong interaction of eq .
( [ 27 ] ) by the substitution @xmath209 .
the influence of the @xmath10-mass splitting on the differential cross section for @xmath35 elastic scattering was discussed in ref .
@xcite , where the single - scattering approximation with allowance for the different charge states of the @xmath10(1232 ) was used . in this approximation
, the csv effect proves to be independent of the scattering angle , with a value proportional to @xmath211 .
nearly the same approach was used for the @xmath212 case in ref .
@xcite .
we denote the different charge states of the @xmath8 via the index @xmath213 and @xmath214 for the @xmath215 , and @xmath216 , respectively .
the mass @xmath217 , corresponding to the isobar @xmath218 , is calculated according to a formula from ref .
@xcite ( page 109 , eq .
( 4.18 ) ) , following ref .
@xcite : @xmath219 where @xmath220 is the 3@xmath221 component of isospin for the @xmath222-term from the @xmath10-multiplet . using the average resonance mass value from the particle data group @xcite , @xmath223 , @xmath224 from ref .
@xcite , and @xmath225 from ref .
@xcite , we get @xmath226 the scalar amplitude for @xmath13 scattering ( see eq .
( [ 2 ] ) ) for each charge state @xmath218 is defined as @xmath227 .
\label{31}\end{aligned}\ ] ] the phases @xmath228 are defined relative to the resonance phase @xmath229 @xmath230 where @xmath231 .
the resonance phase @xmath229 is taken from @xcite . in eq .
( [ 32 ] ) , we neglect the energy dependence of the width @xmath232 because @xmath233 ; in our calculations we use @xmath234 . using this definition of @xmath235 for the @xmath13 scattering amplitudes in eq .
( [ 31 ] ) , we obtain the following expressions for the single pion - nucleus scattering amplitudes : @xmath236 @xmath237 substituting @xmath238 and @xmath239 , the amplitudes @xmath191 and @xmath192 at eqs .
( [ 33 ] ) and ( [ 34 ] ) are transformed to the amplitudes for the isomirror reactions @xmath19 and @xmath240 . for the double - scattering pion - nucleus amplitudes , we obtain @xmath241 ~{\rm and}~ b_2^{ee } = \frac{4}{3}f_1 f_2 ( j_1 + 2 j_2 ) ~{\rm for}~ { \rm { \pi ^+}{^3he},{\pi ^+}{^3h } } \label{35}\end{aligned}\ ] ] instead of eq .
( [ 23 ] ) .
the other amplitudes for @xmath20 and @xmath17 elastic scattering are different : @xmath242 + f_1 ^ 2 [ ( 5 + 3 z ) j_1 + 4 z j_2 ] \phantom{xx } ~{\rm for}~ { \pi ^+}{^3he } , \phantom{x}\cr a_2^{ee } = \frac{4}{3}f_1 f_2 [ 2 ( 1 + z ) j_1 + 4 z j_2 ] + \frac{1}{9}f_2 ^ 2 [ ( 5 + 3 z ) j_1 + 4 z j_2 ] \phantom{x } ~{\rm for}~ { \pi ^+}{^3h } , \phantom{xx}\cr b_2^{cc } = - \frac{2}{9}f_2 ^ 2 ( 2 j_1 + 3 j_2 ) \phantom{xxxxxxxxxxxxxxxxxxxxxx } ~{\rm for}~ { \pi ^+}{^3he } , ~{\rm and}~ \phantom{xx}\cr b_2^{cc } = - \frac{2}{3}f_2 ^ 2 ( 2 j_1 + 5 j_2 ) \phantom{xxxxxxxxxxxxxxxxxxxxxxxxxx } ~{\rm for}~ { \pi ^+}{^3h}. \phantom{xxx } \label{36}\end{aligned}\ ] ] the amplitudes @xmath165 and @xmath166 of eqs .
( [ 35 ] ) and ( [ 36 ] ) are also transformed to their isomirror reaction amplitudes , _
i.e. _ , @xmath243 and @xmath244 , by substituting @xmath238 and @xmath239 . we do not include any difference for the different charge states in the nonresonant amplitudes @xmath245 and @xmath246 .
the difference in the structure of the wfs of @xmath2 and @xmath3 is related not only to the electromagnetic interaction , but also to the the part of the strong interaction which violates isospin . in the strong - interaction sector
, there are terms that violate isospin directly @xcite .
isospin violation inside nuclei can relate to both nucleon and quark degrees of freedom . in terms of quark degrees of freedom
, isospin violation relates to the mass difference of the @xmath11 and @xmath12 quarks . at present
, there is no quantitatively good estimate of isospin violation due to the strong interaction for the @xmath54 nuclei .
if we assume that the strong interaction conserves isospin , then the main reason for the difference in the structure of the wfs of @xmath2 and @xmath3 is the additional coulomb repulsion between the two protons in @xmath3 , which is not present for @xmath2 . if there were no coulomb interaction between these two protons ( _ i.e.
_ , if the wfs of @xmath2 and @xmath3 were isotopically symmetric ) , the neutron distribution for @xmath3 ( the odd " neutron ) would be the same as the proton distribution for @xmath2 ( the odd " proton ) , and the proton distribution for @xmath3 ( the even " protons ) would be the same as the neutron distribution for @xmath2 ( the even " neutrons )
. however , the proton distributions for @xmath2 and @xmath3 can still be different and , as a consequence , so can the charge form factors of @xmath2 and @xmath3 .
if , however , isospin is violated for @xmath2 and @xmath3 , the even- and odd - nucleon distributions can also be different . in ref .
@xcite , the difference between these distributions has been formulated in terms of nonzero parameters @xmath247 and @xmath248 , where @xmath249 here , @xmath250 is the neutron ( n ) or proton ( p ) radius for the even ( e ) or odd ( o ) nucleons . as was shown in ref .
@xcite , the superratio @xmath6 at @xmath251 is very sensitive to @xmath252 and @xmath248 ; a reasonable description of @xmath6 versus scattering angle has been obtained for @xmath253 and @xmath254 .
these differences between @xmath255 and @xmath256 or between @xmath257 and @xmath258 due to isospin violation result in additional changes in the charge radii and form factors of @xmath2 and @xmath3 .
we vary the parameters of the wfs ( [ 7 ] ) and ( [ 8 ] ) to introduce such differences .
for wf ( [ 7 ] ) , we follow the recipe suggested in ref .
@xcite : we fix the slope @xmath259 for @xmath2 at @xmath260 and vary the slope for @xmath3 to obtain the best description of the data for the ratios @xmath4 and @xmath5 .
an analogous procedure is followed for wf ( [ 8 ] ) , using parameters for the wf for @xmath3 suggested in ref .
@xcite ( see sec . [
sec : theor1 ] ) .
then , for the wf of @xmath2 , we use three different variations of the wf : ( i ) a variation of the slope @xmath261 , ( ii ) a variation of the slope @xmath262 , and ( iii ) a variation of both slopes @xmath261 and @xmath262 , which are proportional to each other : @xmath263 and @xmath264 .
although this variation of the parameters of wfs ( [ 7 ] ) and ( [ 8 ] ) can not be compared directly with the refined procedure used in ref .
@xcite , this way of taking into account the internal coulomb interaction allows us to take the the @xmath210-mass splitting into account , which was not done in ref .
therefore , the quantities @xmath252 and @xmath248 obtained from the experimental data can differ from the values obtained in ref .
the results of our calculations for the ratios @xmath4 , @xmath5 , and the superratio @xmath6 which take into account the external coulomb interaction and the @xmath10-mass splitting but exclude the internal coulomb effect are shown in figs .
[ f7 ] @xmath43 [ f10 ] , as the dotted curves for single scattering and the dashed curves for both single and double scattering .
there are no free parameters associated with these results .
the main purpose of figs . [ f7 ] @xmath43 [ f10 ] ( as for figs .
[ f3 ] @xmath43 [ f6 ] ) is to show that the inclusion of double scattering is essential to be able to follow the trend of the data in the non - spin - flip - dip region .
as is seen in these figures , there is qualitative agreement between the results of our calculations and the data for @xmath7 = 180 , 220 , and 256 mev .
also , in contrast with the case for the differential cross section ( see figs .
[ f3 ] @xmath43 [ f6 ] ) , there is little sensitivity to the wf here . for @xmath7 = 180 mev
, the ratio @xmath4 is reproduced very well for both wfs ( [ 7 ] ) and ( [ 8 ] ) , but the peaks near @xmath265 in the ratio @xmath5 and superratio @xmath6 are reproduced better by wf ( [ 7 ] ) . at the same time , the description of @xmath5 and @xmath6 in the backward direction is not good for either wf . figure [ f8 ] shows that even taking into account single scattering and the external coulomb interaction can not reproduce the experimental data there .
but overall , figs .
[ f8 ] @xmath43 [ f10 ] show that taking into account the @xmath10-mass splitting consistently with both the single- and double - scattering contributions reproduces the main structures of the angular distribution and that we do not require a detailed knowledge of the nuclear wf .
we note here that the data for @xmath5 and @xmath6 for @xmath7 = 142 mev are not reproduced by our model approach .
apparently , some other mechanism must play a role at this energy .
we return to this problem below .
we now describe the csv effects by taking into account the internal coulomb interaction as well .
the procedure for variation of the wfs is described above . for this case
, we vary a single free parameter to obtain the best fit .
again , we fit @xmath4 and @xmath5 only , since the superratio @xmath6 ( @xmath266 ) is not an independent quantity .
the best - fit results for this approach and for both wfs ( [ 7 ] ) and ( [ 8 ] ) are shown in figs .
[ f11 ] @xmath43 [ f14 ] by the solid curves . by comparison , the results without the internal coulomb interaction ( from figs .
[ f7 ] @xmath43 [ f10 ] ) are shown by the dashed curves .
both free parameters @xmath261 and @xmath262 have been varied simultaneously , following our prescription ( iii ) of sec .
[ sec : theor2 ] .
the results of varying either @xmath261 or @xmath262 independently are very similar , as listed in the table . in this table ,
the best - fit results obtained by variations of the radii of the nuclei and corresponding @xmath267 values are listed as well . in fig .
[ f12 ] , we see that our calculations reproduce all the ratios @xmath4 , @xmath5 , and @xmath6 for t@xmath268 = 180 mev rather well over the entire angular range .
in fact , we reproduce the superratio @xmath6 for @xmath269 much as gibbs and gibson did @xcite . taking into account
the difference in the wfs of @xmath2 and @xmath3 results in a much better reproduction both of the quantity @xmath6 at @xmath195 and the scattering at backward angles .
thus , taking into account the internal coulomb interaction provides a substantial improvement in @xmath267 compared with the case where only the external coulomb interaction and the @xmath10-mass splitting are included . finally , we consider the description of the data for @xmath7 = 220 and 256 mev to be qualitatively satisfactory , while the description of the data for @xmath7 = 142 mev is not .
consider the situation for @xmath7 = 220 mev , shown in fig .
our theoretical curves reproduce the sharp dip - bump angular dependence of the ratios @xmath5 and @xmath6 remarkably well . at @xmath7 = 256 mev , shown in fig .
[ f14 ] , we predict a more gradual dip - bump angular dependence , which also agrees with the experimental data . at the same time , our theoretical curves do not reproduce the data for @xmath7 = 142 mev , shown in fig .
[ f11 ] , even though the angular dependence is much smoother .
we note , however , that the amount of experimental data for @xmath7 = 220 and 256 mev is significantly less than for @xmath7 = 142 and especially for 180 mev .
evaluation of the @xmath7 = 142 mev data may require an additional mechanism to reproduce the behavior of the data .
= 180 mev . on the other hand , the calculation of ref .
@xcite does not reproduce the csv effect for 142 mev either . ]
we have seen that one of the principal sources of csv is the @xmath10-mass splitting . the difference in the total @xmath41 cross sections
is determined by the @xmath10-mass splitting @xcite . here
, we calculate this effect for the ratios @xmath270 and @xmath271 for the _ total _ cross sections of @xmath272 scattering , defined as @xmath273 we predict a considerably larger csv effect for the three - nucleon system than for the deuteron , as shown in fig .
moreover , the crossover of @xmath270 and @xmath271 at the peak of the @xmath10 , for either of the wfs used , establishes a unique signature for this effect .
the high sensitivity of csv to the @xmath274-mass splitting ( within multiplets ) allows us to suggest the use of these ratios as a method to determine the mass - splitting in heavy @xmath274s and @xmath275s .
we think that this method would be preferable to the traditional @xmath13 partial - wave analysis for baryon spectroscopy . in our theoretical approach to single and double scattering
, we neglect the fermi motion of the nucleons inside the nucleus .
the amplitude of the @xmath13 scattering is extracted from an integral over incident pion energies that are on - shell .
we also neglect the recoil of the nucleons in the green s function @xmath137 of eq .
( [ 14 ] ) for the double - scattering amplitude .
we call this approach the fixed - centers approximation . "
then , we observe that when only certain corrections are taken into account , we obtain worse agreement with the experimental data ( see ref .
@xcite for details ) .
since the fixed - centers approximation gives results that are close to the data , there must be cancellations of the main corrections to leading order .
these cancellations for hadron - deuteron scattering have been discussed in detail in previous work .
the cancellation of nonadiabatic corrections within the glauber approach for differential cross sections at high energies was shown in ref .
@xcite . in the @xmath35 scattering - length calculations ,
the cancellation of off - shell and recoil corrections was discussed in ref .
the cancellation of nonadiabatic corrections for @xmath35 elastic scattering was found in the range of the @xmath8 resonance in ref .
apparently , the analogous cancellation of corrections holds as well for the @xmath54 nuclei .
therefore , the use of only a few of the corrections to the fixed - centers approximation can result in worse agreement with the data than when all the corrections are ignored .
inclusion of all the corrections ( including the so - called binding corrections ) is a rather complicated task and is not a goal of our present study .
for example , we limit ourselves to the consideration of single- and double - scattering terms , and this approach allows us to take into account the leading terms of the amplitude of the pion - nuclear interaction at energies and scattering angles where we can qualitatively reproduce the shape of the distribution of the differential cross sections .
triple - scattering contributions to the differential cross sections can smooth the angular shape , but they are more difficult to calculate and can not be simplified by the transformation of the integrals @xmath158 and @xmath159 discussed in appendix [ sec : tensor ] . at small scattering angles , @xmath276 ,
there is reasonable agreement between our theoretical approach and the experimental data for @xmath7 = 180 mev and above , but for @xmath7 = 142 mev the experimental cross sections are smaller than the results of our model calculations for both wfs used ( figs .
[ f3 ] @xmath43 [ f6 ] ) . but because for modest scattering angles the range of momentum transfer is small ( @xmath277 ) , our use of the simple wfs ( [ 7 ] ) and ( [ 8 ] ) is reasonable . we therefore infer that the suppression of the cross section for @xmath7 = 142 mev results from absorption , which is absent in our model approach for the amplitude of the @xmath278 interaction . usually , the absorption on a nucleus is due to the reaction @xmath279 , where @xmath280 is a pair of correlated nucleons .
but the total cross section of the @xmath281 reaction has its maximum at @xmath7 = 140 mev , and the absorption cross section is somewhat suppressed at @xmath7 = 180 mev @xcite .
however , @xmath7 = 180 mev corresponds to the maximum of the total @xmath282 cross section , as shown in fig .
[ f16 ] . the pion absorption cross section for @xmath3 , measured at psi @xcite ,
also peaks at @xmath7 = 140 @xmath43 150 mev , as shown in fig .
[ f16 ] as well .
therefore , if absorption were responsible for the suppression of the scattering cross sections at small scattering angles for @xmath7 = 142 mev , then this effect would be considerably smaller for @xmath7 = 180 mev , where our fit to the data is much better .
data for the elastic @xmath284 differential cross sections have been obtained at lampf for @xmath7 below , at , and above the @xmath10 resonance @xcite .
the spin - flip amplitudes for elastic @xmath283 in the single - scattering approximation do not contribute to the @xmath8 resonance .
preliminary analysis of the charge asymmetry for @xmath7 = 180 mev shows a statistically significant effect in @xmath27 , of the order of 10% at scattering angles @xmath285 @xcite .
the size of @xmath27 for @xmath286 and the possible influence of absorption there are still unknown . because of the larger role played by pion absorption on heavier nuclei , csv is expected to be suppressed relative to the three- and four - body nuclei . if present , csv in heavier nuclei probably depends more on their geometrical properties than on the @xmath10-mass splitting .
we have performed theoretical calculations for the simple ratios @xmath4 and @xmath5 and the superratio @xmath6 for elastic @xmath45 scattering , for t@xmath268 = 142 , 180 , 220 , and 256 mev and over a broad angular range .
we have found reasonable agreement between the results of our calculations and the experimental data , shown in fig .
[ f1 ] , over most of the range of the data .
our calculations were done in an approach utilizing the sum of the single- and double - scattering @xmath13 contributions , as indicated in fig .
we took into account three sources of csv @xmath43 the @xmath10-mass splitting and the external and internal coulomb interactions .
we used @xmath87-shell wfs for @xmath2 and @xmath3 .
this approach enabled us to use simple analytical expressions for the double - scattering contribution to pion - nuclear scattering , taking into account all spin and isospin amplitudes .
we used two different radial wfs for the @xmath54 nuclei : + i ) a simple gaussian distribution ( eq . ( [ 7 ] ) ) with the slope describing the charge densities of @xmath2 and @xmath3 obtained from electron scattering @xcite .
we used the wf of ref .
+ ii ) a sum of two gaussian wfs ( eq . (
[ 8 ] ) ) , as used in ref .
@xcite for the description of the differential cross sections of the reaction @xmath287 .
this wf reproduces the minimum of the @xmath3 charge form factor at @xmath288 , but shows an enhancement at smaller momentum transfer .
+ the calculated cross sections , shown in figs .
[ f3 ] @xmath43 [ f6 ] , agree qualitatively with the experimental data . for t@xmath268 = 180 mev ,
the theoretical curves have minima at @xmath289 and reproduce the gradual growth of the cross sections as @xmath162 approaches @xmath290 , indicating the importance of the inclusion of double scattering for the differential cross sections .
of course , the absolute cross sections are very sensitive to the wf , and are not reproduced well by the simple @xmath87-shell approach used here .
the main goal of our study is the calculation of the csv effects for @xmath45 differential cross sections in terms of the observables @xmath4 , @xmath5 , and @xmath6 .
no free parameters are used in our approach in taking into account the @xmath10-mass splitting and the external coulomb interaction .
figures .
[ f8 ] @xmath43 [ f10 ] show that these factors alone account qualitatively for major features of the data .
these figures also show that there is little sensitivity to the choice of the wave function .
figures [ f12 ] @xmath43 [ f14 ] show that when the internal coulomb effect is included as well , there is reasonable agreement between our theoretical calculations and the experimental data .
the best agreement is found for t@xmath268 = 180 mev ( at the peak of the @xmath8 resonance ) .
both the @xmath10-mass splitting and the internal coulomb interaction are important for the reproduction of the shape of the angular distribution , both near the non - spin - flip dip at @xmath195 and at large scattering angles . although the influence of the internal coulomb interaction on csv has been shown before @xcite , our investigation shows that including the @xmath10-mass splitting results in a still better description of the effect of csv , as it should : the @xmath8-mass splitting exists , so its effects should not be ignored .
we also predict the simple mirror ratios for the total cross sections , as shown in fig .
[ f15 ] . finally , however , as seen from figs . [ f7 ] and [ f11 ] , our calculations do not reproduce the data for @xmath7 = 142 mev . although we tried to take into account a number of different approaches beyond the framework of our model ( more accurate amplitudes for single and double scattering and fermi motion , _ etc . _ ) , we were not able to improve the agreement for @xmath7 = 142 mev
. therefore , the question of the nature of the effect of csv for @xmath7 = 142 mev remains open .
perhaps there is an additional mechanism at @xmath7 = 142 mev which does not manifest itself at higher energies ; fig .
[ f16 ] shows that quite possibly , pion absorption plays a major role .
the authors acknowledge very useful communications with b. v. geshkenbein , s. s. kamalov , and c. l. morris . this work was supported in part by the u. s. department of energy under grants de fg0295er40901 and de fg0299er41110 and by the russian grants for basic research n 980217618 and n 001596562 .
a. k. acknowledges the hospitality extended by the center for nuclear studies of the george washington university .
i. s. acknowledges partial support from jefferson lab , by the southeastern universities research association under doe contract de ac0584er40150 .
let us introduce relative coordinates of the nucleons @xmath291 and @xmath292 instead of @xmath293 . in terms of these new variables ,
the function @xmath294 of eq .
( [ 5 ] ) yields @xmath295 this function @xmath294 does not depend on the selection of a basis @xmath296 .
the fourier transformation of the function @xmath294 is defined by @xmath297 where @xmath298 and @xmath299 are relative momenta and ( @xmath131 , @xmath300 , @xmath301 ) are the momenta of the nucleons of the nucleus . for the wf ( [ 8 ] )
, @xmath126 has the form @xmath302 where @xmath303 the charge form factor for elastic scattering is defined by @xmath304 an analytical expression for the form factor ( [ a4 ] ) corresponding to the wf ( [ 8 ] ) is @xmath305 , \label{a5}\end{aligned}\ ] ] where @xmath306
the expression for the tensor @xmath154 in terms of the wave function @xmath308 ) ( [ a2 ] ) is @xmath309 this integral is suitable for calculations in coordinate space . to do this
, we follow a transformation first used in ref .
@xcite : @xmath310 where @xmath311 and @xmath312 then , using eq .
( [ b2 ] ) and the coordinate expression for the wave function ( [ a1 ] ) , we get the expression for the tensor @xmath154 : @xmath313 , \label{b3}\end{aligned}\ ] ] where @xmath314 . if the wf @xmath315 is expressed as a sum of several gaussians , the integral ( [ b3 ] )
can be represented in the form of expression ( [ 20 ] ) .
then the integrals @xmath158 and @xmath159 are transformed into one - dimensional integrals .
for the wf ( [ 8 ] ) , they have the form @xmath316 dr , \phantom{x}\cr e_n = \int\limits_{-1}^{1}\exp(i \xi r z ) z^n dz,\phantom{xxxxxxxxxxxx } \label{b4}\end{aligned}\ ] ] @xmath317 , @xmath318 , and @xmath319 is given by eq .
( [ a3 ] ) . in the case of wf ( [ 7 ] ) ,
the expressions for the integrals @xmath320 are computed easily from eq .
( [ b4 ] ) .
meissner , in : _ proc . of 9th international symposium on meson - nucleon physics and the structure of the nucleon , washington , dc , usa ,
july 2631 , 2001 _ , eds . h. haberzettl and w. j. briscoe , @xmath13 newslett .
* 16 * , 1 ( 2001 ) , to be published
. w. j. briscoe _
_ , in preparation for submission to phys . rev . c. b. l. berman g. c. anderson , w. j. briscoe , a. mokhtari , a. m. petrov , m. e. sadler , d. b. barlow , b. m. k. nefkens , and c. pillai , phys
c * 51 * , 1882 ( 1995 ) .
s. k. matthews , w. j. briscoe , c. bennhold , b. l. berman , r. w. caress , k. s. dhuga , s. n. dragic , s. s. kamalov , n. j. nicholas , m. f. taragin , l. tiator , s. j. greene , d. b. barlow , b. m. k. nefkens , c. pillai , j. w. price , l. d. isenhower , m. e. sadler , i. slaus , and i. supek , phys .
c * 51 * , 2534 ( 1995 ) .
s. k. matthews , elastic scattering of pions from tritium and @xmath3 in the backward hemisphere in the region of the @xmath321 resonance , " ph . d. thesis , the george washington university , 1993 ( unpublished ) .
juster , s. auffret , j .-
cavedon , j .- c .
clemens , b. frois , d. goutte , m. huet , p. leconte , j. martino , y. mizuno , x .- h .
phan , s. platchkov , and s. williamson , phys .
* 55 * , 2261 ( 1985 ) .
g. fldt , phys .
scripta , * 16 * , 81 ( 1977 ) ; v. m. kolybasov and v. g. ksenzov , jetp * 71 * , 13 ( 1976 ) ; v. g. ksenzov , phys . atom . nucl .
( former sov .
j. nucl . phys . )
* 28 * , 644 ( 1978 ) ; o. d. dalkarov , v. m. kolybasov , and v. g. ksenzov , nucl . phys . *
a397 * , 498 ( 1983 ) ; v. v. baru and a. e. kudryavtsev , phys .
( former sov .
j. nucl . phys . )
* 60 * , 1475 ( 1997 ) . the ratios @xmath4 , @xmath5 , and the superratio @xmath6 for @xmath327 elastic scattering for various incident pion kinetic energies : ( a ) t@xmath268 = 142 mev , ( b ) 180 mev , ( c ) 220 mev , and ( d ) 256 mev .
experimental data are from @xcite ( diamonds ) , @xcite ( circles ) , @xcite ( triangles ) , and @xcite ( squares ) .
differential cross sections for @xmath329 elastic scattering for @xmath7 = 142 mev . plotted are results for ( a , b ) wfs ( [ 7 ] ) and ( c , d ) wf ( [ 8 ] ) .
experimental data are from @xcite ( diamonds ) , @xcite ( circles ) , @xcite ( triangles ) , and @xcite ( squares ) , with @xmath330 ( filled ) and @xmath331 ( open ) .
the cross sections ( a , c ) are for @xmath20 and @xmath19 and ( b , d ) for @xmath17 and @xmath18 .
the solid curves give the total contribution .
results for single and double scattering alone are shown by the dashed and dotted curves , respectively . the ratios @xmath4 and @xmath5 and the superratio @xmath6 for t@xmath268 = 142 mev .
the notation for the experimental data is the same as for fig .
[ f1 ] . only the @xmath10-mass splitting and external coulomb contributions are taken into account .
the full calculations take into account both single and double scattering , and are shown by the dashed curves .
the results for single scattering alone are shown by the dotted curves . plotted
are the results for ( a ) wf ( [ 7 ] ) and ( b ) wf ( [ 8 ] ) . the ratios @xmath4 and @xmath5 and the superratio @xmath6 for t@xmath268 = 142 mev .
the notation for the experimental data is the same as for fig .
the @xmath10-mass splitting with external ( and internal ) coulomb contributions are shown by the dashed ( solid ) curves . plotted are the results for ( a ) wf ( [ 7 ] ) and ( b ) wf ( [ 8 ] ) . version ( iii ) of the variation of the wfs is shown , as described in sec .
[ sec : theor2 ] .
predictions for the ratios @xmath270 ( solid ) , and @xmath271 ( dashed ) for the total @xmath328 cross sections .
calculations were done for single and double scattering with the @xmath10-mass splitting .
the coulomb interactions are not taken into account . plotted are the results for ( a ) wf ( [ 7 ] ) and ( b ) wf ( [ 8 ] ) .
total @xmath0 cross sections . plotted
are cross sections for @xmath333 ( solid ) and @xmath334 ( multiplied by a factor of 20 ) ( dashed ) from a recent combined fit of the @xmath335 and @xmath35 elastic scattering with the @xmath334 data @xcite . the @xmath40 absorption data ( multiplied by a factor of 10 ) are from @xcite ( filled circles ) ; it can be seen that they peak near t@xmath268 = 140 mev . | we discuss the experimental and theoretical status of charge - symmetry violation ( csv ) in the elastic scattering of @xmath0 and @xmath1 on @xmath2 and @xmath3 .
analysis of the experimental data for the ratios @xmath4 , @xmath5 , and @xmath6 at @xmath7 = 142 , 180 , 220 , and 256 mev provides evidence for the presence of csv .
we describe pion scattering from the three - nucleon system in terms of single- and double - scattering amplitudes .
external and internal coulomb interactions as well as the @xmath8-mass splitting are taken into account as sources of csv .
reasonable agreement between our theoretical calculations and the experimental data is obtained for @xmath9 = 180 , 220 , and 256 mev . for these energies
, it is found that the @xmath10-mass splitting and the internal coulomb interaction are the most important contributions for csv in the three - nucleon system .
the csv effects are rather sensitive to the choice of pion - nuclear scattering mechanisms , but at the same time , our theoretical predictions are much less sensitive to the choice of the nuclear wave function .
it is found , however , that data for @xmath5 and @xmath6 at @xmath7 = 142 mev do not agree with the predictions of our model , which may indicate that there are additional mechanisms for csv which are important only at lower energies . |
this work was supported in part by the jsps core - to - core program international research network for non - equilibrium dynamics of soft matter " , and kyushu university interdisciplinary programs in education and projects in research development . | two phase picture is a simple and effective methodology to capture the nonequilibrium dynamics of polymer associated with tension propagation .
when applying it to the driven translocation process , there is a point to be noted , as briefly discussed in our recent article [ phys .
rev .
e * 85 * , 061803 ( 2012 ) ] . in this article
, we address this issue in detail and modify our previous prediction [ euro .
phys .
j. e * 34 * , 135 ( 2011 ) ] by adopting an alternative steady - state ansatz .
the modified scaling prediction turns out to be the same as that of the iso - flux model recently proposed by rowghanian and grosberg [ j. phys .
chem .
b * 115 * , 14127 - 14135 ( 2011 ) ] .
# 1 # 1 tension - propagation mechanism is inherent in the dynamical response of polymers to local stimuli .
sometime ago , it has been proposed as an essential ingredient governing the driven translocation process @xcite .
such a view seems to have been well validated in recent studies @xcite . in ref .
@xcite , we analyzed in detail the translocation dynamics based on two phase formalism , in which the portion of the translocation polymer in _ cis _ side is pictured to be composed of the moving and the yet quiescent equilibrium domains .
the point of the formalism is to focus on the dynamics of the domain boundary , which is a gross variable describing how the driving force at the pore site is transmitted along the chain backbone .
for such a coarse - graining to be meaningful , other degrees of freedom need to be properly eliminated , the effect of which should be reflected in the behaviors of the gross variable .
most naturally , the moving domain is assumed to be in a steady - state with a characteristic velocity @xmath0 .
then , the steady - state dynamical friction / extension laws are coupled to the continuum equation describing the conservation of the mass . in our earlier article @xcite
, we set @xmath1 , where @xmath2 represents the segment velocity around the rear end of the moving domain .
this steady - state ansatz , however , turned to be not compatible with the mass conservation relation inherent in the translocation process .
this point was shortly mentioned in the appendix of our article @xcite . to make the point clearer
, we present a detailed account in this article , where our previous result @xcite is modified by adopting an alternative steady - state ansatz suggested in refs .
@xcite .
we then point out that the alternative ansatz is intimately related with the iso - flux model @xcite . _
basic equations_ let us consider the driven translocation process as schematically depicted in fig . 1 .
a polymeric chain is initially dissolved in the solution ( @xmath3 ) , taking a random coil conformation with the overall size @xmath4 , where @xmath5 is the total segment number , @xmath6 is the segment length , and @xmath7 is the flory exponent . as those in ref .
@xcite , the length , force and time are made dimensionless by the units @xmath6 , @xmath8 and @xmath9 , respectively ( @xmath10 is the thermal energy ; @xmath11 is the viscosity of the solvent ) .
there is a thin wall at @xmath12 with a small pore , where the driving force with the constant magnitude @xmath13 is locally exerted in @xmath14-direction from the _ cis _ to the _ trans _ side .
one chain end is sucked at time @xmath15 .
the translocation process , then , proceeds with the tension propagation along the chain .
segments are numbered from the first sucked end to the other end ( @xmath5th segment ) . according to the so - called two phase formalism
, the dynamics of driven translocation is described as the growth process of the moving tensed domain , which is specified by the segment @xmath16 at the end of the moving domain , the segment @xmath17 at the pore , the size @xmath18 and the representative ( or average ) velocity @xmath0 of the moving domain at time @xmath19 ( see fig . 1 ) .
these can be determined by the following set of equations : @xmath20 @xmath21 @xmath22 @xmath23 where @xmath24 and @xmath25 are the segment fluxes at the moving domain boundary and at the pore , respectively . as explained in ref .
@xcite , eq .
( [ eq:1 ] ) is the continuum equation , eqs .
( [ eq:2 ] ) and ( [ eq:3 ] ) are the dynamical equations of state describing the steady - state relation among velocity - extension - force ( eq .
( [ eq:2 ] ) ) and mass - extension - force ( eq . ( [ eq:3 ] ) ) for a dragged polymer , where @xmath26 and @xmath27 in trumpet regime @xmath28 , while @xmath29 and @xmath30 for stronger force @xmath31 , @xmath32 and @xmath33 , see ref . @xcite .
strong force regimes are further classified as the stem - flower ( @xmath34 ) and the strong - stretching ( @xmath35 ) . ] .
note that , while the segment line density at the pore @xmath36 is given by the exponent related to equilibrium ( static ) situation , eq . ( [ eq:2 ] )
reflects the dissipation mechanism through the exponent @xmath37 , which contains the dynamical exponent @xmath38 characterizing the dynamics of critical fluctuation " .
_ ansatz ( i)_ in ref .
@xcite , we eliminate the pore flux @xmath39 in eq .
( [ eq:1 ] ) and made the steady - state ansatz for the moving domain by setting @xmath40 , which leads to @xmath41 combining this with eqs .
( [ eq:2 ] ) , ( [ eq:4 ] ) and the segment line density at the boundary @xmath42 , where @xmath43 $ ] for @xmath44 and @xmath45 for larger force , one obtains the tension - propagation law @xmath46 with @xmath47 and @xmath48 , thus , the scaling formula for propagation time is @xmath49 with @xmath50 .
_ ansatz ( ii)_ an alternative steady - state ansatz is to set @xmath51 as was done in ref .
@xcite .
then , combining eqs .
( [ eq:3 ] ) and ( [ eq:5 ] ) , we obtain @xmath52 where eqs .
( [ eq:2 ] ) , ( [ eq:4 ] ) and the expression for the line density at the pore ( see eq .
( [ eq:3 ] ) ) are used .
the tension propagation law and the propagation time in leading order can be expressed in the form of eqs .
( [ eq:7 ] ) and ( [ eq:8 ] ) , respectively , but now the exponent @xmath53 and @xmath54 are modified as @xmath55 and @xmath56 .
the deduced exponents coincide with those based on the iso - flux model @xcite ( see below ) .
the difference in the dynamical evolutions of the number @xmath17 of the translocated segments depending on the adopted ansatz is plotted in fig .
[ fig2 ] .
the inflection points in fig .
[ fig2](a ) correspond to the end of the propagation stage , when most of segments have been already sucked .
the subsequent post - propagation stage can be analyzed using eqs .
( [ eq:6])/([fore_mass_cons ] ) @xcite , where , in contrast to the propagation stage , the growth rate in @xmath17 increases with time due to the decrease in the overall friction with the process advanced . as shown in fig .
2(b ) , the growth of @xmath17 in the propagation stage follows the asymptotic dynamical scaling @xmath57 , with the slight deviation in the early period due to the finite - size effect . of the translocated segments as a function of time calculated according to the steady - state ansatz ( i ) or ( ii ) with @xmath58 , @xmath59 and @xmath60 .
( a ) entire process in normal scale .
( b ) propagation stage in double - logarithmic scale .
the dashed lines represent the asymptotic scaling with the slope @xmath61 . ] _ iso - flux model_ we now attempt to resolve the discrepancy between the above two results . the analysis of eqs .
( [ eq:1 ] ) , ( [ eq:3 ] ) and ( [ eq:4 ] ) reveals a unique property of the moving domain in the translocation process .
given the average relation eq .
( [ eq:4 ] ) between @xmath18 and @xmath16 , one can show that the last terms in eqs .
( [ eq:1 ] ) and ( [ eq:3 ] ) are irrelevant , i.e. , @xmath62 and @xmath63 .. since @xmath64 , this leads to the inequality @xmath65 . eliminating @xmath66 through @xmath67}$ ] , one finds @xmath68 , which is equivalent to be proven with eq .
( [ eq:4 ] ) .
second inequality : using eq .
( [ eq:4 ] ) , this can be rewritten as @xmath69 , which holds under the condition @xmath70 of interest . ]
it then follows that @xmath71 and @xmath72 in the scaling level , the fluxes into and out of moving domains balance , as was recently proposed by rowghanian and grosberg @xcite . to check the validity of the steady - state approximation ,
let us set @xmath73 with @xmath74 $ ] so that @xmath75 and @xmath76 .
the iso - flux condition eq .
( [ eq:11 ] ) is written as @xmath77 \simeq \sigma_r [ v(t)-\delta v_r(t ) ] \nonumber \\ \leftrightarrow & & 1+\frac{\delta v_0}{v } \simeq w(t)^q \left ( 1- \frac{\delta v_r}{v } \right)^{1-q } \end{aligned}\ ] ] where @xmath78 with @xmath79 for @xmath80 and @xmath81 for @xmath31 satisfies @xmath82 ( eq .
( [ eq:2 ] ) ) in the stem - flower regime . ] . given the longest relaxation time @xmath83 of the steadily driven domain and the typical shear rate @xmath84 , the steady - state approximation implicitly assumes the condition @xmath85 .
this self - consistency condition is satisfied for the ansatz ( ii ) , i.e. , @xmath86 v < v$ ] , but not for the ansatz ( i ) , i.e. , @xmath87 v$ ] in the force range @xmath44 where @xmath88 . to summarize ,
the unique driving mode inherent in the translocation process is manifested in the iso - flux condition , which is to be contrasted with the mechanical driving by pulling chain end @xcite .
in the latter , the moving domain is constituted of all the @xmath16 segments , so that eqs .
( [ eq:3 ] ) and ( [ eq:4 ] ) are modified as @xmath89 and @xmath90 , respectively .
it has turned out that our previous ansatz and corresponding predictions @xcite are incompatible with the iso - flux condition . from purely theoretical point of view @xcite ,
the scaling of the propagation time ( thus , the translocation time ) is given by @xmath91 it is somewhat puzzling that the prediction based on the ansatz ( i ) seem to be better correlated with numerical and real experiments @xcite . this might be ascribed to finite - size effects and various crossovers detailed in ref .
@xcite ;
( i ) the last term in eq .
( [ eq:9 ] ) serves as a correction term to the tension - propagation scaling law ( see the inequality above eq .
( [ eq:10 ] ) ) .
( ii ) the post - propagation stage adds a correction term to the translocation time .
( iii ) iso - flux condition ( eqs . ( [ eq:10 ] ) and ( [ eq:11 ] ) ) is also influenced by the finite - size correction .
these effects may be enhanced in realistic situations due to the wall and/or the _ trans_-side effects , which would be dependent on the size and properties of the pore etc . since the pore flux would be sensitive to such specific pore effects , one
may then be tempted to remove the pore flux ( eq .
( [ eq:5 ] ) ) from eq .
( [ eq:1 ] ) and to follow the route of the ansatz ( i ) .
if the outgoing flux @xmath92 is suppressed enough compared to that expected from the iso - flux condition , the description based on the steady - state ansatz ( i ) may attain its apparent suitability .
a very recent theoretical study based on a more accurate " two phase description seems to support such a view @xcite , where it has been shown that the frictional interaction between the pore and the polymer significantly affects the dynamics for polymers with experimentally relevant , finite chain length . |
recently , the study of thermodynamical properties of the black holes using techniques explored in statistical physics and fluids has received a special interest @xcite .
these researches have brought new understanding of the fundamental physics associated with the critical behaviors of several black holes in various dimensions using either numerical or analytic methods@xcite . in particular ,
ads black holes in arbitrary dimensions have been extensively investigated in many works @xcite .
more precisely , the state equations @xmath5 have been established by considering the cosmological constant as the thermodynamic pressure and its conjugate as the thermodynamic volume .
in fact , this issue has opened a new way to study the behavior of the rn - ads black hole systems using the physics of van der waals fluids .
indeed , it has been shown that the corresponding @xmath6-@xmath7 criticality can be linked to the liquid - gas systems of statistical physics .
moreover , it has been realized that the criticality depends on many parameters including the dimension of the spacetime @xcite .
more recently , a special interest has been devoted to study of maxwell s equal for some black hole solutions .
more precisely , the state equations of various ads black hole in @xmath8 diagrams have been worked out showing the existence of undesirable negative pressure .
they appear also thermodynamic unstable regions associated with the condition @xmath9 . in this way ,
the corresponding system can be contracted and expanded automatically @xcite . in van der waals systems ,
such problems have been overcame by using the maxwell equal area approach .
the aim of this work is to contribute to these researches by studying higher dimensional uncharged and charged gauss - bonnet - anti - de sitter black holes in extended phase space .
considering the cosmological constant @xmath0 as a thermodynamic pressure and its conjugate quantity as a thermodynamic volume , we present the maxwell s equal area law of higher gauss - bonnet - ads black holes in extended space .
these black hole solutions involve critical behaviors like van der waals gas .
it has been shown that below the critical temperature @xmath1 the stable equilibrium is violated .
based on the maxwell s construction , we obtain the isobar line for which the liquid - gas phases coexist . the paper is organized as follow . in section 2 , we give an overview on the thermodynamics of higher dimensional gauss - bonnet black holes in ads geometry . in section @xmath10 , we first study the equal area law of gauss - bonnet - ads black hole in extended phase space for uncharged solutions .
next , we extend the analysis to the charged case in five dimensions . then , we present the results for higher dimensions .
the last section is devoted to conclusion .
in this section , we give an overview on thermodynamics of gauss - bonnet black holes in ads space .
this matter is based on many works given by other authors @xcite . indeed , we start by considering a @xmath11-dimensional einstein - maxwell theory in the presence of the gauss - bonnet terms and a cosmological constant @xmath12 .
the corresponding action reads as @xmath13 , \label{action}\ ] ] where @xmath14 is the gauss - bonnet coefficient with dimension @xmath15 ^ 2 $ ] . in this action , @xmath16 is the maxwell field strength given by @xmath17 where @xmath18 is an abelian gauge field . roughly speaking
, the discussion will be given here corresponds to the case with a positive gauss - bonnet coefficient , namely , @xmath19 .
as shown in @xcite , dynamical solutions appear only in higher dimensional theories ( @xmath20 ) .
for this reason , it should be interesting to concentrate on such models . in this way
, the above action produces a static black hole solution with the following metric @xmath21 where @xmath22 is the line element of a @xmath23-dimensional maximal symmetric einstein space with the constant curvature @xmath24 and volume @xmath25 .
it is noted that @xmath26 takes three values @xmath27 , @xmath28 and @xmath29 , corresponding to the spherical , ricci flat and hyperbolic topology of the black hole horizon , respectively . according to @xcite
, the metric function @xmath30 takes the following form @xmath31 where @xmath32 . in this solution
, @xmath33 represents the black hole mass , @xmath34 is linked to the charge of the black hole and @xmath35 . for well - defined vacuum solution associated with @xmath36 ,
the effective gauss - bonnet coefficient @xmath37 and pressure @xmath6 must satisfy the following constraint @xmath38 it is recalled that the mass @xmath33 can be given in terms of the horizon radius @xmath39 of the black hole determined by the largest real root of the equation @xmath40 . it takes the following form @xmath41 similarly , the hawking temperature of the black hole reads as @xmath42 it is worth noting that in the discussion of the thermodynamics of the black hole in the extended phase space by considering the pressure @xmath35 , the black hole mass @xmath33 should be identified with the enthalpy @xmath43 rather than the internal energy of the gravitational system @xcite .
it turns out that many other thermodynamic quantities can be obtained using thermodynamical equations .
for instance , the entropy @xmath44 , thermodynamic volume @xmath7 and electric potential ( chemical potential ) @xmath45 take the following forms @xmath46 @xmath47 @xmath48 these thermodynamic quantities satisfy the following differential form @xmath49 where @xmath50 is the conjugate quantity to the gauss - bonnet coefficient @xmath51 , being considered as a variable . by the scaling argument , the generalized smarr relation for the black holes can be written as @xmath52 moreover , many solutions treating other black holes are given in @xcite .
this class of black hole solutions has volume @xmath53 and an area given by @xmath54 . in this way
, the black hole horizon has a scalar curvature @xmath55 .
the gauss - bonnet term on the horizon is @xmath56 .
in fact , the first term in ( [ beta ] ) can be put in the form @xmath57 , while the second term takes the form @xmath58 .
it is observed that both terms vanish in the flat solution corresponding to @xmath59 .
it is noted that the first term in ( [ beta ] ) is nothing but the second term in the black hole mass ( [ m ] ) , while the second term in ( [ beta ] ) is identified with the second term of the black hole entropy ( [ s ] ) multiplied by hawking temperature @xmath60 .
using legendre transformations , the gibbs free energy and the helmholtz free energy read as @xmath61 however , it is recalled that the helmholtz free energy @xmath62 can be obtained by removing the contribution of the background of the ads vacuum solution . in this way , only the situation associated with a negative @xmath62 can be regarded as the black hole solution considered as a thermodynamically ads vacuum solution . in the case
associated with the hyperbolic horizon ( @xmath63 ) @xcite , the black hole entropy could be negative .
in fact , a negative entropy has no meaning in statistical physics . for these reasons , the constaints @xmath64 should be added .
it is noted that in the case @xmath63 , the metric function @xmath30 ( [ fr ] ) shows the existence of a minimal horizon radius @xmath65 . however
, the non - negative values of the black hole entropy ( [ s ] ) produce a more strong constraint on the horizon radius given by @xmath66 .
having given the essentials on the thermodynamics of the gauss - bonnet - anti - de sitter black holes , we move to investigate the corresponding maxwell s equal area law .
in the study of der waals system , for an isotherm below the critical temperature @xmath67 the two points , of the plan @xmath68 , solving the equation @xmath69 indicate the the stability limit of the system .
it is observed that the critical point corresponding to the largest volume is interpreted as the stability limit of the gaseous phase .
however , the critical point associated with the smallest volume corresponds to the stability limit of the liquid phase .
the chemical potential of such thermodynamic systems should satisfy @xmath70 in the isotherm transformation , the difference of chemical potential between two states with the pressure @xmath6 and @xmath71 should have the following form @xmath72 it is realized that the gauss - bonnet - ads black holes present similar behaviors as illustrated in figure 1 .
curve below the critical temperature.,title="fig : " ] it observed from this figure that , at the point @xmath73 , the black hole lies at `` gas '' phase .
however , at point `` a '' , the black hole lies at `` liquid '' phase completely .
moreover , the region between @xmath74 and @xmath73 can be considered as a coexistence phase
. however , the oscillating part of the curve between @xmath74 and @xmath73 can not be the coexistence line due to the fact that the part @xmath75 violates the equilibrium conditions . at @xmath74 and @xmath73 , the chemical potentials @xmath76 and @xmath77 are @xmath78 this relation produces the thermodynamic condition for the phase equilibrium .
indeed , the equation @xmath79 gives @xmath80 showing that the area @xmath81 and @xmath82 are equal .
+ in what follows , the main aim is to investigate the position of the points @xmath74 and @xmath73 for gauss - bonnet - ads black holes in higher dimensions .
first , we discuss the uncharged case then we will present numerical solutions for the charged black holes in the next section .
in this subsection , we discus the neutral case corresponding to @xmath83 . identifying the pressure with the cosmological constant and the corresponding conjugate thermodynamic volume , the hawking temperature ( [ t ] )
can be explored to write down the state equation .
the latter reads as @xmath84 to make contact with the van der waals equation , we should use a series expansion with the inverse of specific volume @xmath85 .
the calculation leads to @xmath86 identify the specific volume @xmath85 with the horizon radius of the black holes as proposed in @xcite @xmath87 the specific volume @xmath85 can be related to the horizon radius @xmath39 . using ( [ sv ] ) , one can recover all the results in terms of @xmath85 . in this way , the eq .
( [ pq ] ) becomes @xmath88 to obtain the equal area isobar , @xmath89 , one use the following relations : @xmath90 and @xmath91 the equal area law requires the following equality @xmath92 leading to @xmath93 in the isothermal curves , the points @xmath94 and @xmath95 should satisfy @xmath96 @xmath97 from these equations , one can derive the relations @xmath98 and @xmath99 putting @xmath100,(@xmath101 ) , we get the following relations @xmath102 and @xmath103 and @xmath104 using eqs ( [ a]),([b ] ) and([c ] ) , we get a polynomial equation admitting @xmath105 as a real positive root @xmath106 where the coefficients @xmath107 , @xmath108 and @xmath109 read as @xmath110 solving the above polynomial equation , we get @xmath111 in the limit @xmath112 , one should have @xmath113 . in
what follows , we discuss the convergence of this limit in terms of the topology and the dimension of the space - time .
in fact , three situations can appear .
they are classified as follows : * in the case of @xmath114 corresponding to flat topology , this limit diverges for any dimensions @xmath11 .
this shows the absence of the critical points .
indeed , this can be seen from the equation of state eq .
( [ states ] ) .
the letter reduces to @xmath115 indicating that no phase transition can happen . * in the case of @xmath116 associated with the hyperbolic topology , a negative value of @xmath105 appears in @xmath2 . for @xmath117 , the limit also diverges .
this implies that there does not exist any phase transition . * in the case of @xmath118 corresponding to the spherical topology , this limit diverges only when @xmath117 . in the case of @xmath2
, we have @xmath119 it is noted that a similar result has been found in @xcite .
substituting the ( @xmath120 ) in ( @xmath121 ) to eliminate @xmath95 and setting @xmath122 , the critical temperature reads as @xmath123 we can also obtain the following relation @xmath124 in the table @xmath125 , we present the numerical values of the @xmath126 and @xmath71 for different values of the constant @xmath37 .
.numerical solutions for @xmath127 , @xmath94 , @xmath95 and @xmath71 at different temperature with @xmath83 in five dimension with the spherical topology . [ cols="<,<,<,<,<,<",options="header " , ] it is observed from the table ( [ tab3 ] ) that @xmath71 increases with @xmath34 and @xmath128 .
however , the specific volume @xmath95 decreases with @xmath34 and @xmath128 .
the comparison between table ( [ tab2 ] ) and ( [ tab3 ] ) shows the effect of the black hole dimensions .
in fact , the @xmath71 and @xmath95 increases and decreases resepectively with the dimension . to illustrate this effect , we plot these results in the fig.([ffig4 ] ) showing the maxwell equal area .
= diagram of the charged gauss - bonnet - ads black holes in six dimensions , the dashed blue curve corresponds to the critical temperature @xmath1 , the red one corresponds to isotherm with @xmath129 and the green one is asscociated with @xmath130 .
, title="fig : " ] diagram of the charged gauss - bonnet - ads black holes in six dimensions , the dashed blue curve corresponds to the critical temperature @xmath1 , the red one corresponds to isotherm with @xmath129 and the green one is asscociated with @xmath130 .
, title="fig : " ] +
in this paper , we have studied the the maxwell s equal area law of higher dimensional gauss - bonnetanti - de - sitter black holes .
the corresponding critical behaviors share similarity with van der waals one .
we have shown that this construction can be used to eliminate the region of the violated stable equilibrium @xmath131 .
in particular , we have found the isobar line in which the two real phases coexist .
it has been realized that this construction can be viewed as a simple way to drive the coordinates of the critical points .
we have presented numerical calculations showing that the critical behaviors for the uncharged black holes appear only when @xmath3 .
for the charged case , we have studied solutions in @xmath132 and @xmath4 separately and showed that , up to some constrains , the critical behaviors appear only in the spherical topology .
s. h. hendi , s. panahiyan and e. mahmoudi , _ thermodynamic analysis of topological black holes in gauss - bonnet gravity with nonlinear source _ , eur .
j. c * 74 * , no .
10 , 3079 ( 2014 ) [ arxiv:1406.2357 [ gr - qc ] ] .
d. l. wiltshire , _ spherically symmetric solutions of einstein - maxwell theory with a gauss - bonnet term _ , phys .
b * 169 * , 36 ( 1986 ) .
m. cvetic , s. i .
nojiri and s. d. odintsov , _ black hole thermodynamics and negative entropy in de sitter and anti - de sitter einstein - gauss - bonnet gravity _ , nucl .
b * 628 * , 295 ( 2002 ) [ hep - th/0112045 ] . | interpreting the cosmological constant @xmath0 as a thermodynamic pressure and its conjugate quantity as a thermodynamic volume , we study the maxwell s equal area law of higher dimensional gauss - bonnet - ads black holes in extended space .
these black hole solutions critically behave like van der waals systems .
it has been realized that below the critical temperature @xmath1 the stable equilibrium is violated .
we show through numerical calculations that the critical behaviors for the uncharged black holes only appear in @xmath2 . for the charged case , we analyse solutions in @xmath3 and @xmath4 separately and find that ,
up to some constrains , the critical behaviors only appear in the spherical topology . using the maxwell s construction
, we also find the isobar line for which the liquid - gas - like phases coexist . |
all graphs considered in this paper are undirected , finite and simple .
we refer to the book @xcite for graph theoretical notation and terminology not described here . for a graph @xmath2 ,
we by @xmath7 , @xmath8 , @xmath9 , @xmath5 denote the set of vertices , the set of edges , the line graph and the total graph of @xmath2 , respectively .
the following we state the motivations and our results of this paper .
connectivity and edge - connectivity are two of the most basic concepts of graph - theoretic subjects , both in a combinatorial sense and an algorithmic sense . as we know , menger s theorem is the most important basic result and fundamental theorem of connectivity .
based on this theorem , the ` path ' version of _ connectivity _ of a graph @xmath2 is defined as @xmath10 , where @xmath11 is the maximum number of internally disjoint paths connecting two distinct vertices @xmath12 and @xmath13 in @xmath2 .
similarly , the _ edge - connectivity _ of graph @xmath2 is defined as @xmath14 , where @xmath15 is the maximum number of edge - disjoint paths connecting @xmath12 and @xmath13 .
although there are many elegant and powerful results on these two parameters in graph theory , they also have their defects on measuring connection of a graph .
so people want some generalizations of both connectivity and edge - connectivity .
the generalized connectivity of a graph @xmath2 , introduced by hager @xcite , is a natural and nice generalization of the ` path ' version definition of connectivity . for a graph @xmath16 and a set @xmath17 of at least two vertices , _ an @xmath18-steiner tree _ or _ a steiner tree connecting @xmath18 _ ( or simply , _ an @xmath18-tree _ ) is a subgraph @xmath19 of @xmath2 that is a tree with @xmath20 .
two steiner trees @xmath21 and @xmath22 connecting @xmath18 are said to be _ internally disjoint _ if @xmath23 and @xmath24 . for @xmath25 and @xmath26 , the _ generalized local connectivity _
@xmath27 is the maximum number of internally disjoint steiner trees connecting @xmath18 in @xmath2 .
note that when @xmath28 a minimal steiner tree connecting @xmath18 is just a path connecting the two vertices of @xmath18 .
for an integer @xmath0 with @xmath29 , _
generalized @xmath0-connectivity _ ( or _ @xmath0-tree - connectivity _ ) is defined as @xmath30 .
clearly , when @xmath28 , @xmath31 is the connectivity @xmath3 of @xmath2 , that is , @xmath32 . as a natural counterpart of the generalized connectivity , in @xcite x.li and y.mao introduced the concept of generalized edge - connectivity , which is a generalization of the ` path ' version definition of edge - connectivity .
for @xmath25 and @xmath33 , the _ generalized local edge - connectivity _ @xmath34 is the maximum number of edge - disjoint steiner trees connecting @xmath18 in @xmath2 . for an integer @xmath0 with @xmath29 , the _
generalized @xmath0-edge - connectivity _ @xmath35 of @xmath2 is then defined as latexmath:[$\lambda_k(g)= \min\{\lambda(s)\,|\,s\subseteq v(g ) \ and \ results on the generalized edge - connectivity can be found in @xcite .
recently , book@xcite , written by x. li and y. mao , has been published , where authors bring together the known results , conjectures , and open problems on generalized connectivity and edge generalized connectivity . and
thus many researchers pay more attention to the study for this topic .
see @xcite . _ the total graph _
@xmath5 of @xmath2 is graph with the vertex set is @xmath37 and two vertices @xmath38 of @xmath5 are adjacent if one of the following cases hold : ( i ) @xmath39 and @xmath12 is adjacent with @xmath13 in @xmath2 .
( ii ) @xmath40 and @xmath38 are adjacent in @xmath2 .
( iii ) @xmath41 , @xmath42 , and @xmath38 are incident in @xmath2 .
clearly , total graph is generalized line graph .
t. nonaka et al . , in @xcite determined the connectivity and edge connectivity of the total graph @xmath5 . motivated by this , in this paper ,
we investigate the generalized 3-(edge)-connectivity of total graph for some graphs , such as tree , unicycle graph , complete graph and complete bipartite graph in section 3 . and
further we discuss the bounds on 3-generalized-(edge-)connectivity for the total graph in section 4 .
[ obs2 - 1 ] if @xmath2 is a connected graph , then @xmath43 .
[ obs2 - 2 ] if @xmath44 is a spanning subgraph of @xmath2 , then @xmath45 and @xmath46 .
@xcite[pro2 - 1 ] for every two integers @xmath47 and @xmath0 with @xmath29 , @xmath48 @xcite[pro2 - 2 ] for every two integers @xmath47 and @xmath0 with @xmath29 , @xmath49 @xcite[pro2 - 4 ] let @xmath2 be a connected graph of order @xmath50 . then for @xmath51 , @xmath52 . moreover , the upper bound is always sharp for @xmath53 .
@xcite[pro2 - 5 ] for any graph @xmath2 of order @xmath47 .
moreover , the upper bound is tight .
@xcite[pro2 - 7 ] let @xmath2 be a connected graph of order @xmath47 with minimum degree @xmath55 . if there are two adjacent vertices of degree @xmath55 , then @xmath56 .
moreover , the upper bound is sharp .
@xcite[pro2 - 8 ] let @xmath2 be a connected graph of order @xmath47 with minimum degree @xmath55 . if there are two adjacent vertices of degree @xmath55 , then @xmath57 for @xmath58 .
moreover , the upper bound is sharp .
@xcite[pro2 - 9 ] let @xmath2 be a connected graph with @xmath47 vertices . for every two integers @xmath0 and @xmath59 with @xmath60 and @xmath61 , if @xmath62 , then @xmath63 .
moreover , the lower bound is sharp .
@xcite[th2 - 10 ] given any three positive integers @xmath64 such that @xmath65 and @xmath66 , let @xmath67 denote a complete bipartite graph with a bipartition of sizes @xmath68 and @xmath69 .
then @xmath70 is @xmath68 for @xmath71 and @xmath72 @xmath73 for @xmath74 and @xmath75 is odd .
@xmath76 @xmath77 for @xmath74 and @xmath75 is even .
[ cor2 - 11 ] let @xmath78 be two integers with @xmath79 , and @xmath67 denote a complete bipartite graph with a bipartition of sizes @xmath68 and @xmath69 , respectively .
then @xmath80 @xcite[pro2 - 12 ] let @xmath78 be two integers with @xmath79 , and @xmath67 denote a complete bipartite graph with a bipartition of sizes @xmath68 and @xmath69 , respectively
. then @xmath81
in this section , we determine the generalized 3-connectivity and generalized 3-edge - connectivity of some total graph @xmath5 such as total graphs of tree , unicycle graph , complete graph and complete bipartite graph . for @xmath82 , we by @xmath83 $ ] denote the subgraph of @xmath2 which induced by @xmath18 .
let @xmath21 be a subtree of @xmath2 , then @xmath84 .
a spanning tree of induced subgraph @xmath85 $ ] be called _ a corresponding tree _ of @xmath21 in @xmath9 , denoted by @xmath86 .
now we start our investigation with tree @xmath87 with order @xmath47 .
[ lem3 - 1 ] let @xmath88 be a total graph of tree @xmath87 with order @xmath89 .
then the generalized 3-connectivity of @xmath88 is @xmath90 since @xmath91 , so @xmath92 while @xmath93 . here
we consider @xmath94 . since the minimum degree of @xmath88 is 2 , by observation 2.1
, we have @xmath95 . next we prove @xmath96 .
suppose @xmath97 , @xmath98 , then @xmath99 .
let @xmath18 be a 3-subset of @xmath100 , we only need to show that there exist at least two internally disjoint @xmath18-trees in @xmath88 .
if @xmath101 , assume @xmath102 .
by @xmath103 denote the path connecting vertices @xmath104 and @xmath105 in @xmath87 , then we obtain one @xmath18-tree @xmath106 in @xmath87 .
let @xmath86 be the corresponding tree of @xmath21 in @xmath107 and suppose @xmath108 and @xmath109 are neighbor vertices of @xmath110 and @xmath111 in @xmath21 , respectively .
follow this we get another @xmath18-tree @xmath112
. clearly , these two @xmath18-trees are internally disjoint , as desire .
1 ( a ) .
if @xmath113 , assume @xmath114 . then let @xmath115 and @xmath116 .
since @xmath87 and @xmath107 are both connected subgraphs of in @xmath88 , there exist a @xmath117-tree in @xmath87 , written as @xmath118 , and a @xmath119-tree in @xmath107 , written as @xmath120 .
follow this we can obtain two internally disjoint @xmath18-trees : @xmath121 and @xmath122 , as desire . see fig . 1 ( b ) .
if @xmath123 , assume @xmath124 .
then let @xmath115 and @xmath116 .
similarly we obtain a @xmath117-tree @xmath118 in @xmath87 and a @xmath119-tree @xmath120 in @xmath107 .
thus we construct two internally disjoint @xmath18-trees in @xmath88 as : @xmath125 and @xmath126 , as desire .
see fig . 1 ( c ) .
if @xmath127 , assume @xmath128 and let @xmath115 , then we can obtain a @xmath117-tree in @xmath87 , written as @xmath118 .
follow this we can obtain one @xmath18-tree in @xmath87 is @xmath129 .
since @xmath130 and @xmath107 be connected , we get another @xmath18-tree @xmath21 in @xmath107 . clearly , these two @xmath18-trees are internally disjoint , as desire .
see fig . 1 ( d ) . therefore , we get @xmath131 for @xmath94 . ' '' '' figure 1 : two internally disjoint @xmath18-trees for 3-subset @xmath132 : one is dotted , another is dashed .
[ fig1 ] [ lem3 - 1 ] let @xmath88 be a total graph of tree @xmath87 with order @xmath89 .
then the generalized 3-edge - connectivity of @xmath88 is @xmath133 since @xmath134 , so @xmath135 . while @xmath94 , note that the minimum degree of @xmath88 is 2 , by observation 2.1 , we get @xmath136 . on the other hand , by the observation 2.1 and theorem 3.1 , @xmath137 .
thus @xmath138 for @xmath94 . ' '' '' the following we determine the generalized 3-connectivity and the generalized 3-edge - connectivity of unicycle graph .
[ lem3 - 1 ] let @xmath139 be a total graph of unicycle graph @xmath140 with order @xmath141 .
then the generalized 3-connectivity of @xmath139 is @xmath142 first consider @xmath143 .
since @xmath144 is 4-regular graph , by proposition 2.7 we get @xmath145 . on the other hand ,
since @xmath144 is 4-connected , by proposition 2.9 we get @xmath146 .
thus @xmath147 .
for the general case , if @xmath148 , since the minimum degree of @xmath139 is 2 and thus by observation 2.1 get @xmath149 .
on the other hand , suppose @xmath150 is unique cycle of @xmath140 and @xmath151 is an edge of @xmath150 .
now let @xmath152 , clearly , @xmath153 is a spanning tree of @xmath140 and by theorem 3.1 we have @xmath154 .
combine this with observation 2.2 we get @xmath96 . hence @xmath131 for @xmath148 . ' '' '' [ lem3 - 1 ] let @xmath139 be a total graph of unicycle graph @xmath140 with order @xmath141 .
then the generalized 3-edge - connectivity of @xmath139 is @xmath155 while @xmath143 . since @xmath144 is 4-regular graph , by proposition 2.8 we get @xmath156 . at the same time , by observation 2.1 and theorem 3.3 we have @xmath157 .
thus @xmath158 . while @xmath148 , since the minimum degree of @xmath139 is 2 , by observation 2.1 we have @xmath159 . on the other hand , by observation 2.1 and theorem 3.3 we have @xmath160 , so @xmath161 for @xmath148 . ' '' '' next we determine the generalized 3-connectivity and the generalized 3-edge - connectivity of complete graph . before investigation , we first list a useful lemma .
@xcite[lem3 - 1 ] let @xmath162 be a line graph of complete graph @xmath163 with @xmath164 for @xmath165
. suppose @xmath166 and @xmath167 . if the induced subgraph @xmath168=k_{3}$ ] , then generalized local connectivity @xmath169 .
@xcite let @xmath163 be complete graph with order @xmath141 .
then the generalized 3-connectivity of line graph @xmath162 is @xmath170 .
@xcite let @xmath163 be complete graph with order @xmath141 .
then the generalized 3-edge - connectivity of line graph @xmath162 is @xmath171 .
[ lem3 - 1 ] let @xmath172 be a total graph of complete graph @xmath163 with order @xmath89 .
then the generalized 3-connectivity of @xmath172 is @xmath173 by theorem 3.1 and 3.3 , the result holds for cases when @xmath174 .
now we consider @xmath175 .
suppose @xmath176 and @xmath177 , then @xmath178 .
first let @xmath179 .
clearly , @xmath180 and @xmath168=k_{3}$ ] , by lemma 3.5 there exist at most @xmath181 internally disjoint @xmath182-trees in @xmath162 .
besides these @xmath182-trees , add tree @xmath183 together , we obtain at most @xmath184 internally disjoint @xmath182-trees in @xmath172 .
thus we get @xmath185 .
the following we distinguish four cases to show @xmath186 .
let @xmath187 be a 3-subset of @xmath188 , we only need to show that there exist at least @xmath184 internally disjoint @xmath18-trees in @xmath172 .
* case 1*. @xmath189 this means @xmath190 , assume @xmath191 with @xmath192 .
firstly , path @xmath193 together with trees @xmath194 for @xmath195 are @xmath196 internally disjoint @xmath18-trees .
secondly , paths @xmath197 , @xmath198 and trees @xmath199 for @xmath200 are @xmath201 internally disjoint @xmath18-trees .
total up all we get @xmath202 internally disjoint @xmath18-trees in @xmath172 , as desire . *
case 2*. @xmath203 without loss of generality , assume @xmath204 , @xmath205 and then let @xmath206 with @xmath207 . here
first consider the case for @xmath208 .
if edges @xmath209 and @xmath210 are nonadjacent in @xmath163 , then we form internally disjoint @xmath18-trees as : for every @xmath211 to form @xmath212 and thus get @xmath196 internally disjoint @xmath18-trees ; for every @xmath213 to form @xmath214 and thus get @xmath215 internally disjoint @xmath18-trees . put all @xmath216 , @xmath217 with trees @xmath218 and @xmath219 together , we get @xmath220 internally disjoint @xmath18-trees in @xmath172 . if edges @xmath209 and @xmath210 are adjacent in @xmath163 , by similar procedure as the above we also get @xmath220 internally disjoint @xmath18-trees in @xmath172 .
note that @xmath202 , so the result holds .
now we consider the case @xmath221 , it is clear that for every two integers @xmath222 , we can get three internally disjoint @xmath18-trees such as @xmath223 , @xmath224 and @xmath225 and thus get at least @xmath181 internally disjoint @xmath18-trees . put these trees with @xmath226 together , we obtain at least @xmath184 internally disjoint @xmath18-trees in @xmath172 , as we desire . *
case 3*. @xmath227 assume @xmath228 , @xmath229 and then let @xmath230 with @xmath231 . if @xmath232 , @xmath233 and @xmath234 are nonadjacent each other in @xmath172 , we can form internally disjoint @xmath18-trees as : for every @xmath235 to form @xmath236 and thus get @xmath196 internally disjoint @xmath18-trees ; for every @xmath237 to form @xmath238 and thus get @xmath196 internally disjoint @xmath18-trees .
put all these trees with tree @xmath239 together we get @xmath202 internally disjoint @xmath18-trees in @xmath172 , as desire .
if @xmath232 and @xmath233 are adjacent but nonadjacent to @xmath234 , then assume @xmath240 .
thus we form internally disjoint @xmath18-trees as : for every @xmath241 to form @xmath242 and thus get @xmath196 internally disjoint @xmath18-trees ; for every @xmath243 to form @xmath238 and thus get @xmath215 internally disjoint @xmath18-trees .
put all these trees with trees @xmath244 and @xmath245 together , we get @xmath246 internally disjoint @xmath18-trees in @xmath172 , as desire .
if @xmath232 and @xmath233 are adjacent and one of them adjacent to @xmath234 , then assume @xmath240 and @xmath247 .
thus trees @xmath242 for every @xmath248 are @xmath201 internally disjoint @xmath18-trees . besides these , trees @xmath238 for every @xmath249 are also @xmath196 internally disjoint @xmath18-trees . altogether , we get @xmath246 internally disjoint @xmath18-trees in @xmath172 , as desire . if @xmath232 , @xmath234 and @xmath233 are adjacent each other , then assume @xmath250 .
trees @xmath251 for every @xmath241 and @xmath252 for every @xmath253 are @xmath254 internally disjoint @xmath18-trees .
in addition to these , add tree @xmath255 together , we get @xmath202 internally disjoint @xmath18-trees in @xmath172 , as desire .
* case 4*. @xmath256 it is clear @xmath257 in this case , so by lemma 3.6 there exist at least @xmath181 internally disjoint @xmath18-trees in @xmath258 . put these @xmath18-trees with tree @xmath259 together we get @xmath181 + 1 internally disjoint @xmath18-trees in @xmath172 , as desire .
this complete the proof . ' '' '' [ lem3 - 1 ] let @xmath172 be a total graph of complete graph @xmath163 with order @xmath89 .
then the generalized 3-edge - connectivity of @xmath172 is @xmath260 . since @xmath172 is @xmath261 regular graph , by proposition 2.8 @xmath262 .
next we by constructing @xmath220 edge disjoint @xmath18-trees in @xmath172 for any 3-subset @xmath18 of @xmath188 to prove @xmath263 .
recall of the proof of theorem 3.8 , except case @xmath256 and case @xmath203 for @xmath264 with @xmath221 , there always exist at least @xmath220 internally disjoint @xmath18-trees in @xmath172 , which are also edge disjoint @xmath18-trees in @xmath172 . thus here
we only need to consider the above two exception cases and show there still exist at least @xmath220 edge disjoint @xmath18-trees in @xmath172 . if @xmath256 , assume @xmath265 , by lemma 3.7 there exist at least @xmath266 edge disjoint @xmath18-trees in @xmath162 , in addition these trees , add two edge disjoint @xmath18-trees @xmath259 and @xmath267 together , we obtain at least @xmath220 edge disjoint @xmath18-trees in @xmath172 , as desire .
if @xmath203 for @xmath264 with @xmath221 , this means @xmath268 , then we form @xmath18-trees @xmath269 and @xmath270 for every @xmath271\setminus \ { a , b\}$ ] and thus get @xmath254 edge disjoint @xmath18-trees .
add tree @xmath272 together with all @xmath273 and @xmath274 we get @xmath220 edge disjoint @xmath18-trees in @xmath172 , as desire .
thus we get @xmath263 .
this complete the proof . ' '' '' at end of this section , we determine the generalized 3-connectivity and the generalized 3-edge - connectivity of the total graph of the complete bipartite graph .
we start with definition and lemmas . _ the cartesian product _
@xmath275 of @xmath276 and @xmath277 is a graph which has vertex set @xmath278 with two vertices @xmath279 and @xmath280 adjacent iff for @xmath281 , @xmath282 is adjacent with @xmath283 in @xmath277 or @xmath284 , @xmath285 is adjacent with @xmath286 in @xmath276 .
it is clear that line graph @xmath287 of complete bipartite graph @xmath288 is the cartesian product of @xmath289 and @xmath163 and the generalized 3-connectivity of @xmath287 has been determined in our another paper , which listed as follow .
@xcite let @xmath287 be the line graph of complete bipartite graph @xmath290 , then the generalized 3-connectivity of @xmath287 is @xmath291 .
[ lem3 - 1 ] let @xmath292 be a total graph of complete bipartite graph @xmath288 with @xmath293 .
then the generalized 3-connectivity of @xmath292 is @xmath294 suppose @xmath295 and @xmath296 be the two parts of @xmath288 , ie .
, @xmath297 . then @xmath298 and @xmath299 . clearly , by theorem 3.1
, the results hold for case @xmath300 .
the following we consider cases for @xmath301 . *
case 1*. @xmath302 since @xmath303 is @xmath304 regular , by proposition 2.7 we get @xmath305 . the following we prove @xmath306 .
in fact , for any 3-subset @xmath187 of @xmath307 , here we only need to show that there exist at least @xmath308 internally disjoint @xmath18-trees in @xmath303 . *
subcase 1.1*. @xmath309 @xmath309 means @xmath310 .
then either @xmath311 or @xmath312 and @xmath313 .
if @xmath311 , assume @xmath314 ( see fig 2 ( a ) ) , we can form @xmath315 internally disjoint @xmath18-trees such as : @xmath316 and @xmath317 for @xmath318 .
if @xmath312 and @xmath313 , assume @xmath319 ( see fig 2 ( b ) ) , we form @xmath308 internally disjoint @xmath18-trees as follow : @xmath272 , @xmath320 , @xmath321 and @xmath322 , @xmath323 for @xmath324 .
all as we desire .
figure 2 : the subcase for @xmath325 .
[ fig2 ] * subcase 1.2*. @xmath326 here we need discuss five possible cases .
if @xmath327 are nonadjacent each other in @xmath303 , assume @xmath328 ( see fig 3 ( a ) ) .
then @xmath308 internally disjoint @xmath18-trees be formed as follows : @xmath329 , @xmath330 , @xmath331 , @xmath332 and @xmath333 for @xmath324 , @xmath323 for @xmath334 . if @xmath12 is nonadjacent to @xmath335 but @xmath13 and @xmath336 are adjacent in @xmath303 , assume @xmath337 ( see fig 3 ( b ) )
. then @xmath308 internally disjoint @xmath18-trees are formed as : @xmath338 , @xmath339 , @xmath340 and @xmath341 , @xmath342 for @xmath343 .
if @xmath336 is nonadjacent to @xmath38 but @xmath12 and @xmath13 are adjacent in @xmath303 , assume @xmath344 ( see fig 3 ( c ) ) .
then @xmath308 internally disjoint @xmath18-trees be formed as : @xmath345 , @xmath346 , @xmath347 and @xmath348 , @xmath349 for @xmath324 .
if @xmath12 and @xmath336 are both adjacent to @xmath13 but @xmath12 and @xmath336 are nonadjacent in @xmath303 , assume @xmath350 ( see fig 3 ( d ) ) .
then @xmath308 internally disjoint @xmath18-trees be formed as : @xmath226 , @xmath351 , @xmath352 and @xmath353 , @xmath354 for @xmath324 .
if @xmath327 are adjacent each other in @xmath303 , assume @xmath355 ( see fig 3 ( e ) ) .
then @xmath272 , @xmath356 , @xmath357 and @xmath358 , @xmath359 for @xmath324 are @xmath308 internally disjoint @xmath18-trees in @xmath303 . figure 3 : the subcases for @xmath360 .
[ fig3 ] * subcase 1.3*. @xmath361 similarly , we also need discuss five possible cases while @xmath361 . if @xmath327 are nonadjacent each other in @xmath303 , assume @xmath362 ( see fig 4 ( a ) ) .
then @xmath308 internally disjoint @xmath18-trees be formed as : @xmath363 , @xmath364 , @xmath365 , @xmath366 , @xmath367 and @xmath368 , @xmath369 for @xmath370 .
if @xmath335 are adjacent but they are nonadjacent to @xmath12 in @xmath303 , assume @xmath371 ( see fig 4 ( b ) ) .
then @xmath308 internally disjoint @xmath18-trees be constructed as : @xmath372 , @xmath340 , @xmath373 , @xmath374 , @xmath375 and @xmath376 , @xmath377 for @xmath370 .
if @xmath38 are adjacent but they are nonadjacent to @xmath336 in @xmath303 , assume @xmath378 ( see fig 4 ( c ) ) .
then @xmath308 internally disjoint @xmath18-trees be constructed as : @xmath379 , @xmath363 , @xmath380 , @xmath381 , @xmath382 and @xmath383 , @xmath384 for @xmath370 .
if @xmath12 and @xmath336 are both adjacent to @xmath13 but @xmath12 and @xmath336 are nonadjacent in @xmath303 , assume @xmath385 ( see fig 4 ( d ) ) . then @xmath308 internally disjoint @xmath18-trees be constructed as : @xmath226 , @xmath386 , @xmath387 and @xmath388 , @xmath389 for @xmath343 .
if @xmath327 are adjacent each other in @xmath303 , assume @xmath390 ( see fig 4 ( e ) ) .
then @xmath308 internally disjoint @xmath18-trees be constructed as : @xmath391 , @xmath392 , @xmath393 and @xmath394 , @xmath395 for @xmath343 .
figure 4 : the subcases for @xmath396 .
[ fig5 ] * subcase 1.4*. @xmath397 @xmath397 means @xmath398 , without loss generality , suppose @xmath399 , then by lemma 3.10 there always exist @xmath400 internally disjoint @xmath18-trees in @xmath401 , named as @xmath402 .
now we add two @xmath18-trees @xmath403 and @xmath404 to all @xmath216 , here @xmath405 . in total , we get @xmath308 internally disjoint @xmath18-trees in @xmath303 , as desire .
* case 2*. @xmath406 since the minimum degree of @xmath292 is @xmath304 , by observation 2.1 we get @xmath407 .
let @xmath187 be a 3-subset of @xmath408 , now we by forming @xmath304 internally disjoint @xmath18-trees in @xmath292 to prove @xmath409 .
* subcase 2.1*. @xmath325 clearly , either @xmath311 or @xmath312 and @xmath313 .
if @xmath311 , by the proof of subcase 1.1 , there exist @xmath304 internally disjoint @xmath18-trees in @xmath292 . here
we consider the case for @xmath312 and @xmath313 , assume @xmath319 ( see fig 2 ( b)),we can form @xmath304 internally disjoint @xmath18-trees as : @xmath272 , @xmath367 and @xmath410 , @xmath411 for @xmath412 . as desire .
* subcase 2.2*. @xmath360 here we need discuss five possible cases .
if @xmath327 are nonadjacent each other in @xmath292 , assume @xmath413 ( see fig 3 ( a ) ) .
then we form @xmath304 internally disjoint @xmath18-trees as : @xmath329 , @xmath414 and @xmath415 , @xmath416 for @xmath412 . if @xmath13 and @xmath336 are nonadjacent to @xmath12 but @xmath13 and @xmath336 are adjacent in @xmath292 , assume @xmath417 ( see fig 3 ( b ) ) .
then @xmath304 internally disjoint @xmath18-trees be constructed as : @xmath418 , @xmath419 and @xmath420 , @xmath421 for @xmath422 . if @xmath12 and @xmath13 are nonadjacent to @xmath336 but @xmath12 and @xmath13 are adjacent in @xmath292 , assume @xmath344 ( see fig 3 ( c ) ) .
then @xmath304 internally disjoint @xmath18-trees be constructed as : @xmath345 , @xmath423 , @xmath424 , @xmath425 and @xmath348 , @xmath349 for @xmath324 .
if @xmath12 and @xmath336 are both adjacent to @xmath13 but @xmath12 and @xmath336 are nonadjacent in @xmath292 , assume @xmath350 ( see fig 3 ( d ) ) .
then @xmath304 internally disjoint @xmath18-trees be constructed as follows : @xmath226 , @xmath351 , @xmath352 , @xmath426 and @xmath353 , @xmath354 for @xmath324 .
if @xmath327 are adjacent each other in @xmath292 , assume @xmath355 ( see fig 3 ( e ) ) .
then @xmath272 , @xmath427 , @xmath356 , @xmath357 and @xmath358 , @xmath359 for @xmath324 are @xmath304 internally disjoint @xmath18-trees in @xmath292 , as desire . * subcase 2.3*. @xmath396 similarly , we also need discuss five possible cases .
if @xmath327 are nonadjacent each other in @xmath292 , assume @xmath362 ( see fig 4 ( a ) ) .
then @xmath304 internally disjoint @xmath18-trees be constructed as : @xmath363 , @xmath364 , @xmath365 , @xmath366 , @xmath367 , @xmath428 and @xmath429 , @xmath430 for @xmath370 .
if @xmath13 and @xmath336 are adjacent but they are nonadjacent to @xmath12 in @xmath292 , assume @xmath371 ( see fig 4 ( b ) ) .
then @xmath304 internally disjoint @xmath18-trees be constructed as : @xmath372 , @xmath340 , @xmath431 , @xmath432 , @xmath433 , @xmath434 and @xmath376 , @xmath377 for @xmath370 .
if @xmath12 and @xmath13 are adjacent but they are nonadjacent to @xmath336 in @xmath292 , assume @xmath435 ( see fig 4 ( c ) ) .
then @xmath304 internally disjoint @xmath18-trees be constructed as : @xmath436 , @xmath330 , @xmath437 , @xmath438 and @xmath439 , @xmath384 for @xmath324 .
if @xmath12 and @xmath336 are both adjacent to @xmath13 but @xmath12 and @xmath336 are nonadjacent in @xmath292 , assume @xmath385 ( see fig 4 ( d ) ) . then @xmath304 internally disjoint @xmath18-trees be constructed as : @xmath226 , @xmath386 , @xmath440 , @xmath441 and @xmath388 , @xmath389 for @xmath343 .
if @xmath327 are adjacent each other in @xmath292 , assume @xmath390 ( see fig 4 ( e ) ) .
then @xmath304 internally disjoint @xmath18-trees be constructed as : @xmath391 , @xmath442 , @xmath443 , @xmath444 and @xmath394 , @xmath395 for @xmath343 . *
subcase 2.4*. @xmath445 @xmath445 means @xmath446 , assume @xmath399 , then by lemma 3.10 there exist @xmath447 internally disjoint @xmath18-trees in @xmath287 , named as @xmath448 .
now we add two @xmath18-trees @xmath403 and @xmath404 to all @xmath216 , here @xmath405 .
thus we get @xmath449 internally disjoint @xmath18-trees in @xmath292 , as desire .
by now we complete the proof . ' '' '' note that the fact @xmath292 is @xmath304 -regular graph while @xmath302 and its minimum degree is @xmath304 while @xmath406 .
combine this with observation 2.1 and theorem 3.11 , we immediately get theorem 3.12 .
let @xmath292 be a total graph of complete bipartite graph @xmath290 .
then the generalized 3-edge - connectivity of @xmath292 is @xmath450
in this section we give some bounds for the generalized 3-connectivity of total graph @xmath5 .
since the minimum degree of @xmath5 is @xmath457 , so by proposition 2.6 we get @xmath458 . on the other hand , by lemma 4.1 , we know @xmath459 and let @xmath460 with @xmath461 .
then we get @xmath462 .
combine this with proposition 2.9 we have @xmath463 suppose @xmath475 and @xmath476 , let @xmath187 be a 3-subset of @xmath477 .
the following we by constructing edge disjoint @xmath18-trees in @xmath5 to prove @xmath474 . now distinguish four cases to complete the proof .
@xmath478 means @xmath25 , since there are @xmath479 edge disjoint @xmath18-trees in @xmath2 , name them as @xmath216 .
in addition to these , it is clear that every @xmath216 has a corresponding tree @xmath480 in @xmath9 and every @xmath480 can be formed as a @xmath18-tree by connecting it with @xmath18 .
thus altogether we can get at least @xmath481 edge disjoint @xmath18-trees in @xmath5 .
suppose @xmath39 , @xmath483 and let @xmath484 . if @xmath485 $ ] is a triangle , since there are @xmath471 edge disjoint @xmath486-paths in @xmath2 and thus assume @xmath487 is a @xmath486-path .
based on this @xmath486-path we can form 2 edge disjoint @xmath18-trees in @xmath5 as : @xmath488 and @xmath489 .
thus we total get @xmath490 edge disjoint @xmath18-trees in @xmath5 . if @xmath491 but @xmath492 , assume @xmath493 .
since there exist @xmath479 edge disjoint @xmath117-trees in @xmath2 , name them as @xmath216 . for every pair trees @xmath216 and its corresponding tree @xmath480 , we by symmetric difference operation on @xmath2 and @xmath9 can obtain @xmath481 edge disjoint @xmath18-trees in @xmath5 .
if @xmath494 and @xmath492 , assume @xmath495 .
since there exist @xmath479 edge disjoint @xmath117-trees in @xmath2 , named as @xmath216 . for every pair trees @xmath216 and its corresponding tree @xmath480 , we also can obtain at least@xmath496 edge disjoint @xmath18-trees in @xmath5 .
assume @xmath41 , @xmath498 and let @xmath499 .
no matter what vertices @xmath327 they are , there exist 3 elements among @xmath500 which can form a 3-subset @xmath501 .
since there exist @xmath479 edge disjoint @xmath117-trees in @xmath2 , named as @xmath216 .
similarly , based on every pair @xmath216 and its corresponding tree @xmath480 we get at least @xmath496 edge disjoint @xmath18-trees in @xmath5 .
this case means @xmath503 , by lemma 4.2 there exist @xmath479 internally disjoint @xmath18-trees in @xmath9 , which are also edge disjoint @xmath18-trees . based on this , using vertices in graph @xmath2 can obtain at least @xmath496 edge disjoint @xmath18-trees in @xmath5 . | the generalized @xmath0-connectivity @xmath1 of a graph @xmath2 , introduced by hager in 1985 , is a natural generalization of the concept of connectivity @xmath3 , which is just for @xmath4 .
total graph is generalized line graph and a large graph which obtained by incidence relation between vertices and edges of original graph .
t. hamada and t. nonaka et al . , in @xcite determined the connectivity of the total graph @xmath5 for a graph @xmath2 . in this paper , we determine the generalized @xmath6-(edge)-connectivity of some total graphs and give the bounds on the generalized @xmath6-(edge)-connectivity for total graph . +
* keywords : * generalized connectivity , generalized edge connectivity , total graph . + * ams subject classification 2010 : * 05c05 , 05c40 , 05c70 , 05c75 . |
since halmos s observation @xcite in 1943 that automorphisms of compact groups automatically preserve haar measure , these maps have provided a rich class of examples in dynamics . in case
the group is abelian , its dual group is a module over the laurent polynomial ring @xmath0 $ ] .
such modules have a well - developed structure theory , which enables a comprehensive analysis of general automorphisms in terms of basic building blocks that can be completely understood .
the roots of the study of several commuting algebraic maps can be traced back to the seminal 1967 paper of furstenberg @xcite , where he considered the joint dynamical properties of multiplication by different integers on the additive torus . in 1978
ledrappier @xcite gave a simple example of a mixing action of @xmath1 by automorphisms of a compact abelian group that was not mixing of higher orders .
for an action by @xmath2 commuting automorphisms , the dual group is a module over the laurent polynomial ring @xmath3 $ ] , i.e. , the integral group ring @xmath4 of @xmath5 .
the commutative algebra of such modules provides effective machinery for analyzing such actions .
this point of view was initiated in 1989 by kitchens and the second author @xcite , and a fairly complete theory of the dynamical properties of such actions is now available @xcite .
let @xmath6 denote an arbitrary countable group , and let @xmath7 denote an action of @xmath6 by automorphisms of a compact abelian group , or an _
algebraic @xmath6-action_. the initial steps in analyzing such actions were taken in ( * ? ? ?
* chap . 1 ) , and give general criteria for some basic dynamical properties such as ergodicity and mixing . in 2001
einsiedler and rindler @xcite investigated the particular case when @xmath8 , the discrete heisenberg group , as a first step towards algebraic actions of noncommutative groups . here
the concrete nature of @xmath9 suggests that there should be specific answers to the natural dynamical questions , and they give several instances of this together with instructive examples .
however , the algebraic complexity of the integral group ring @xmath10 prevents the comprehensive analysis available in the commutative case .
a dramatic new development occurred in 2006 with the work of deninger on entropy for principal @xmath6-actions .
let @xmath11 , and let @xmath12 denote the principal left ideal generated by @xmath13 .
then @xmath6 acts on the quotient @xmath14 , and there is a dual @xmath6-action @xmath15 on the compact dual group , called a _
principal @xmath6-action_. deninger showed in @xcite that in many cases the entropy of @xmath15 equals the logarithm of the fuglede - kadison determinant the linear operator corresponding to @xmath13 on the group von neumann algebra of @xmath6 . in case @xmath16 , this reduces to the calculation in @xcite of entropy in terms of the logarithmic mahler measure of @xmath13 .
subsequent work by deninger , li , schmidt , thom , and others shows that this and related results hold in great generality ( see for example @xcite , @xcite , and @xcite ) . in @xcite
the authors proved that three different concepts connected with @xmath6-actions , namely entropy , fuglede - kadison determinants , and @xmath17-torsion , coincide , revealing deep connections that are only partly understood .
these ideas have some interesting consequences .
for example , by computing the entropy of a particular heisenberg action in two different ways , we can show that @xmath18 for almost every pair @xmath19 of real numbers . despite its simplicity
, this fact does not appear to follow from known results on random matrix products .
our purpose here is to survey what is known for the heisenberg case @xmath8 , and to point out many of the remaining open questions . as @xmath9 is the simplest noncommutative example ( other than finite extensions of @xmath20 , which are too close to the abelian case to be interesting ) , any results will indicate limitations of what a general theory can accomplish
also , the special structure of @xmath9 should enable explicit answers to many questions , and yield particular examples of various dynamical phenomena .
it is also quite instructive to see how a very general machinery , used for algebraic actions of arbitrary countable groups , can be made quite concrete for the case of @xmath9 .
we hope to inspire further work by making this special case both accessible and attractive .
let @xmath6 be a countable discrete group .
the _ integral group ring _
@xmath21 of @xmath6 consists of all finite sums of the form @xmath22 with @xmath23 , equipped with the obvious ring operations inherited from multiplication in @xmath6 .
the _ support _ of @xmath24 is the subset @xmath25 .
suppose that @xmath6 acts by automorphisms of a compact abelian group @xmath26 .
such actions are called _
algebraic @xmath6-actions_. denote the action of @xmath27 on @xmath28 by @xmath29 .
let @xmath30 be the ( discrete ) dual group of @xmath26 , with additive dual pairing denoted by @xmath31 for @xmath28 and @xmath32 .
then @xmath30 becomes a module over @xmath21 by defining @xmath33 to be the unique element of @xmath30 so that @xmath34 for all @xmath35 , and extending this to additive integral combinations in @xmath21 . conversely , if @xmath30 is a @xmath21-module , its compact dual group @xmath36 carries a @xmath6-action @xmath37 dual to the @xmath6-action on @xmath30 .
thus there is a 1 - 1 correspondence between algebraic @xmath6-actions and @xmath21-modules .
let @xmath38 be the additive torus .
then the dual group of @xmath39 can be identified with @xmath40 via the pairing @xmath41 where @xmath42 and @xmath43 . for @xmath44 , the action of @xmath45 on @xmath46 is defined via duality by @xmath47 for all @xmath48 . by taking @xmath49
, we obtain that @xmath50 .
it is sometimes convenient to think of elements in @xmath51 as infinite formal sums @xmath52 , and then @xmath53 .
this allows a well - defined multiplication of elements in @xmath51 by elements from @xmath21 , both on the left and on the right .
we remark that the shift - action @xmath54 is opposite to the traditional shift direction when @xmath6 is @xmath55 or @xmath5 , but is forced when @xmath6 is noncommutative .
this has sometimes caused confusion ; for example the last displayed equation in @xcite is not correct .
now fix @xmath11 .
let @xmath12 be the principal left ideal generated by @xmath13 .
the quotient module @xmath14 has dual group @xmath56 .
an element @xmath46 is in @xmath57 iff @xmath58 for all @xmath48 .
this is equivalent to the condition that @xmath59 for all @xmath48 , where @xmath60 .
hence @xmath61 exactly when @xmath62 , using the conventions above for right multiplication of elements in @xmath51 by members of @xmath21 .
in other words , if we define @xmath63 to be right convolution by @xmath64 , then @xmath57 is the kernel of @xmath65 . in terms of coordinates
, @xmath66 is in @xmath57 precisely when @xmath67 for all @xmath44 .
our focus here is on the discrete heisenberg group @xmath9 , generated by @xmath68 , @xmath69 , and @xmath70 subject to the relations @xmath71 , @xmath72 , and @xmath73 .
alternatively , @xmath9 is the subgroup of @xmath74 generated by @xmath75 we will sometimes use the notation @xmath76 for the integral group ring @xmath10 of @xmath9 when emphasizing its ring - theoretic properties .
the center of @xmath9 is @xmath77 .
the center of @xmath76 is then the laurent polynomial ring @xmath78 $ ] .
hence every element of @xmath76 can be written as @xmath79 where @xmath80 and @xmath81 . for @xmath82 , define the _
newton polygon _ @xmath83 of @xmath24 to be the convex hull in @xmath84 of those points @xmath85 for which @xmath86 .
in particular , @xmath87 . because @xmath88 is an integral domain ,
it is easy to verify that @xmath89 equals the minkowski sum @xmath90 for all @xmath91 .
this shows that @xmath10 has no nontrivial zero - divisors .
however , a major difference between the commutative case and @xmath10 is that unique factorization into irreducibles fails for @xmath10 .
it is easy to verify that @xmath92 each of the linear factors is clearly irreducible by a newton polygon argument .
we claim that @xmath93 can not be factored in @xmath10 .
note that @xmath94 ^ 2 $ ] .
suppose that @xmath95 .
adjusting by units and reordering factors if necessary , we may assume that @xmath24 and @xmath96 have the form @xmath97 and @xmath98 .
expanding @xmath99 , we find that @xmath100 hence @xmath101 and @xmath102 .
then we would have @xmath103 this proves that @xmath13 has no nontrivial factorizations in @xmath10 .
since @xmath9 is nilpotent of rank 2 , it is polycyclic , and so @xmath76 is both left- and right - noetherian , i.e. @xmath76 satisfies the ascending chain condition on both left ideals and on right ideals @xcite .
suppose now that @xmath30 is a finitely generated left @xmath76-module , say generated by @xmath104 .
the map @xmath105 defined by @xmath106\mapsto g_1m_1+\dots g_l m_l$ ] is surjective .
its kernel @xmath107 is a left @xmath76-submodule of the noetherian module @xmath108 , hence also finitely generated , say by @xmath109 , \dots,[f_{k1},\dots , f_{kl}]$ ] .
let @xmath110\in r^{k\times l}$ ] be the rectangular matrix whose rows are the generators of @xmath107 .
then @xmath111 , and @xmath112 .
we will denote the corresponding algebraic @xmath9-action for this presentation of @xmath30 by @xmath113 . notice that when @xmath114 we are reduced to the case @xmath115 $ ] , corresponding to the quotient module @xmath116 and the principal @xmath9-action @xmath15 .
let @xmath26 be a compact abelian group and let @xmath117 denote haar measure on @xmath26 , normalized so that @xmath118 . if @xmath119 is a continuous automorphism of @xmath26 , then the measure @xmath120 defined by @xmath121 is also a normalized translation - invariant measure .
hence @xmath122 , and @xmath117 is @xmath119-invariant .
this shows that if @xmath7 is an algebraic action of a countable group @xmath6 on @xmath26 , then @xmath7 is @xmath117-measure - preserving .
a measurable set @xmath123 is _ @xmath7-invariant _ if @xmath124 agrees with @xmath125 off a null set for every @xmath27 .
the action @xmath7 is _ ergodic _ if the only @xmath7-invariant sets have measure 0 or 1 .
the following , which is a special case of a result due to kaplansky @xcite , gives an algebraic characterization of ergodicity .
let @xmath6 be a countable discrete group , and @xmath7 be an algebraic @xmath6-action on a compact abelian group @xmath26 whose dual group is @xmath30
. then @xmath7 is ergodic if and only if the @xmath6-orbit of every nonzero element of @xmath30 is infinite .
roughly speaking , this result follows from the observation that the existence of a bounded measurable @xmath7-invariant function on @xmath26 is equivalent to the existence of a nonzero finite @xmath6-orbit in @xmath30 . for actions of the heisenberg group @xmath9
, this raises the question of characterizing those @xmath126 for which @xmath113 is ergodic .
the first result in this direction is due to ben hayes .
[ thm : hayes ] for every @xmath127 the principal algebraic @xmath9-action @xmath15 is ergodic . we give a brief sketch of the proof .
first recall that @xmath88 is a unique factorization domain .
define the _ content _ @xmath128 of @xmath129 to be the greatest common divisor in @xmath88 of the nonzero coefficient polynomials @xmath130 , and put @xmath131 .
a simple variant of the proof of gauss s lemma shows that @xmath132 for all @xmath91 .
now fix @xmath133 .
the case @xmath134 is trivial , so assume that @xmath135 .
suppose that @xmath136 has finite @xmath9-orbit in @xmath116 . then there are @xmath137 such that @xmath138 and @xmath139 for some @xmath140 . then @xmath141 so that @xmath142 divides @xmath143 in @xmath88 . also , @xmath144 so that @xmath145=c(g_2)$ ] , and hence @xmath146 divides @xmath147 .
thus @xmath148 , and so @xmath149f\in rf$ ] , showing that @xmath150 in @xmath116 .
hayes called a group @xmath6 _ principally ergodic _ if every principal algebraic @xmath6-action is ergodic .
he extended theorem [ thm : hayes ] to show that the following classes of groups are principally ergodic : torsion - free nilpotent groups that are not virtually cyclic ( i.e. , do not contain a cyclic subgroup of finite index ) , free groups on more than one generator , and groups that are not finitely generated . clearly @xmath55 is not principally ergodic , since for example the action of @xmath68 on the module @xmath0/{\langle}x^k -1{\rangle}$ ] dualizes to a @xmath151 permutation matrix on @xmath152 , which is not ergodic .
recently li , peterson , and the second author used a very different approach to proving principal ergodicity , based on cohomology @xcite .
these methods greatly increased the collection of countable discrete groups known to be principally ergodic , including all such groups that contain a finitely generated amenable subgroup that is not virtually cyclic .
we will describe now how their ideas work in the case of @xmath9 .
we begin by describing two important properties of finite - index subgroups of @xmath9 , namely that they are amenable and have only one end . for an arbitrary discrete group @xmath6
let @xmath153 and @xmath154 , so that @xmath155 is the dual space to @xmath156 .
fix @xmath157 , and put @xmath158 .
let @xmath159 , the finite - index subgroup of @xmath9 generated by @xmath160 , and @xmath161 .
[ lem : amenability ] let @xmath162 and suppose that @xmath163 is an action of @xmath164 by continuous affine operators on @xmath165 . if @xmath166 is a weak*-compact , convex subset of @xmath165 such that @xmath167 for every @xmath168 , then there is a common fixed point @xmath169 for all the @xmath170 .
put @xmath171 the condition of the powers of @xmath70 is imposed so that any distortion caused by left multiplication of @xmath172 by a given element @xmath168 is eventually small .
more precisely , for every @xmath168 we have that @xmath173 where @xmath174 denotes symmetric difference and @xmath175 denotes cardinality .
now fix @xmath176 , and let @xmath177 .
then @xmath178 since @xmath166 is convex .
since @xmath166 is weak*-compact , there is a subsequence @xmath179 converging weak * to some @xmath169 .
note that @xmath180 by compactness of @xmath166 .
then since each @xmath181 is continuous , we have that @xmath182 for every @xmath183 .
furthermore , @xmath184 it follows that @xmath185 for all @xmath186 .
the essential point in the previous proof is that @xmath164 is amenable , and that @xmath187 forms a flner sequence .
we call a set @xmath188 _ almost invariant _ if @xmath189 is finite for every @xmath168 .
clearly , if @xmath190 is almost invariant , then so is @xmath191 for every @xmath192 .
[ lem : ends ] let @xmath188 be an infinite almost invariant subset .
then @xmath193 is finite .
let @xmath194 be a set of generators for @xmath164 .
the _ cayley graph _
@xmath195 of @xmath164 with respect to @xmath196 has as vertices the elements of @xmath164 , and for every vertex @xmath197 and @xmath198 there is a directed edge from @xmath197 to @xmath199 .
let @xmath125 be the union of @xmath200 over @xmath198 , so @xmath125 is finite .
we can therefore enclose @xmath125 in a box of the form @xmath201 . since @xmath190 is infinite , choose @xmath202 .
then for every @xmath203 there is a finite directed path in @xmath195 from @xmath204 to @xmath205 that avoids @xmath206 , say with vertices @xmath204 , @xmath207 , @xmath208 , @xmath209 , @xmath210 , and by definition of @xmath125 each of these is in @xmath190 .
hence @xmath211 , and so is finite .
suppose that @xmath212 with @xmath213 having finite @xmath9-orbit in @xmath214 .
we will prove that @xmath212 .
there are @xmath215 such that @xmath216 and @xmath217 .
then @xmath218 also stabilizes @xmath213 .
let @xmath158 . then there are @xmath219 such that @xmath220 , @xmath221 , and @xmath222 .
consider the finite - index subgroup @xmath162 of @xmath9 as above .
for every @xmath168 there is a @xmath223 such that @xmath224 , and this is unique since @xmath10 has no zero - divisors .
then @xmath225 is a _ cocycle _ , that is , it obeys @xmath226 for all @xmath227 .
consider @xmath10 as a subset of @xmath165 .
we claim that @xmath228 is a uniformly bounded cocycle , i.e. , that @xmath229 .
the reason for this is that we can calculate the value of @xmath230 for arbitrary @xmath197 using left shifts of the generators @xmath231 that are sufficiently spread out to prevent large accumulations of coefficients .
for example , if @xmath232 , then applying the cocycle equation first for powers of @xmath233 , then powers of @xmath234 , and then powers of @xmath161 , we get that @xmath235 since the supports of the @xmath231 are finite , there is a uniform bound @xmath236 so that for every @xmath237 and every @xmath168 , there are at most @xmath238 summands in the expression for @xmath230 whose support contains @xmath239 .
hence @xmath240 , establishing our claim .
now let @xmath166 be the closed , convex hull of @xmath241 in @xmath165 , which is weak*-compact since the @xmath230 are uniformly bounded .
consider the continuous affine maps @xmath242 defined by @xmath243 .
then @xmath244 by the cocycle property of @xmath228 , and @xmath245 , so that @xmath167 for all @xmath168 . by lemma [ lem : amenability ]
there is a common fixed point @xmath246 for the @xmath170 , so that @xmath247 for all @xmath168 .
write each @xmath248 with @xmath249 and @xmath250 . then @xmath251 so that @xmath252 also satisfies that @xmath253 , where now @xmath254 has integer coordinates . replacing @xmath252 with @xmath255 , we have found a @xmath256 with @xmath257 and @xmath258 for all @xmath168 .
next , we use lemma [ lem : ends ] to show that we can replace @xmath252 with an element of @xmath259 having finite support , and so is an element of @xmath10 .
fix @xmath237 .
for @xmath260 consider the `` level set '' for the restriction of @xmath252 to the right coset @xmath261 , @xmath262 .
we claim that for each @xmath239 there is exactly one @xmath263 for which @xmath264 is infinite . for suppose that @xmath264 is infinite .
let @xmath186 . since @xmath265 has finite support , @xmath266 for all but finitely many @xmath168 .
hence @xmath267 for every @xmath186 . by lemma [ lem : ends ] , we see that @xmath268 is finite . hence for every @xmath237 we can adjust the value of @xmath252 on the coset @xmath261 so that the restriction of @xmath252 to @xmath261 has finite support . doing this for each of the finitely many right cosets @xmath261 results in a @xmath252 with finite support on @xmath9 , so that @xmath269 .
thus @xmath270 for every @xmath168 . since @xmath271
, we obtain @xmath272 , as required .
theorem [ thm : hayes ] answers the @xmath273 case of the following natural question .
[ prob : ergodicity ] describe or characterize those @xmath126 for which @xmath113 is ergodic , or , equivalently , those noetherian @xmath76-modules @xmath30 for which @xmath37 is ergodic .
is there a finite algorithm that will decide whether or not a given @xmath113 is ergodic
? are there easily checked sufficient conditions on @xmath274 for ergodicity of @xmath113 ?
einsiedler and rindler provided one answer to problem [ prob : ergodicity ] , which involves the notion of prime ideals in @xmath76 .
a two - sided ideal @xmath275 in @xmath76 is _ prime _
if whenever @xmath276 and @xmath277 are two - sided ideals in @xmath76 with @xmath278 , then either @xmath279 or @xmath280 . if @xmath281 is an @xmath76-submodule of an @xmath76-module @xmath30 , then the annihilator @xmath282 of @xmath281 is defined as @xmath283 , which is a two - sided ideal in @xmath76 .
a prime ideal @xmath275 is _ associated to _
@xmath30 if there is a submodule @xmath284 such that for every nonzero submodule @xmath285 we have that @xmath286 .
every noetherian @xmath76-module has associated prime ideals , and there are only finitely many of them .
call a prime ideal @xmath275 of @xmath76 _ ergodic _ if the subgroup @xmath287 of @xmath9 has infinite index in @xmath9 . for instance , the ideal @xmath275 in example [ exam : nonergodic ] is prime ( being the kernel of a ring homomorphism onto an commutative integral domain ) , but is not ergodic .
it is easy to verify that a prime ideal is ergodic if and only if @xmath288 is ergodic , and that the only ideal associated with @xmath289 is @xmath275 .
let @xmath30 be a noetherian @xmath76-module
. then @xmath37 is ergodic if and only if every prime ideal associated with @xmath30 is ergodic .
[ exam : nonergodic ] let @xmath275 be the left ideal in @xmath76 generated by @xmath290 and @xmath291
. then @xmath292 and so the map @xmath293 defined by @xmath294 is a well - defined surjective ring homomorphism with kernel @xmath275 .
thus @xmath275 is a prime ideal with @xmath289 isomorphic to @xmath55 .
the dual @xmath9-action is simply the identity map on @xmath295 , which is nonergodic .
hence @xmath275 is a left ideal generated by two elements with @xmath288 nonergodic , showing that theorem [ thm : hayes ] does not extend to nonprincipal actions .
we remark that if we consider @xmath296 instead of @xmath9 , then the characterization of ergodic prime ideals in ( * ? ? ?
6.5 ) shows that their complex variety is finite , and in particular by elementary dimension theory they must have at least three generators .
a relatively explicit description of all prime ideals in @xmath76 is given in @xcite .
characterize the ergodic prime ideals in @xmath76 .
an answer to this problem would reduce problem [ prob : ergodicity ] to computing the prime ideals associated to a given noetherian @xmath76-module .
however , this appears to be difficult , even for modules of the form @xmath116 , although it follows from theorem [ thm : hayes ] that all prime ideals associated to @xmath116 must be ergodic .
let @xmath6 be a countable discrete group , and let @xmath30 be a left @xmath21-module . denote by @xmath117 haar measure on @xmath297 .
the associated algebraic @xmath6-action @xmath37 is called _ mixing _ if , for every pair of measurable sets @xmath298 , we have that @xmath299 , where @xmath300 refers to the one - point compactification of @xmath6 . for @xmath32 ,
the _ stabilizer _ of @xmath301 is the subgroup @xmath302 .
[ prop : general - mixing ] the algebraic @xmath6-action @xmath37 is mixing if and only if the stabilizer of every nonzero @xmath32 is finite . in the case
@xmath8 , this is equivalent to requiring that for every nonzero @xmath32 the map @xmath303 is injective on @xmath9 . using this together with some of the ideas from the previous section , we can give a simple criterion for @xmath304 $ ] so that @xmath305 is mixing .
let @xmath306 .
then the principal @xmath9-action @xmath305 is mixing if and only if @xmath307 has no roots that are roots of unity .
suppose first that @xmath307 has a root that is a root of unity , so that @xmath307 has a factor @xmath308 dividing @xmath146 for some @xmath309 .
then @xmath310 , but @xmath311g\in{{\mathbb{z}}{\gamma}}g$ ] , so that @xmath305 is not mixing by prop .
[ prop : general - mixing ] .
conversely , suppose that @xmath24 has no root that is a root of unity .
recall that the content @xmath143 of @xmath312 is the greatest common divisor in @xmath88 of the @xmath313 .
then @xmath314 if and only if @xmath315 .
suppose that @xmath316 with @xmath317 .
then @xmath24 divides @xmath318 , showing that @xmath314 .
similarly , if @xmath319 , then @xmath320 , and by assumption @xmath24 is relatively prime to @xmath321 .
hence again @xmath322 , and so @xmath314 . then prop .
[ prop : general - mixing ] shows that @xmath305 is mixing . using more elaborate algebra , hayes found several sufficient conditions on @xmath127 for @xmath15 to be mixing . to make the cyclotomic nature of these conditions clear , recall that the @xmath323-th _ cyclotomic polynomial
@xmath324 is given by @xmath325 , where the product is over all primitive @xmath323-th roots of unity .
each @xmath324 is irreducible in @xmath326 $ ] , and @xmath327 is the irreducible factorization of @xmath328 in @xmath326 $ ] .
let @xmath329 be @xmath330 commuting variables .
then a _ generalized cyclotomic polynomial in _
@xmath331 $ ] is one of the form @xmath332 for some @xmath333 and choice of integers @xmath334 , not all @xmath335 .
there is a well - defined ring homomorphism @xmath336 $ ] , where @xmath337 and @xmath338 are commuting variables , given by @xmath339 , @xmath340 , and @xmath341 . for @xmath342 , its image under @xmath343 is @xmath344 .
each of the following conditions on @xmath127 is sufficient for @xmath345 to be mixing : 1 .
@xmath346 $ ] and @xmath13 is not divisible by a generalized cyclotomic polynomial in @xmath68 and @xmath70 .
2 . @xmath347 $ ] and @xmath13 is not divisible by a generalized cyclotomic polynomial in @xmath69 and @xmath70 .
3 . @xmath348 with the content @xmath142 not divisible by a cyclotomic polynomial in @xmath70 , and @xmath349 not divisible by a cyclotomic polynomial in @xmath337 and @xmath338 .
\(a ) if @xmath350 , then @xmath345 is mixing by part ( 3 ) .
\(b ) if @xmath351 , then @xmath345 is mixing by part ( 1 ) , yet @xmath352 is cyclotomic , showing that part ( 3 ) is not always necessary .
\(c ) a generalized cyclotomic polynomial in @xmath353 and @xmath354 has a root both of whose coordinates have absolute value 1 .
it follows that , for example , @xmath355 can not be divisible by any generalized cyclotomic polynomial . then part ( 3 ) above implies that for every choice of nonzero integers @xmath356 , the polynomial @xmath357 yields a mixing @xmath345 .
more generally , if @xmath358 is not divisible by a generalized cyclotomic polynomial , and @xmath359 all satisfy @xmath360 and have no common root that is a root of unity , then @xmath361 results in a mixing @xmath15 .
does there exist a finite algorithm that decides , given @xmath127 , whether or not @xmath15 is mixing ? more generally , is there such an algorithm that decides mixing for @xmath9-actions of the form @xmath362 , where @xmath363 ? there is another , simply stated , sufficient condition for @xmath15 to be mixing . recall that @xmath364 is a banach algebra under convolution , with identity element 1 . if @xmath127 is invertible in @xmath364 , then @xmath15 is mixing .
if @xmath15 were not mixing , there would be an @xmath365 and an infinite subgroup @xmath366 such that @xmath367 for all @xmath168 , say @xmath368 .
let @xmath369 be the inverse of @xmath13 .
then @xmath370 . letting @xmath371 in @xmath164
, we see that @xmath372 .
hence @xmath373 , showing that @xmath15 is mixing
. we will see in theorem [ thm : expansive ] that invertibility of @xmath13 in @xmath364 corresponds to an important dynamical property of @xmath345 .
let @xmath6 be a countable discrete group and @xmath7 be an algebraic @xmath6-action on a compact abelian group @xmath26 .
then @xmath7 is called _ expansive _ if there is a neighborhood @xmath374 of @xmath375 in @xmath26 such that @xmath376 .
all groups @xmath26 we consider are metrizable , so let @xmath2 be a metric on @xmath26 compatible with its topology . by averaging @xmath2 over @xmath26
, we may assume that @xmath2 is translation - invariant .
then @xmath7 is expansive provided there is a @xmath377 such that if @xmath378 for all @xmath27 , then @xmath379 .
expansiveness is an important and useful property , with many implications .
it is therefore crucial to know when algebraic actions are expansive . for principal actions
there is a simple criterion .
[ thm : expansive ] let @xmath6 be a countable discrete group and let @xmath11
. then the following are equivalent : 1 .
the principal algebraic action @xmath15 on @xmath57 is expansive , 2 .
the principal algebraic action @xmath380 on @xmath381 is expansive , 3 .
f is invertible in @xmath156 , 4 .
@xmath382 is injective on @xmath155 . before sketching the proof
, we isolate a crucial property of @xmath156 called _ direct finiteness _ : if @xmath383 with @xmath384 , then @xmath385 , where @xmath386 denotes the usual convolution product in @xmath156 .
this was originally proved by kaplansky @xcite using von neumann algebra techniques .
later montgomery @xcite gave a short proof using @xmath387-algebra methods . a more self - contained argument using only elementary ideas
was given by passman @xcite .
all these arguments use a key feature of @xmath156 , that it has a faithful trace function @xmath388 given by @xmath389 .
this function has the properties that it is linear , @xmath390 , @xmath391 , and @xmath392 , with equality iff @xmath393 .
the key argument is that if @xmath394 , then @xmath395 , and @xmath396 implies that @xmath397 . if @xmath384 , then @xmath398 satisfies that @xmath399 , and @xmath400 , hence @xmath385 .
let @xmath401 be the usual metric on @xmath38 defined by @xmath402 .
it is straightforward to check that we may use the pseudometric @xmath403 on @xmath57 defined by @xmath404 to determine expansiveness ( see ( * ? ? ?
2.3 ) for details ) .
first suppose that there is a @xmath405 such that @xmath406 .
let @xmath407 be the usual projection map , and extend @xmath408 to @xmath155 coordinatewise .
for every @xmath409 we have that @xmath410 , so that @xmath411 . since @xmath412 can be made arbitrarily small , it follows that @xmath15 is not expansive .
conversely , if @xmath15 is not expansive , there is a point @xmath61 with @xmath413 for all @xmath27 .
pick @xmath414 with @xmath415 for all @xmath27 , and let @xmath416 . then @xmath417 ) , and @xmath418 , hence @xmath419 .
this shows that ( 1 ) @xmath420 ( 4 ) .
now suppose that @xmath65 is injective on @xmath155 . then @xmath421 is dense in @xmath156 by the hahn - banach theorem .
since the set of invertible elements in @xmath156 is open , there is a @xmath422 with @xmath423 . by direct finiteness
, @xmath64 is invertible in @xmath156 with inverse @xmath252 .
hence @xmath424 .
this shows that ( 4 ) @xmath425 ( 3 ) , and the implication ( 3 ) @xmath425 ( 4 ) is obvious . since @xmath13 is invertible if and only if @xmath64 is invertible , ( 2 ) @xmath420 ( 1 ) .
let us call @xmath11 _ expansive _ if @xmath15 is expansive .
for the case @xmath426 , wiener s theorem on invertibility in the convolution algebra @xmath427 shows that @xmath428 $ ] is expansive if and only if @xmath13 does not vanish on @xmath429 . the usual proof of wiener s theorem via banach algebras is nonconstructive since it uses zorn s lemma to create maximal ideals .
paul cohen @xcite has given a constructive treatment of this and similar results .
we are grateful to david boyd for showing us a simple algorithm for deciding expansiveness in this case .
there is a finite algorithm , using only operations in @xmath326 $ ] , that decides , given @xmath428 $ ] , whether or not @xmath13 is expansive .
we may assume that @xmath430 $ ] with @xmath431 , say of degree @xmath2 .
we can compute the greatest common divisor @xmath432 of @xmath433 and @xmath434 using only finitely many operations in @xmath326 $ ] .
observe that any root of @xmath433 on @xmath429 must also be a root of @xmath432 .
let the degree of @xmath24 be @xmath435 .
then @xmath436 , so that the coefficients of @xmath432 are symmetric . if @xmath435 is odd , then @xmath437 is a root of @xmath24 since all other possible roots come in distinct pairs . if @xmath435 is even , it is simple to compute @xmath438 $ ] so that @xmath439 . any root of @xmath24 on @xmath429 corresponds to a root of @xmath96 on @xmath440 $ ]
. we can then apply sturm s algorithm , which uses a finite sequence of calculations in @xmath326 $ ] and sign changes of rationals , to compute the number of roots of @xmath96 in @xmath440 $ ] .
decidability of expansiveness for other groups @xmath6 , even just @xmath9 , is a fascinating open question .
[ prob : expansiveness ] is there a finite algorithm that decides , given @xmath127 , whether or not @xmath13 is expansive ?
there is one type of polynomial in @xmath21 that is easily seen to be expansive . call @xmath11 _
lopsided _ if there is a @xmath441 such that @xmath442 .
this terminology is due to purbhoo @xcite .
if @xmath13 is lopsided with dominant coefficient @xmath443 , adjust @xmath13 by multiplying by @xmath444 so that @xmath445
. then @xmath446 , where @xmath447 .
we can then invert @xmath13 in @xmath156 by geometric series : @xmath448 thus lopsided polynomials are expansive .
the product of lopsided polynomials need not be lopsided ( expand @xmath449 ) .
surprisingly , if @xmath11 is expansive , then there is always a @xmath48 such that @xmath450 is lopsided .
this was first proved by purbhoo @xcite for @xmath16 using a rather complicated induction from the case @xmath426 , but his methods also provided quantitative information he needed to approximate algorithmically the complex amoeba of a laurent polynomial in several variables .
the following short proof is due to hanfeng li .
[ prop : lopsided - is - invertible ] let @xmath6 be a countable discrete group and let @xmath11 be expansive .
then there is @xmath48 such that @xmath450 is lopsided . since @xmath13 is expansive , by theorem [ thm : expansive ]
there is a @xmath422 such that @xmath451 .
there is an obvious extension of the definition of lopsidedness to @xmath156 .
note that lopsidedness is an open condition in @xmath156 .
first perturb @xmath252 slightly to @xmath452 having finite support , and then again slightly to @xmath453 having finite support and rational coordinates .
this results in @xmath454 such that @xmath455 is lopsided .
choose an integer @xmath323 so that @xmath456
. then @xmath450 is lopsided .
one consequence of the previous result is that if @xmath11 is expansive , then the coefficients of @xmath457 must decay exponentially fast . recall that if @xmath6 is finitely generated , then a choice of finite symmetric generating set @xmath196 induces the _ word norm _
@xmath458 , where @xmath459 is the length of the shortest word in generators in @xmath196 whose product is @xmath460 . clearly @xmath461 . a different choice @xmath462 for symmetric generating set gives an equivalent word norm @xmath463 in the sense that there are two constants @xmath464 such that @xmath465 for all @xmath27 .
[ prop : exponential - decay ] let @xmath6 be a finitely generated group , and fix a finite symmetric generating set @xmath196 .
suppose that @xmath11 is invertible in @xmath156 . then there are constants @xmath466 and @xmath467 such that @xmath468 for all @xmath27 . by the previous proposition ,
there is a @xmath48 such that @xmath469 is lopsided .
we may assume that the dominant coefficient of @xmath96 occurs at @xmath470 , so that @xmath471 , where @xmath472 and @xmath473 has @xmath474 .
let @xmath475 and put @xmath476 .
now @xmath477 and @xmath478 for all @xmath333 .
furthermore , if @xmath479 , then @xmath480 . hence if @xmath481 , then @xmath482 .
thus @xmath483 whenever @xmath481 .
this shows that @xmath484 with @xmath485 and suitable @xmath466 .
since @xmath486 , we obtain the result with the same @xmath330 and different @xmath166 .
we remark that a different proof of this proposition , using functional analysis and under stricter hypotheses on @xmath6 , was given in ( * ? ? ?
if @xmath426 and @xmath428 $ ] is invertible in @xmath487 , then @xmath13 does not vanish on @xmath429 .
then @xmath488 is holomorphic in an annular region around @xmath429 in @xmath489 , and so its laurent expansion decays exponentially fast , giving a direct proof of the proposition in this case .
next we focus on the case @xmath8 and obstructions of invertibility coming from representations of @xmath9 .
let @xmath490 be a complex hilbert space , and let @xmath491 be the algebra of bounded linear operators on @xmath490 .
denote by @xmath492 the group of unitary operators on @xmath490 .
an _ irreducible unitary representation _ of @xmath9 is a homomorphism @xmath493 for some complex hilbert space @xmath490 such that there is no nontrivial closed subspace of @xmath490 invariant under all the @xmath494 .
then @xmath343 extends to an algebraic homomorphism @xmath495 by @xmath496 .
if there is a nonzero @xmath497 with @xmath498 , then clearly @xmath13 can not be invertible in @xmath364 .
the converse is also true .
[ thm : representations ] let @xmath127 .
then @xmath13 is not invertible in @xmath364 if and only if there is an irreducible unitary representation @xmath493 on some complex hilbert space @xmath490 and a nonzero @xmath497 such that @xmath498 .
this result is stated in ( * ? ? ?
8.2 ) , and a detailed proof is given in ( * ? ? ?
a key element of the proof is that @xmath499 is _ symmetric _ , i.e. , that @xmath500 is invertible for every @xmath501 , and in particular for every @xmath502 . there are examples of countable amenable groups that are not symmetric @xcite .
we remark that theorem [ thm : representations ] remains valid when @xmath9 is replaced by any nilpotent group , in particular by @xmath55 or @xmath5 . in the latter cases all irreducible unitary representations
are 1-dimensional , and obtained by evaluation at a point in @xmath503 .
this is exactly the wiener criterion for invertibility in @xmath504 .
the usefulness of theorem [ thm : representations ] is at best limited since @xmath9 is not of type i and so its representation theory is murky .
however , there is an extension of gelfand theory , called allan s local principle , that detects invertibility in the noncommutative banach algebra @xmath499 .
the use of this principle for algebraic actions was initiated in @xcite . in the case of @xmath9-actions
this principle has an explicit and easily verified form , which we now describe . to simplify notation ,
let @xmath505 be the complex convolution banach algebra of @xmath9 , and @xmath506 be its center .
the maximal ideals of @xmath166 all have the form @xmath507 , where @xmath508 . for @xmath509 the quotient norm of @xmath510
is easily seen to be @xmath511 .
let @xmath512 denote the two - sided ideal in @xmath206 generated by @xmath513 , so that @xmath514 then for @xmath515 , the quotient norm of @xmath516 is @xmath517 to give a concrete realization of @xmath518 , introduce variables @xmath374 , @xmath519 , subject to the skew commutativity relation @xmath520 .
then the skew convolution banach algebra @xmath521 consists of all sums @xmath522 with @xmath523 and @xmath524 , and with multiplication the usual convolution modified by skew commutativity .
then the map @xmath525 given by @xmath526 is an isometric isomorphism of complex banach algebras .
identifying these algebras , the quotient map @xmath527 takes the concrete form @xmath528 . in ( * ? ? ?
* section 3 ) , _ allan s local principle _ was introduced as a convenient device for checking expansiveness of principal actions of @xmath529 and applied to a number of examples .
[ thm : allan - principle ] an element @xmath530 is invertible if and only if @xmath531 is invertible in @xmath521 for every @xmath508 . using the notations introduced above , clearly if @xmath252 is invertible in @xmath206 , then its homomorphic images @xmath531 must all be invertible .
conversely , suppose @xmath532 is not invertible , but @xmath531 is invertible in @xmath521 for every @xmath533 .
the left ideal @xmath534 is proper since @xmath252 is not invertible , so let @xmath277 be a maximal left ideal in @xmath206 containing @xmath534 .
it is easy to see that @xmath535 must be a maximal ideal in @xmath166 , hence @xmath536 for some @xmath508 . then @xmath537 . by assumption
, @xmath538 is invertible in @xmath518 , and @xmath539 , so that @xmath540 , a contradiction .
we now turn to some concrete examples where expansiveness can be analyzed geometrically . before doing so ,
let us introduce a quantity that we will use extensively .
let @xmath428 $ ] .
the _ logarithmic mahler measure _ of @xmath13 is defined as @xmath541 .
the _ mahler measure _ of @xmath13 is defined as @xmath542 .
suppose that @xmath543 with @xmath544 .
factor @xmath433 over @xmath489 as @xmath545 .
then jensen s formula shows that @xmath546 has the alternative expression as @xmath547 where @xmath548 for @xmath549 .
mahler s motivation was to derive important inequalities in transcendence theory . using that @xmath550
, he showed that if @xmath551 $ ] , then @xmath552 , and that the constant is best possible .
let us consider the case of @xmath127 that is linear in @xmath69 , so that @xmath553 .
we will find 1-dimensional @xmath15-invariant subspaces of @xmath554 as follows .
let @xmath555 be the subgroup of @xmath9 generated by @xmath68 and @xmath70 .
to make calculations in @xmath554 and @xmath556 more transparent , we will write elements as formal sums @xmath557 and @xmath558
. fix @xmath559 and define @xmath560 observe that @xmath561 and similarly @xmath562 . hence the 1-dimensional space @xmath563 is a common eigenspace for @xmath564 and @xmath565 .
it follows that @xmath566 for every @xmath567 .
let @xmath568 be a sequence of complex constants to be determined , and consider the point @xmath569 . using the relations @xmath570 for all @xmath571 and all @xmath567
, the condition @xmath406 becomes @xmath572 this calculation shows that @xmath573 if and only if the @xmath574 satisfy @xmath575 since @xmath576 , one way to create nonexpansive @xmath13 s of this form is to find conditions on @xmath24 and @xmath96 that guarantee the existence of a nonzero bounded solution @xmath568 to for some choice of @xmath577 and @xmath578 .
suppose that both @xmath24 and @xmath96 do not vanish on @xmath579 .
fix a nonzero value of @xmath580 .
then by , the other values of @xmath574 are determined : @xmath581^{-1 } & \text{for $ n{\leqslant}-1$}. \end{cases}\ ] ] let @xmath582 and consider the map @xmath583
given by @xmath584 then @xmath585 satisfies the cocycle equation @xmath586 for all @xmath587 and @xmath588 .
furthermore , there is a nonzero bounded solution @xmath568 to if and only if @xmath589 is bounded above .
suppose that @xmath508 is irrational , and that there is a @xmath588 for which has a nonzero bounded solution .
observe that @xmath590 is continuous on @xmath579 since neither @xmath24 nor @xmath96 vanish there by assumption . by the ergodic theorem , @xmath591 thus @xmath592 .
there is a similar necessary condition when @xmath578 is rational , but here the integral is replaced by a finite sum over a coset of the finite orbit of @xmath578 in @xmath429 .
it turns out that these necessary conditions are also sufficient .
[ thm : linear - expansive ] let @xmath593 , and suppose that both @xmath24 and @xmath96 do not vanish anywhere on @xmath579 .
let @xmath594 .
then @xmath15 is expansive if and only if 1 .
@xmath595 for every irrational @xmath508 , and 2 . for every @xmath323-th root of unity @xmath508 and @xmath588 we have that @xmath596 .
suppose first that @xmath15 is nonexpansive . by theorem [ thm : expansive ]
, the @xmath9-invariant subspace @xmath597 is nontrivial . by restricting the @xmath9-action on @xmath107 to the commutative subgroup @xmath164
, we can apply the argument in ( * ? ? ?
* lemma 6.8 ) to find a 1-dimensional @xmath164-invariant subspace @xmath598 .
in other words , there are a nonzero @xmath599 and @xmath600 such that @xmath601 and @xmath602 .
it follows from this and the above discussion that @xmath252 must have the form @xmath603 with the @xmath604 given by , and with @xmath568 a bounded sequence .
if @xmath578 is an @xmath323-th root of unity , then clearly @xmath605 by boundedness of the @xmath574 , while if @xmath578 is irrational , then the discussion above shows that @xmath606 .
for the converse , suppose first that @xmath578 is rational , say @xmath607 .
if there is a @xmath588 with @xmath588 with @xmath608 , then and show that there is a bounded sequence @xmath568 , in this case even periodic with period @xmath263 , so that @xmath609 with @xmath573 , showing that @xmath15 is nonexpansive
. finally , suppose that @xmath578 is irrational and @xmath606 .
if there were a continuous coboundary @xmath205 on @xmath429 such that @xmath610 , then the cocycle @xmath611 would be bounded for every value of @xmath577 , and as before this means we can form an nonzero point @xmath612 with @xmath573 , so @xmath15 is nonexpansive .
so suppose no such coboundary exists .
let @xmath613 , and define a skew product transformation @xmath614 by @xmath615 . since @xmath590 is not a coboundary , but @xmath606 ,
the homeomorphism @xmath196 is topologically transitive ( see @xcite or ( * ? ? ?
* thms . 1 and 2 ) ) . by @xcite
, there exists a point @xmath588 such that the entire @xmath196 orbit of @xmath616 has its second coordinate bounded above . since @xmath617
, it follows that for this choice of @xmath577 , the sequence @xmath568 from and is also bounded , and therefore @xmath618 with @xmath573 , and
so @xmath15 is nonexpansive . with the assumptions of theorem [ thm : linear - expansive ] that neither @xmath24 nor @xmath96 vanish on @xmath579 , the functions @xmath619 and @xmath620 are both continuous functions of @xmath578 .
hence the conditions ( 1 ) and ( 2 ) in theorem [ thm : linear - expansive ] combine to say that the graphs of these functions never cross ( for rational @xmath578 use the mean value theorem ) .
in particular , if @xmath621 , then since @xmath622 , the condition for expansiveness of @xmath623 becomes simply that @xmath624 for all @xmath578 .
the polynomial @xmath625 , although not lopsided , was shown to be expansive in ( * ? ? ?
* example 7.4 ) . in (
* example 3.6 ) four different ways to verify its expansiveness are given : ( 1 ) using irreducible unitary representations , ( 2 ) direct computation of its inverse in @xmath364 , ( 3 ) using allan s local principle , and ( 4 ) using the geometric argument in theorem [ thm : linear - expansive ] .
many more examples illustrating various aspects of expansiveness ( or its lack ) for polynomials in @xmath10 are contained in @xcite .
[ exam : x+y+z+2 ] let @xmath626 , where @xmath627 . since @xmath628 , it does not quite satisfy the hypothesis of theorem [ thm : linear - expansive ] .
however , we can directly use the violation of ( 2 ) there to show that @xmath15 is nonexpansive .
let @xmath629 .
we want to find @xmath588 such that @xmath630 .
this amounts to solving @xmath631 , and this has the four solutions @xmath632 .
let @xmath633 , and consider the point @xmath634 as defined in .
then the coefficients @xmath574 of the point @xmath635 satisfy , and are hence alternately multiplied by @xmath636 and by @xmath637 , where @xmath638 and @xmath639 . thus @xmath640 is bounded , and hence @xmath15 is nonexpansive . note that here @xmath641 , which vanishes only at @xmath642 .
hence for all irrational @xmath578 , condition ( 1 ) of theorem [ thm : linear - expansive ] is satisfied .
it is an easy exercise to show that if @xmath643 is rational , then condition ( 2 ) is also satisfied .
so here the only values of @xmath644 leading to a bounded solution @xmath568 of are @xmath645
. this example appears as ( * ? ? ?
* example 10.6 ) , but was claimed there to be expansive .
[ exam : y^2-xy-1 ] let @xmath646 . for @xmath559
let @xmath647 be defined by , and let @xmath648 .
although @xmath13 is now quadratic in @xmath69 , we can still calculate @xmath649 as before , finding that @xmath573 if and only if @xmath650 thus we need conditions on @xmath644 for which has a bounded solution . for @xmath309 put @xmath651 then the recurrence shows that @xmath652 and there is a similar formula for @xmath653 .
we are therefore reduced to finding @xmath644 such that the matrix - valued cocycle @xmath654 is bounded .
the easiest place to look is at @xmath655 , since @xmath656 .
now @xmath657 has eigenvalues on @xmath429 provided that the roots of @xmath658 are there .
this happens exactly when @xmath659 . hence if we define @xmath574 by @xmath660 then @xmath661 , and so @xmath15 is nonexpansive .
for some values of @xmath644 the growth rate of @xmath662 can be strictly positive , for example @xmath663 .
however , as we will see later , the growth rate of @xmath662 as @xmath664 is zero for almost every @xmath559 , although this is not at all obvious .
if @xmath24 or @xmath96 are allowed to vanish on @xmath579 , there is a completely different source of nonexpansive behavior .
let @xmath593 , and suppose that @xmath665 and @xmath666 have a common zero @xmath559 .
consider @xmath647 as a point in @xmath554 .
then @xmath667 so that @xmath65 is not expansive .
there is a simple geometric method to create examples of this situation .
start with two polynomials @xmath665 and @xmath668 whose unitary varieties in @xmath579 do not intersect , but which have the property that the unitary variety of the skewed polynomial @xmath669 intersects that of @xmath24 for sufficiently large @xmath301 .
an example of this form are described in ( * ? ? ?
* example 8.5 ) , which also contains other results about expansiveness in cases when at least one unitary variety is nonempty .
however , this analysis leaves open one very interesting case .
the simplest version of this question is the following .
let @xmath670 $ ] , and suppose there is a @xmath508 for which @xmath671 and such that @xmath672 vanishes for at least one value of @xmath588 .
is there a value of @xmath577 for which the partial sums @xmath673 are bounded above for all @xmath323 ?
fraczek and lemaczyk showed in @xcite that these sums are unbounded for almost every @xmath588 .
the argument of besicovitch @xcite to prove the existence of bounded sums makes essential use of continuity , and does not apply in this case where there are logarithmic singularities .
[ rem : algebraic ] by invoking some deep results in diophantine approximation theory , we can show that the second alternative in the last paragraph of the proof of theorem [ thm : linear - expansive ] never occurs .
for we claim that if @xmath606 , then @xmath578 must be an algebraic number in @xmath429 . assuming this for the moment
, then a result of gelfond @xcite shows that for every @xmath409 there is a @xmath466 such that @xmath674 for every @xmath309 .
since @xmath675 is smooth , its fourier coefficients @xmath676 decay rapidly as @xmath677 , and so the formal solution @xmath678 to @xmath610 , with @xmath679 for @xmath680 , also decays rapidly , giving a continuous solution @xmath205 .
to justify our claim , write @xmath681 where @xmath682 and ) @xmath683 are laurent polynomials with integer coefficients and the @xmath684 and @xmath685 are algebraic functions .
the condition @xmath606 becomes , via jensen s formula , @xmath686 which is an algebraic equation in @xmath578 , so @xmath578 is algebraic . from this , and the preceding proof , under our assumptions that neither @xmath24 nor @xmath96 vanish anywhere on @xmath579 ,
we conclude that if @xmath606 and @xmath578 is irrational , then _ every _ choice of @xmath577 will yield a nonzero point @xmath687 with @xmath573 .
this idea was observed independently by evgeny verbitskiy ( oral communication ) . to illustrate some of the preceding ideas , we provide an informative example .
this was chosen so that the diophantine estimates mentioned in remark [ rem : algebraic ] can be given an elementary and self - contained proof , rather than appealing to difficult general diophantine results .
in addition , the constants in our analysis are effective enough to rule out nonexpansive behavior at all rational @xmath578 .
one consequence is that for this algebraic @xmath9-action , nonexpansiveness can not be detected by looking at only finite - dimensional representations of @xmath9 .
[ exam : nonexpansive ] let @xmath688 , where @xmath689 and @xmath690 .
it is easy to check that neither @xmath204 nor @xmath228 vanishes on @xmath429 , so that @xmath24 does not vanish anywhere on @xmath579 . also , @xmath691 has roots @xmath692 and @xmath693 .
consider @xmath623 , so that here @xmath694
. then @xmath695=\log|a(\xi)|+\log|c(\zeta)|$ ] , and define @xmath696 as in . call @xmath559 _ nonexpansive for _ @xmath24 if the sequence @xmath697 is bounded .
we claim there are eight values @xmath698 , which are algebraically conjugate algebraic integers of degree 48 , such that the nonexpansive points for @xmath24 are exactly those of the form @xmath699 , where @xmath577 is any element of @xmath429 and @xmath700 .
we start with two simple results .
[ lem : coboundary ] suppose that @xmath508 is irrational , and that there are constants @xmath466 and @xmath467 such that @xmath701 for all @xmath309 . if @xmath702 with @xmath703 , then the function @xmath704 on @xmath429 is a continuous coboundary for @xmath578 , i.e. , there is a continuous @xmath705 such that @xmath706 .
since @xmath707 , the taylor series for @xmath708 converges , so that @xmath709 let @xmath710 . then @xmath711 provided that @xmath712 $ ] for all @xmath309 .
the assumption on @xmath578 implies that @xmath713 , so that the series for @xmath206 converges uniformly on @xmath429 .
then @xmath714 gives the required coboundary .
[ lem : diophantine ] let @xmath715 $ ] be monic and irreducible .
suppose that @xmath716 has a root @xmath508 that is irrational .
then there is a constant @xmath466 such that @xmath717 where @xmath718 is the mahler measure of @xmath716 .
factor @xmath716 over @xmath489 as @xmath719 , where @xmath720 is the complex conjugate of @xmath578 .
the polynomial @xmath721 has integer coefficients , and @xmath722 , hence @xmath723 . then using the trivial estimates that @xmath724 if @xmath725 and @xmath726 if @xmath727 , we obtain that @xmath728 hence is valid with @xmath729 recall that @xmath730 , so that @xmath731 . by theorem [ thm : linear - expansive ]
, we need to find just those @xmath508 for which @xmath732 that is , we must solve the equation @xmath733 .
any such solution @xmath508 would also satisfy @xmath734 , or , equivalently , satisfy @xmath735.\ ] ] here @xmath736 has degree 24 , and has eight roots @xmath698 . to show these are algebraic integers , multiply @xmath736 by its galois conjugate over @xmath737 , yielding the following polynomial @xmath738
whose irreducibility is confirmed by _
mathematica_. hence the @xmath739 are conjugate algebraic integers of degree 48 as claimed .
now @xmath740 has 10 roots outside the unit circle , whose product is @xmath741 .
then@xmath742 , which plays a crucial role in dealing with rational @xmath578 . by theorem [ thm : expansive ] , the only irrational @xmath578 we need to consider are the eight @xmath739 . by lemma [ lem : diophantine ]
, there is a constant @xmath466 such that @xmath743 , where @xmath744 .
since @xmath745 , by lemma [ lem : coboundary ] , we can find continuous coboundaries @xmath746 and @xmath747 so that @xmath748 hence @xmath749 is also a coboundary
. thus @xmath699 is nonexpansive for every @xmath588 .
this is an example of the algebraic phenomenon discussed in remark [ rem : algebraic ] . to complete our analysis
, we turn to the rational case , say @xmath750 , a primitive @xmath323-th root of unity .
the idea of the following argument is that is to show that there is a large enough @xmath751 such that for all @xmath752 , the variation of the @xmath323-periodic function @xmath753 is small compared to @xmath754 .
the estimates are sharp enough to obtain the bound @xmath755 , and the remaining cases @xmath756 can be checked by hand .
first observe that @xmath757 since @xmath758 for all @xmath702 with @xmath759 , it follows that @xmath760 for all @xmath588 and all @xmath309 .
recalling that @xmath761 , we obtain that @xmath762^n|.\ ] ] thus , if @xmath763 then we must have @xmath764 for all @xmath588 . to obtain a reasonable bound for @xmath751 so that holds for all @xmath752
, we need to make a more careful estimate in the proof of lemma [ lem : diophantine ] for the polynomial @xmath765 . here
@xmath765 has ten roots @xmath766 outside the unit circle .
thus besides these and @xmath578 and @xmath720 , there are 36 roots of @xmath740 on or inside the unit circle .
the estimate in the proof of lemma [ lem : diophantine ] can be refined as @xmath767 it is easy to check that @xmath768 has a maximum of about 37.94 at @xmath769 .
hence @xmath770 since @xmath771 , it follows that @xmath772 . verifying breaks into two cases , depending on whether or not @xmath773 is close to some @xmath739 .
let @xmath774 .
it is an exercise in calculus to show that if @xmath775 for every @xmath263 , then @xmath776 , while if @xmath777 for some @xmath263 , then the derivative of @xmath778 has absolute value @xmath779 .
a glance at figure [ fig : graph ] should make clear the meaning of these statements .
[ fig : graph ] ] in the first case , the inequality is satisfied if @xmath780 , which is true for all @xmath781 . in the second case ,
using the lower bound on the absolute value of the derivative , @xmath782 since @xmath783 , the last term is eventually greater than @xmath784 , and in fact this holds for all @xmath785 .
one can then check by hand that that holds for all @xmath756 , completing our analysis of this example .
it is sometimes useful to make a change of variables to transform a polynomial into a form that is easier to analyze ( see for example ( * ? ? ?
* ( 36 ) ) ) .
let @xmath6 be a countable discrete group , and @xmath786 denote the group of automorphisms of @xmath6 .
if @xmath787 , then @xmath788 will act on various objects associated with @xmath6 .
for example , if @xmath789 , then @xmath790 , and so @xmath791 and @xmath792 .
analogous formulae hold for @xmath793 , @xmath156 , and @xmath155 .
if @xmath7 is an action of @xmath6 , there is a new @xmath6-action @xmath794 defined by @xmath795 .
[ lem : change - of - variables ] let @xmath787 and @xmath11 .
then @xmath788 induces a continuous group isomorphism @xmath796 intertwining the @xmath6-actions @xmath797 and @xmath798 , so that @xmath799 and @xmath800 are topologically conjugate @xmath6-actions . if @xmath801 , then @xmath61 iff @xmath62 iff @xmath802 iff @xmath803 , so that @xmath788 induces the required isomorphism . since @xmath804
, we see that @xmath788 intertwines @xmath797 and @xmath798 .
[ rem : automorphism - effects ] it is important to emphasize what lemma [ lem : change - of - variables ] does _ not _ say .
although @xmath798 and @xmath805 are both @xmath6-actions on @xmath806 , there is no obvious relation between them , even if @xmath6 is commutative .
for example , if @xmath807 , @xmath808 , and @xmath809 , then @xmath810 is conjugate to the @xmath55-action of @xmath811 on @xmath812 , while @xmath800 is conjugate to the @xmath55 action of @xmath813 on @xmath812 . but
@xmath190 and @xmath813 do not even have the same eigenvalues , so can not give algebraically conjugate @xmath55-actions . however , certain dynamical properties are clearly shared between @xmath345 and @xmath805 , for example ergodicity , mixing , and expansiveness .
automorphisms of @xmath9 have an explicit description .
[ lem : automorphism - heisenberg ] every automorphism of @xmath9 is uniquely determined by integers @xmath204 , @xmath205 , @xmath228 , @xmath2 , @xmath330 , and @xmath814 with @xmath815 , and is given by @xmath816 , @xmath817 , and @xmath818 .
clearly @xmath788 induces an automorphism of @xmath819 , hence @xmath815 .
the condition @xmath820 shows that @xmath818 .
these necessary conditions are easily checked to also be sufficient .
an analysis of expansiveness for nonprincipal actions has been carried out by chung and li @xcite .
let @xmath821 be a square matrix over @xmath10 .
then @xmath822 is a left @xmath10-module , whose dual gives an algebraic action @xmath362 on @xmath823 .
an argument similar to the proof of ( 3 ) implies ( 1 ) in theorem [ thm : expansive ] shows that if @xmath274 has an inverse in @xmath824 , then @xmath362 is expansive on @xmath823 .
the converse is also true , and leads to a description of all expansive @xmath9-actions .
let @xmath821 and @xmath362 be the associated algebraic @xmath9-action on @xmath823 . then @xmath362 is expansive if and only if @xmath274 is invertible in @xmath825 .
moreover , every expansive @xmath9-action is isomorphic to the restriction of such an @xmath362 to a closed @xmath362-invariant subgroup of @xmath823 .
let @xmath826 . by simply formally solving the equations for the inverse ,
one arrives at @xmath827 where the inverses appearing in the entries are all in @xmath364 by lopsidedness .
obviously an algorithm for invertibility of square matrices would immediately answer problem [ prob : expansiveness ] .
but if the answer to the latter is affirmative , would this provide an algorithm to decide invertibility of square matrices ?
suppose there is an algorithm which decides whether or not an element in @xmath10 has an inverse in @xmath364 .
is there then an algorithm that decides , given @xmath828 , whether or not @xmath274 has a inverse with entries in @xmath364 ? if @xmath7 is an expansive algebraic @xmath6-action on @xmath26 , and @xmath829 is a closed @xmath7-invariant subgroup , then clearly the restriction of @xmath7 to @xmath829 is also expansive .
however , whether the quotient action @xmath830 on @xmath831 is expansive is much more difficult .
when @xmath16 , expansiveness of the quotient is always true ( ( * ? ?
* thm.3.11 ) ) , but the proof uses commutative algebra and is not dynamical . chung and li conjecture @xcite that for nilpotent groups @xmath6 , quotients of expansive actions are always expansive . even for the heisenberg group
this is not known .
if @xmath7 is an expansive action of @xmath9 on a compact abelian group @xmath26 , and if @xmath829 is a closed @xmath7-invariant subgroup of @xmath26 , then must the quotient action of @xmath7 on @xmath831 be expansive ?
let @xmath6 be a countable discrete group and @xmath7 an algebraic @xmath6-action on a compact abelian group @xmath26 .
an element @xmath28 is called _
homoclinic for @xmath7 _ , or simply _ homoclinic _ , if @xmath832 as @xmath300 .
the set of homoclinic points forms a subgroup of @xmath26 called the _ homoclinic group of @xmath7_. the established notation in the literature for this group is @xmath833
. it will always be clear from context ( and a slight font change ) what @xmath6 ( or @xmath834 ) refers to .
homoclinic points are an important technical device for localizing the behavior of points in the group .
for example , they are used to construct periodic points , to prove a strong orbit tracing property called specification , and to estimate entropy .
they are also a natural starting point for constructing symbolic covers of algebraic actions . for @xmath16
, many properties of homoclinic groups were studied in detail in @xcite , especially for principal actions .
let us briefly describe some of the main results there , with a view to extensions to @xmath9
. for @xmath835 $ ] , define the _ complex variety _ of @xmath13 to be @xmath836 where @xmath837 , and the _ unitary variety _ of @xmath13 to be @xmath838 then by theorem [ thm : expansive ] and wiener s theorem , @xmath15 is expansive if and only if @xmath839 . in this case , let @xmath840 . as
before , let @xmath841 be given by @xmath842 , which clearly commutes with the left @xmath6-actions .
put @xmath843 . since @xmath844 , we see that @xmath845 , and is also homoclinic .
furthermore , @xmath846 is _ fundamental _ , in the sense that every homoclinic point is a finite integral combination of translates of @xmath846 ( cf .
* lemma 4.5 ) ) . in this case
all homoclinic points decay rapidly enough to have summable coordinates . in order to describe homoclinic points of principal @xmath6-actions
@xmath15 , we first `` linearize '' @xmath57 as follows .
put @xmath847 suppose now that @xmath11 is expansive , and define @xmath848 .
then @xmath65 is invertible on @xmath155 , and @xmath849 , where @xmath850 for every @xmath851 .
[ prop : cover ] let @xmath6 be a countable discrete group , and @xmath11 be expansive , so that @xmath13 is invertible in @xmath156 . put @xmath852 and let @xmath853 be defined as @xmath854 , where @xmath408 is reduction of coordinates @xmath855 .
then : 1 .
@xmath856 is surjective , and in fact the restriction of @xmath343 to the set of those @xmath857 with @xmath858 is also surjective ; 2 .
@xmath859 ; 3 .
@xmath343 commutes with the relevant left @xmath6-actions ; and 4 .
@xmath343 is continuous in the weak * topology on closed , bounded subsets of @xmath860 .
suppose that @xmath61 .
there is a unique lift @xmath861 with @xmath862 and @xmath863 for all @xmath27 .
then @xmath864 in @xmath57 , hence @xmath865 , and in fact @xmath866 .
furthermore @xmath867 this proves ( 1 ) , and the remaining parts are routine verifications .
if @xmath11 is expansive , let @xmath843 and call @xmath846 the _ fundamental homoclinic point _ of @xmath15 .
this name is justified by the following .
[ prop : fundamental - homoclinic - point ] let @xmath6 be a countable discrete group , and @xmath11 be expansive
. put @xmath848 and @xmath868 .
then every element of @xmath869 is a finite integral combination of left translates of @xmath846 .
suppose that @xmath870 , and lift @xmath66 to @xmath871 as in the proof of proposition [ prop : cover ] .
then @xmath865 , and since @xmath872 as @xmath300 , the coordinates of @xmath873 must vanish outside of a finite subset of @xmath6 , i.e. , @xmath874 . then @xmath875 has the required form .
next we show that expansive principal actions have a very useful orbit tracing property called specification .
[ prop : specification ] let @xmath6 be a countable discrete group , and @xmath11 be expansive .
then for every @xmath409 there is a finite subset @xmath876 of @xmath6 such that if @xmath877 and @xmath878 are arbitrary subsets of @xmath6 with @xmath879 and if @xmath880 and @xmath881 are arbitrary points in @xmath57 , then we can find @xmath61 such that @xmath882 for every @xmath883 , for @xmath884 .
let @xmath409 .
the set @xmath876 is chosen so that @xmath885 .
lift each @xmath886 to @xmath887 , and then truncate each @xmath888 to a @xmath889 having support in @xmath890 .
it is then easy to verify that @xmath891 as the required properties ( see ( * ? ? ?
4.4 ) for details ) .
a point @xmath61 with @xmath892 is called _
summable_. let @xmath893 denote the group of all summable homoclinic points for @xmath15 .
summability is crucial in using homoclinic points for dynamical purposes .
let @xmath894 $ ] .
it is shown in ( * ? ? ?
* example 7.3 ) that @xmath869 is uncountable ( indeed , the fourier series of every smooth density on @xmath895 decays to 0 at infinity , and so gives a homoclinic point ) , but @xmath896 . despite their large number , the nonsummable homoclinic points here are essentially useless .
summable homoclinic points may still exist for nonexpansive actions .
for example , consider @xmath897 .
the formal inverse @xmath252 of @xmath64 via geometric series is well - defined and has coordinates decaying to 0 at infinity , so that @xmath898 is homoclinic , but the decay is so slow that @xmath252 is not summable ( see ( * ? ? ?
* example 7.2 ) ) . define a function @xmath899 by putting @xmath900 .
then @xmath901 is integrable on @xmath812 , and @xmath252 is just the fourier transform of @xmath901 .
now @xmath901 has a singularity at @xmath902 , and we can try to cancel this by multiplying it by a sufficiently high power @xmath281 of another polynomial @xmath903 that also vanishes at @xmath902 so that @xmath904 has absolutely convergent fourier series , resulting in a summable homoclinic point @xmath905 . for this example
, taking @xmath906 , a detailed analysis in @xcite shows that @xmath907 is the smallest power such that @xmath904 has absolutely convergent fourier series , providing a summable homoclinic point for @xmath15 .
this `` multiplier method '' can be generalized to all @xmath908 $ ] provided that the dimension of @xmath909 is at most @xmath910 .
more precisely , with this condition , there is another polynomial @xmath911 $ ] , not a multiple of @xmath13 , such that @xmath912 .
the corresponding quotient @xmath904 has absolutely convergent fourier series for sufficiently large @xmath281 @xcite , and hence @xmath15 has summable homoclinic points .
however , if @xmath913 , this method fails , and in fact there are no nonzero summable homoclinic points ( * ? ? ?
let us turn to considering homoclinic points for principal actions of @xmath9 .
if @xmath127 is expansive , we have already seen how to describe @xmath869 , and that this agrees with @xmath893 .
consider @xmath914 .
if @xmath915 , then @xmath13 is in the kernel of the algebra homomorphism @xmath916 given by @xmath917 , @xmath918 , and @xmath341 .
hence @xmath13 is not expansive .
however , it is shown in @xcite that the formal inverse of @xmath13 can be smoothed by using the multiplier @xmath919 to create a summable homoclinic point for @xmath15 .
the proof uses highly nontrivial combinatorial arguments , starting with the noncommutative expansion @xmath920 if @xmath921 the authors of @xcite state that they shown that the multiplier @xmath922 results in a summable homoclinic point , and conjecture that @xmath919 actually suffices . for more complicated nonexpansive polynomials in @xmath10 , it is not at all clear what could substitute for the dimension condition on the unitary variety in the commutative case . for @xmath127 ,
determine explicitly both @xmath869 and @xmath893 .
anticipating the entropy material from the next section , we remark that chung and li ( * ? ? ?
1.1 ) , generalizing earlier work for @xmath5 in @xcite , showed that @xmath923 if and only if @xmath15 has positive entropy , and that @xmath869 is dense in @xmath57 if and only if @xmath15 has completely positive entropy . for expansive @xmath6-actions @xmath15 , proposition [ prop : cover](1 )
gives a continuous , equivariant , and surjective map @xmath343 from the full @xmath6-shift with symbols @xmath924 to @xmath57 , which allows us to view this shift space as a symbolic cover of @xmath57 . in 1992
anatoly vershik showed @xcite that for certain hyperbolic toral automorphisms , this symbolic cover could be pruned to a shift of finite type for which the covering map @xmath343 is one - to - one almost everywhere .
this provided an arithmetic approach to constructing markov partitions , which were originally found geometrically by adler and weiss , and are one of the main motivations for symbolic dynamics .
vershik s arithmetic construction was further investigated in @xcite , @xcite , @xcite , and @xcite .
einsiedler and the second author @xcite considered the problem of extending this idea to obtain symbolic representations of algebraic @xmath5-actions , and gave an example of an algebraic @xmath1-action for which the symbolic cover could be pruned to a shift of finite type to obtain a map that is one - to - one almost everywhere , but the proof involved a complicated percolation argument . even for heisenberg actions virtually nothing is known about the existence of good symbolic covers .
find general sufficient conditions on an expansive @xmath127 so that the symbolic cover from proposition [ prop : cover ] ( 1 ) can be pruned to one that is ( a ) of finite type or at least sofic , ( b ) of equal entropy , or ( c ) one - to - one almost everywhere .
we give here several equivalent definitions of the ( topological ) entropy of an algebraic action , sketch some background material on von neumann algebras , and then describe recent results relating entropy to fuglede - kadison determinants .
let @xmath6 be a countable discrete group . for finite subsets @xmath925 , define @xmath926 .
for what follows to make sense , we require that @xmath6 be _ amenable _ , namely that there is a sequence @xmath927 of finite subsets of @xmath6 such that for every finite subset @xmath107 of @xmath6 we have that @xmath928 such a sequence is called a _
right - flner sequence_. suppose that @xmath7 is an algebraic @xmath6-action on a compact abelian group @xmath26 .
we assume there is a translation - invariant metric @xmath2 on @xmath26 , and let @xmath117 denote normalized haar measure on @xmath26 . as before ,
we abbreviate the ( left ) @xmath7-action of @xmath6 on @xmath26 by using @xmath29 for @xmath929 .
then @xmath6 acts on subsets @xmath123 by @xmath930 .
although this differs from the traditional action of transformations on subsets using inverse images , this seems better suited to our purposes , since all @xmath931 are invertible , and its use is consistent with the action of @xmath6 on functions on @xmath26 : if @xmath932 is the indicator function of @xmath125 , then @xmath933 . to define topological entropy
, we consider open covers @xmath934 of @xmath26 .
if @xmath935 are open covers , define their _ span _ as @xmath936 . if @xmath934 is an open cover and @xmath274 is a finite subset of @xmath6 , let @xmath937 , where @xmath938 .
for an open cover @xmath934 let @xmath939 denote the cardinality of the open subcover with fewest elements , which is finite by compactness .
it is easy to check that @xmath940 .
we define the _ open cover entropy _ of @xmath7 to be @xmath941 where the supremum is taken over all open covers of @xmath26 .
recall the elementary fact that if @xmath942 is a sequence of nonnegative reals with @xmath943 , then @xmath944 converges to a limit as @xmath945 , and this limit equals @xmath946 .
hence for @xmath426 and @xmath947 , it follows that the @xmath948 in is actually a limit .
there is a general version of this argument valid for amenable groups , due to lindenstrauss and weiss .
[ prop : lindenstrauss - weiss ] suppose that @xmath949 is a real - valued function defined for all nonempty finite subsets @xmath274 of @xmath6 , satisfying : 1 .
@xmath950 , 2 . if @xmath951 , then @xmath952 , 3 .
@xmath953 for all @xmath954 , and 4 .
@xmath955 if @xmath956 .
then for every right - flner sequence @xmath187 the numbers @xmath957 converge to a finite limit , and this limit is independent of the choice of right - flner sequence . roughly speaking , this fact is proved by showing that if @xmath107 is a large finite subset of @xmath6 and @xmath958 is small , then any @xmath274 with @xmath959 can be almost exactly tiled by left translates of @xmath172s of various sizes
. then subadditivity and translation - invariance of @xmath119 show that @xmath960 is bounded above , within a small error , by @xmath961 .
fix an open cover @xmath934 of @xmath26 , and put @xmath962 for every nonempty finite subset @xmath274 of @xmath6 . since each @xmath963 is a homeomorphism of @xmath26
, it follows that @xmath964 conditions ( 1 ) , ( 2 ) , and ( 4 ) in proposition [ prop : lindenstrauss - weiss ] are trivially satisfied for this @xmath119 . hence for every open cover @xmath934 , the @xmath948 in is a limit , and this limit does not depend on the choice of right flner sequence @xmath187 .
the open cover definition of topological entropy is due to adler , konheim , and mcandrew @xcite .
bowen @xcite introduced equivalent definitions that are better suited for many purposes , which we now describe .
if @xmath274 is a finite subset of @xmath6 and @xmath409 , a subset @xmath965 is called _ @xmath966-spanning _ if for every @xmath35 there is a @xmath967 such that @xmath968 for every @xmath969 .
dually , a set @xmath123 is called _ @xmath970-separated _ if for distinct elements @xmath971 there is a @xmath969 such that @xmath972 .
let @xmath973 denote the smallest cardinality of any @xmath966-spanning set , and @xmath974 be largest cardinality of any @xmath966-separated set . put @xmath975
if @xmath976 denotes the open cover of @xmath26 by @xmath958-balls , the elementary inequalities @xmath977 show that @xmath978 , and so all three are independent of choice of right flner sequence .
one more variant of the entropy definition , using volume decrease , is also useful .
let @xmath979 , and for finite @xmath980 put @xmath981 .
define @xmath982 if @xmath125 is @xmath966-spanning , then @xmath983 , so that @xmath984 , and hence @xmath985 . if @xmath125 is @xmath966-separated , then the sets @xmath986 are disjoint , so that @xmath987 , proving that @xmath988
. thus all these notions of entropy coincide , and we let @xmath989 denote their common value .
we remark that these are also equal to the measure theoretic entropy of @xmath7 with respect to haar measure , but we will not be using this fact .
paper @xcite has complete proofs of these facts , and also that the @xmath948 s in these definitions can be replaced by @xmath990 s without affecting the results .
if @xmath787 and @xmath187 is a right flner sequence , then clearly so is @xmath991 .
it follows that @xmath992 , i.e. , entropy is invariant under a change of variables .
suppose that @xmath7 is an algebraic @xmath6-action on @xmath26 , and that @xmath829 is a closed @xmath6-invariant subgroup of @xmath26 .
let @xmath993 denote the restriction of @xmath7 to @xmath829 , and @xmath830 be the quotient action on @xmath831 .
an important property of entropy is that it adds over the exact sequence @xmath994 .
let @xmath6 be an amenable group , let @xmath7 be an algebraic @xmath6-action on @xmath26 , and suppose that @xmath829 is a closed , @xmath6-invariant subgroup of @xmath26 .
then @xmath995 .
the addition formula has a long history .
the basic approach is to take a borel cross - section to the quotient map @xmath996 , and regard @xmath7 as a skew product with base action @xmath830 and fiber actions that are affine maps of @xmath829 with the same automorphism part @xmath993 but with different translations .
the idea is then to show that the translation parts of these affine maps , being isometries , do not affect entropy .
the case @xmath426 was proved by bowen @xcite , and the case @xmath16 is handled in ( * ? ? ?
* appendix b ) using arguments due originally to thomas @xcite .
fiber entropy for amenable actions was dealt with in @xcite .
there is a serious difficulty in generalizing these ideas to noncommutative @xmath6 , namely the lack of a scaling argument used to eliminate a universal constant due to overlaps of open sets in a cover .
however , machinery developed by ollagnier @xcite handles this issue , and this was used by li to give the most general result cited above . when @xmath16 there are explicit formulas for entropy .
first consider the case @xmath426 . without loss of generality
, we can assume that @xmath428 $ ] has the form @xmath997 with @xmath544 .
factor @xmath433 over @xmath489 as @xmath998 .
then yuzvinskii @xcite showed that @xmath999 an interpretation of from @xcite shows that term @xmath1000 is due to geometric expansion , while the term @xmath1001 is due to @xmath716-adic expansions for those primes @xmath716 dividing @xmath574 , an adelic viewpoint that has been useful in other contexts as well .
mahler measure is defined for polynomials @xmath1002=r_d$ ] by the formulas @xmath1003 and @xmath1004 @xcite .
one of the main results in @xcite is that for nonzero @xmath1005 we have that @xmath1006 .
with this information , entropy for arbitrary algebraic @xmath5-actions can be easily found .
let @xmath1007 be an ideal in @xmath1008 that is not principal .
a simple argument ( * ? ? ?
* thm.4.2 ) shows that @xmath1009 .
any finitely generated @xmath1008-module @xmath30 has a prime filtration @xmath1010 with @xmath1011 , where the @xmath1012 are prime ideals in @xmath1008 .
then the addition formula shows that @xmath1013 , and each summand can be computed from our preceding remarks . to conclude this discussion of the @xmath16 case , we point out that there is a complete characterization for which principal actions have zero entropy .
recall the definition of generalized cyclotomic polynomial from [ sec : mixing ] .
[ prop : zero - entropy ] let @xmath1002 $ ] . then @xmath1014 if and only if @xmath13 is a product of generalized cyclotomic polynomials times a monomial or its negative .
this result was originally proved by boyd @xcite using deep results of schinzel , but was later given a simpler and more geometric proof by smyth @xcite .
turning now to noncommutative @xmath6 , we sketch some background material on related von neumann algebras .
the functional analysis used can be found , for example , in ( * ? ? ?
* chaps .
vii , viii ) .
let @xmath1015 which is a complex hilbert space with the standard inner product @xmath1016=\sum_{{\delta}}w_{\delta}\overline{v_{\delta}}$ ] . as in the case of @xmath1017
, there is a left - action of @xmath6 on @xmath1018 given by @xmath1019 .
the _ group von neumann algebra _ @xmath1020 of @xmath6 consists of all those bounded linear operators @xmath1021 commuting with the left @xmath6-action , i.e. , @xmath1022 .
there is a natural inclusion of @xmath1023 into @xmath1020 given by @xmath1024 , where as before @xmath1025 .
in addition , there is a faithful normalized trace function @xmath1026 given by @xmath1027 .
this means that @xmath1028 is linear , @xmath1029 , @xmath1030 for every @xmath1031 , and @xmath1032 . using this trace , fuglede and
kadison @xcite defined a determinant function on @xmath1020 as follows .
let @xmath1033 .
then @xmath1034 , so the spectral measure @xmath120 of the self - adjoint operator @xmath1035 is supported on @xmath1036 , in fact on @xmath1037 $ ] . using the functional calculus for @xmath1038
, we can form the operator @xmath1039 , where the lower limit @xmath1040 indicates that we ignore any point mass at @xmath335 that @xmath120 may have .
we then define @xmath1041.\ ] ] this fuglede - kadison determinant has the following very useful properties ( see @xcite for details ) : @xmath1042 we remark that multiplicativity of @xmath1043 is not obvious , essentially being a consequence of the campbell - baker - hausdorff formula and vanishing of trace on commutators , although for technical reasons a complex variables approach is more efficient .
let @xmath16 , and write elements of @xmath5 as @xmath1044 . the fourier transform identifies @xmath1045 with @xmath1046 , with @xmath1047 being identified with the function @xmath1048 , where @xmath1049 $ ] for @xmath1050 .
any bounded linear operator @xmath1051 on @xmath1045 commuting with the @xmath5-action must have the form of convolution with some element @xmath1052 .
hence the fourier transform @xmath519 of @xmath1053 must give a bounded linear operator on @xmath1054 via pointwise multiplication , and this forces @xmath1055 .
conversely , every @xmath1055 corresponds to an element of @xmath1056 .
this identifies @xmath1056 with @xmath1057 . under this identification , @xmath1058.\ ] ] for @xmath1059 , we observed that @xmath1060 , and this corresponds to multiplication by @xmath1061 .
hence in this case @xmath1062 and so @xmath1063 . indeed , it was the equality of mahler measure , entropy , and the fuglede - kadison determinant in lck s book ( * ? ? ?
* example 3.13 ) that originally inspired deninger to investigate whether this phenomenon extended to more general groups .
he was able to show in @xcite that with some conditions it did .
further work @xcite , @xcite extended the generality , culminating in the comprehensive results of li and thom @xcite . to describe their work , recall that @xmath65 is a bounded linear operator on @xmath1018 . more generally , if @xmath1064 , then there is a bounded linear operator @xmath1065\to{\ell^2({\delta},{\mathbb{c}})}^l$ ] given by right multiplication by @xmath1066 , where @xmath1067
there is an extension of @xmath1043 to such @xmath274 ( see @xcite ) for details .
[ thm : li - thom ] let @xmath6 be a countable discrete amenable group , and let @xmath11 .
suppose that @xmath1068 is injective
. then @xmath1069 . more generally , if @xmath1064 and if @xmath1070 is injective , then @xmath1071 . if @xmath1072 , then @xmath1073
in particular , @xmath1074 .
this is a highly nontrivial fact , since there is no direct connection between @xmath1075 and @xmath380 .
the computation , or even estimation , of the values of fuglede - kadison determinants is not easy . in the next two sections we will explicitly calculate entropy for certain principal actions of the heisenberg group . as pointed out by deninger @xcite , there are examples of lopsided polynomials @xmath13 for which @xmath1076 can be computed by a rapidly converging series .
[ exam : traces ] let @xmath1077 .
write @xmath1078 , where @xmath1079 .
it is easy to see ( * ? ? ?
* lemma 2.7 ) that the spectrum of @xmath24 , considered as an element in @xmath1080 , is contained in @xmath1081 $ ] .
hence we can apply the functional calculus to compute @xmath1082 via the power series for @xmath1083 , yielding @xmath1084 now @xmath1085 is the value of the constant term of @xmath1086 . if @xmath1087 denotes the number of words over @xmath1088 of length @xmath323 whose product is 1 , then clearly @xmath1089 . in any word over @xmath196 with product 1 ,
the number of @xmath68 and @xmath1090 must be equal and similarly with @xmath69 and @xmath1091 , so that @xmath1092 for @xmath323 odd .
the numbers @xmath1093 grow rapidly : @xmath1094 ( see @xcite , although the result stated there is off by a factor of 2 ) .
we thank david wilson for providing us a short _ mathematica _ program that computes @xmath1093 up to @xmath1095 , which is a 33 digit number .
hence @xmath1096 using the trivial estimate @xmath1097 , the last sum has value less than @xmath1098 .
the remaining part therefore gives the value for @xmath1099 , correct to six decimal places .
we remark that @xmath1100 , considered as an operator in @xmath1080 , has been studied intensively , and is closely related to kac s famous ten martini problem .
indeed , the image of @xmath1101 in the rotation algebra factor @xmath1102 of @xmath1080 is harper s operator @xmath1103 ( cf .
kac conjectured that for every irrational @xmath45 the spectrum of @xmath1103 is a cantor set of zero lebesgue measure .
this conjecture was recently confirmed by deep work of avila .
the calculations in the previous example can be carried out just as well in any group @xmath6 containing two elements @xmath68 and @xmath69 .
the only change is that the number @xmath1104 of words over @xmath1088 whose product is 1 will be different , depending on @xmath6 . for comparison with the example
, we work this out for @xmath1105 with commuting generators @xmath1106 , and for @xmath1107 , the free group with generators @xmath1106 .
recalling that @xmath1078 , let us define @xmath1108 the preceding example shows that @xmath1109 . for @xmath1105
, we are computing entropy for @xmath13 considered as a polynomial in the commuting variables @xmath1110 , and this is given by mahler measure to be @xmath1111 observe that any word in @xmath1112 whose product is 1 abelianizes to one in @xmath1 , and so @xmath1113 , which is reflected in the inequality @xmath1114 .
the case @xmath1107 is more interesting .
here any word in the generators @xmath1110 whose product is 1 must also give a word in any group @xmath6 containing @xmath1106 with product 1 , so that @xmath1115 for all groups @xmath6 .
hence @xmath1116 , so that @xmath1117 gives a universal upper bound . to compute @xmath1117 ,
start with the generating function for @xmath1118 , which by ( * ? ? ?
* i.9 ) is known to be @xmath1119 letting @xmath1120 then a calculation with _ mathematica _ shows that @xmath1121\cong 1.514787\ ] ] indeed @xmath1122 , but only in the fifth decimal place .
we remark that lewis bowen @xcite has extended a notion of entropy to actions of sofic groups , and in particular free groups ( although these do not have flner sequences ) . in (
* example 1.1 ) he shows that in the previous discussion @xmath1117 equals the sofic entropy of the corresponding algebraic @xmath1123-action .
it is somewhat surprising that here @xmath1117 is the logarithm of an algebraic number , since in the case @xmath16 with @xmath1124 this appears not to generally be the case .
for example , when @xmath1125 and @xmath1126 , smyth @xcite has computed the logarithmic mahler measure to be @xmath1127 $ ] , where @xmath578 is the riemann zeta - function .
next we extend the face entropy inequality described for principal @xmath1128-actions in ( * ? ? ?
* remark 5.5 ) to principal @xmath10-actions .
this inequality proves that many such actions have strictly positive entropy .
we start with the basic case .
[ prop : basic - inequality ] let @xmath1129 with @xmath1130
. then @xmath1131 . by definition
the map @xmath1132 has kernel @xmath1133 , and is surjective since multiplication by @xmath1134 is injective on @xmath1135 $ ] .
define @xmath1136 by @xmath1137 .
then for every @xmath571 we have that @xmath1138 iff @xmath1139 .
the algebraic @xmath1-action @xmath1140 on @xmath1133 has entropy @xmath1141 .
fix @xmath1142 and @xmath1143 .
then for sufficiently large rectangles @xmath1144 there is a @xmath1145-separated set @xmath1146 with @xmath1147 .
note that since @xmath401 is translation invariant , for every @xmath1148 the translated set @xmath1149 is also @xmath1145-separated .
let @xmath1150 be arbitrary , and put @xmath1151 .
the condition for @xmath66 to be in @xmath57 is that @xmath1152 , which in terms of the @xmath1153 becomes @xmath1154 let @xmath1155 . for each @xmath1156
we will construct a sequence @xmath1157 in @xmath1158 that will be used to create an @xmath1159-separated set in @xmath57 .
put @xmath1160 for all @xmath309 , so that is trivially satisfied for @xmath333 .
define @xmath1161 for @xmath1162
. then is satisfied at @xmath1163 since @xmath1164 .
since @xmath1165 is surjective on @xmath1158 , for each @xmath1166 there is a @xmath1167 such that @xmath1168 let @xmath1169 for @xmath1170 .
since @xmath1171 , it follows that is satisfied at @xmath1172 for all choices of @xmath1166 and @xmath1173 .
similarly , for each @xmath1174 there is a @xmath1175 with @xmath1176 put @xmath1177 for @xmath1178 . then holds for @xmath1179 and all choices of @xmath1180 .
continuing in this way , for every @xmath1181-tuple @xmath1156 we have constructed @xmath1182 so that is satisfied for @xmath1183 .
each choice can be further extended to find @xmath1153 for @xmath1184 for which the resulting point is in @xmath57 .
identify @xmath296 with @xmath9 via @xmath1185 , and consider @xmath1186 .
we claim that the @xmath1187 points in @xmath57 constructed above are @xmath1188-separated . for at the first index @xmath263 for which @xmath1189 , the points @xmath1190 and @xmath1191 differ by at least @xmath1159 at some coordinate of @xmath1192 .
finally , if we choose @xmath1193 be be very long in the @xmath68-direction compared with the @xmath70-direction , and make @xmath1181 small compared with both , we can make @xmath1194 as right - invariant as we please .
hence there is a flner sequence @xmath1195 in @xmath9 with @xmath1196 and thus @xmath1131 .
recall from section [ sec : algebraic - actions ] that the _ newton polygon _ @xmath1197 of @xmath1198 is the convex hull in @xmath84 of those points @xmath85 for which @xmath1199 .
a _ face _ of @xmath1197 is the intersection of @xmath1197 with a supporting hyperplane , which is either a point or a line segment . for each face @xmath274 of @xmath1197 let @xmath1200 . for every face @xmath274 of @xmath1197
there is a change of variables followed by multiplication by a monomial transforming @xmath13 so that @xmath274 now lies on the @xmath68-axis with the rest of @xmath1197 in the upper half - plane .
since entropy is invariant under such transformations , we can apply proposition [ prop : basic - inequality ] to obtain the following face entropy inequality .
[ cor : face - entropy ] if @xmath127 and @xmath274 is a face of @xmath1197 , then @xmath1201 .
face entropies are essentially logarithmic mahler measures of polynomials in commuting variables , and so easy to compute . observe that if @xmath1202 , then the corollary shows that @xmath1203 for every face @xmath274 of @xmath1197 , and that proposition [ prop : zero - entropy ] gives a complete characterization of what @xmath1204 can be . indeed , smyth used the face entropy inequality as the starting point for his proof of proposition [ prop : zero - entropy ] .
however , the algebraic complexity of @xmath10 prevents a direct extension of his methods , leaving open a very interesting question .
characterize those @xmath127 for which @xmath1202 .
recall that the pinsker @xmath1205-algebra of a measure - preserving action @xmath7 is the largest @xmath1205-algebra for which the entropy of @xmath7 on this @xmath1205-algebra is zero .
an action has _ completely positive entropy _ if its pinsker @xmath1205-algebra is trivial .
an old argument of rohlin shows that the pinsker @xmath1205-algebra of an algebraic action @xmath7 on @xmath26 is invariant under translation by every periodic point .
hence if the periodic points are dense , then the pinsker @xmath1205-algebra is invariant under all translations , and so arises from the quotient map @xmath996 , where @xmath829 is a compact @xmath7-invariant subgroup .
thus the restriction of @xmath7 to its pinsker @xmath1205-algebra is again an algebraic action ( see ( * ? ? ?
6.2 ) ) , providing one reason for the importance of the problem above . for algebraic @xmath1128-actions
there is an explicit criterion for completely positive entropy in terms of associated prime ideals ( see ( * ? ? ?
* thm . 6.5 ) ) , but even in the case of heisenberg actions no similar criterion is known .
characterize the algebraic @xmath9-actions with completely positive entropy . for algebraic @xmath1128-actions , completely positive entropy is sufficient to imply that they are isomorphic to bernoulli shifts @xcite .
is the same true for algebraic @xmath9-actions ?
if an algebraic @xmath9-actions has completely positive entropy , is it necessarily measurably isomorphic to a bernoulli @xmath9-action ?
let @xmath6 be a countable discrete group , and @xmath7 be an algebraic @xmath6-action on @xmath26 .
a point @xmath28 is _ periodic _ for @xmath7 if its @xmath6-orbit is finite .
the stabilizer @xmath1206 of such a point @xmath66 has finite index in @xmath6 , and we will need a generous supply of such subgroups . call @xmath6 _ residually finite _ if , for every finite subset @xmath107 of @xmath6 , there is a finite - index subgroup @xmath164 of @xmath6 such that @xmath1207 .
every finite - index subgroup @xmath164 of @xmath6 contains a further finite - index subgroup @xmath1208 of @xmath164 that is normal in @xmath6 , so that residual finiteness can be defined using finite - index normal subgroups .
if @xmath1209 is a sequence of finite - index subgroups of @xmath6 , we will say @xmath1210 if , for every finite set @xmath1211 , there is an @xmath1212 such that @xmath1213 for all @xmath1214 . for a finite - index subgroup @xmath164 of @xmath6 ,
let @xmath1215 if @xmath164 is normal in @xmath6 , then for @xmath168 and @xmath27 we have that @xmath1216 . hence in this case
@xmath1217 is @xmath6-invariant , since if @xmath1218 then @xmath1219 .
we will focus on expansive principal actions .
let @xmath6 be a countable residually finite discrete group , and let @xmath11 be expansive .
recall the notations and results from proposition [ prop : cover ] . for a finite - index subgroups @xmath164 of @xmath6 ,
let a superscript of @xmath164 on a space denote the set of those elements in the space fixed by @xmath164 , so for example @xmath1220 [ prop : fix ] let @xmath6 be a countable discrete group , and @xmath11 be expansive . using the notations of proposition [ prop : cover ] , for every finite - index subgroup @xmath164 of @xmath6 we have that @xmath1221 any point @xmath1222
can be lifted to a point @xmath1223 , and then the proof of proposition [ prop : cover ] yields @xmath164-invariant points at every step .
note that @xmath1224 is a free abelian group of rank @xmath1225 $ ] , the index of @xmath164 in @xmath6 , and that @xmath65 is an injective endomorphism on this group by expansiveness .
[ cor : fixed - point - count ] under the hypotheses in proposition [ prop : fix ] , @xmath1226 observe that @xmath65 is an injective linear map on the real vector space @xmath1227 of dimension @xmath1228 $ ] and maps the lattice @xmath1224 to itself .
if @xmath190 is any @xmath1229 matrix with integer entries and nonzero determinant , then it is easy to see , for example from the smith normal form , that @xmath1230 .
since we will need to complexify some spaces in order to use complex eigenvalues , let us say a word about conventions regarding determinants .
if @xmath190 is an @xmath1229 real matrix , we could regard @xmath190 as an @xmath1229 complex matrix acting on @xmath1231 , or as a @xmath1232 real matrix acting on @xmath1233 , and these have different determinants .
we will always use the first interpretation . with corollary [ cor : fixed - point - count ]
we begin to see the connections among periodic points , entropy , and fuglede - kadison determinants . for expansive actions ,
periodic points are separated , and so @xmath1234 should approximate , or at least provide a lower bound for , the entropy @xmath1235 .
but by , this is also a finite - dimensional approximation to the logarithm @xmath1236 of the fuglede - kadison determinant of @xmath65 .
the main technical issue is then to show both of these approximations converge to the desired limits . for expansive @xmath15
is is relatively easy , but for general @xmath15 the are numerous difficulties to overcome .
we can now deduce two important properties of expansive principal actions .
[ prop : dense - and - positive - entropy ] let @xmath6 be a residually finite countable discrete amenable group , and let @xmath11 be expansive
. then 1 .
the @xmath15-periodic points are dense in @xmath57 , and 2 .
if @xmath1237 i.e. , if @xmath13 is not invertible in @xmath21 , then @xmath1238 . for ( 1 )
, let @xmath61 and find @xmath1239 with @xmath1240 . by prop .
[ prop : cover](4 ) , there is a finite subset @xmath107 of @xmath6 such that if @xmath1241 denotes the restriction of @xmath857 to @xmath107 and 0 elsewhere , then @xmath1242 is close to @xmath66 . by enlarging @xmath107 if necessary , we can assume that @xmath1243 is small .
let @xmath164 be a finite - index subgroup of @xmath6 such that @xmath1244 for distinct @xmath227 .
let @xmath1245 .
then @xmath1246 and @xmath1247 is close to @xmath66 . for ( 2 ) ,
again choose a finite @xmath107 so that @xmath1243 is small , where we may assume that @xmath1248 .
find @xmath164 with @xmath1249 for distinct @xmath227 .
if @xmath274 is any sufficiently invariant flner set , then @xmath1250 is about @xmath1251 $ ] . then for every choice of @xmath1252 for @xmath1253 , the points @xmath1254 are @xmath1255-separated .
hence if @xmath187 is any flner sequence , then for every @xmath1256 we have that @xmath1257}>0 . \qedhere\ ] ] for @xmath16 , it turns out that the periodic points for @xmath1258 are always dense in @xmath297 for every finitely generated @xmath4-module @xmath30 ( * ? ? ?
the simple example of multiplication by @xmath1259 on @xmath1260 dualizes to an automorphism of a compact group with no nonzero periodic points , since @xmath1261 is invertible in @xmath1260 ( see ( * ? ? ?
* example 5.6(1 ) ) ) .
we do not know the answer to the following .
let @xmath6 be a countable discrete residually finite group , and @xmath30 be a finitely generated @xmath21-module .
must the @xmath1258-periodic points always be dense in @xmath297 ?
we now focus on using periodic points to calculate entropy .
it is instructive to see how these calculations work in a simple example .
[ exam : periodic - golden - mean ] let @xmath426 , and let @xmath1262 $ ] .
let @xmath1263 and @xmath947 .
then @xmath187 is a flner sequence in @xmath55 that is also a fundamental domain for @xmath1264 .
denote by @xmath1265 the set of all @xmath323-th roots of unity in @xmath489 . for @xmath1266 let @xmath1267 .
then @xmath1268 , so the @xmath1269 form an eigenbasis for the shift on @xmath1270 .
hence @xmath1271 for each @xmath1266 .
we can consider the elements of @xmath1270 as elements in the @xmath323-dimensional complex vector space @xmath1272 .
the matrix of @xmath65 with respect to the basis @xmath1273 is the circulant matrix @xmath1274 -1 & -1 & 1 & 0 & \ldots&0&0 \\ 0 & -1 & -1 & 1 & \ldots&0 & 0\\ & & & \ddots & & & \\ 1 & 0 & & \ldots & & -1 & -1 \\
-1 & 1 & & \ldots & & 0 & -1 \end{bmatrix*}.\ ] ] we can compute the determinant of @xmath1275 using the eigenbasis @xmath1276 .
factor @xmath1277 , where @xmath692 and @xmath1278 .
then @xmath1279 then @xmath1280 since @xmath15 is expansive , periodic points are separated , and so @xmath1281 is certainly a lower bound for @xmath1235 .
but it is also easy in this case to see that it is an upper bound , using for example approximations from homoclinic points as in the proof of proposition [ prop : dense - and - positive - entropy ] .
note that if @xmath1282 $ ] had a root @xmath588 , then the factor @xmath1283 in the calculation of determinant would occasionally be very small , which could cause the limit not to exist .
this is one manifestation of the difficulties with nonexpansive actions .
we turn to the heisenberg case @xmath8 . for @xmath1284 put @xmath1285 , which is a normal subgroup of @xmath9 of index @xmath1286 .
let @xmath127 be expansive . recall that @xmath1287 is an expansive constant for @xmath15 .
in particular , if @xmath164 is a finite - index subgroup of @xmath9 , and if @xmath1288 , then for any fundamental domain @xmath1193 of @xmath164 there is a @xmath1289 such that @xmath1290 .
a bit of notation about the limits we will be taking is convenient .
if @xmath1291 is a quantity that depends on @xmath1292 , we write @xmath1293 to mean that for every @xmath409 there is a @xmath1294 such that @xmath1295 for every @xmath1296 and all sufficiently large @xmath330 and @xmath814 . [ thm : periodic - point - limit ] let @xmath127 be expansive , and define @xmath1292 as above .
then @xmath1297 } \log|{\operatorname{\mathsf{fix}}}_{{{\lambda}_{rq , sq , q}}}({\alpha_f})|= { \mathsf{h}}({\alpha_f}).\ ] ] in light of example [ exam : periodic - golden - mean ] , it is tempting to use a fundamental domain for @xmath1292 of the form @xmath1298 , but such a @xmath1193 is far from being right - flner , since right multiplication by @xmath68 will drastically shear @xmath1193 in the @xmath70-direction . the method used in (
5.7 ) is to decompose @xmath1193 into pieces , each of which is thin in the @xmath69-direction , and translate these pieces to different locations in @xmath9 .
the union of these translates will still be a fundamental domain , but now it will also be flner , and so can be used for entropy calculations .
this method depends in general on a result of weiss @xcite , and ultimately goes back to the @xmath958-quasi - tiling machinery of ornstein and weiss @xcite . in our case
, we can give a simple description of this decomposition .
choose integers @xmath1299 such that @xmath1300 but @xmath1301 as @xmath1302 .
consider the set @xmath1303 it is easy to verify that since @xmath1299 is small compared with @xmath1304 , right multiplication by a fixed @xmath237 creates only small distortions , and so for every @xmath237 we have that @xmath1305 define @xmath1306 - 1}x^{rqj}f_{rq , a(q)j , q}\,\,,\ ] ] where if @xmath1299 does not evenly divide @xmath1307 , make the obvious modification in the last set .
then @xmath1308 is also a fundamental domain for @xmath1292 , but now it is also a flner sequence as @xmath1302 by . by the separation property of periodic points , for all
@xmath1309 we have that @xmath1310 . since @xmath1311 is flner , @xmath1312 for the reverse inequality , let @xmath1143 and let @xmath1313 .
choose a finite set @xmath1314 such that @xmath1315 .
the sets @xmath1316 also form a flner sequence , and @xmath1317 as @xmath1302 .
fix @xmath1304 , @xmath330 , and @xmath814 for the moment , and choose a @xmath1318-separated set @xmath196 of maximal cardinality .
for every @xmath1319 , let @xmath1320 be its lift , with @xmath1321 and @xmath862 .
write @xmath1322 for the unique point with @xmath1323 for all @xmath1324 .
our choice of @xmath125 implies that the points in @xmath1325 are @xmath1326-separated , hence distinct .
hence @xmath1327 .
since @xmath1328 is flner and @xmath1317 as @xmath1302 , we see that @xmath1329 completing the proof .
we apply this result , combined with corollary [ cor : fixed - point - count ] , to compute entropy for expansive principal @xmath9-actions @xmath15 . using the above notations ,
let @xmath1330 , so that @xmath1331 we will compute this determinant by decomposing @xmath1332 into @xmath65-invariant subspaces , each having dimension 1 or @xmath1304 . to do this , for each @xmath1333 let @xmath1334 observe that @xmath1335 , and similarly @xmath1336 , so that @xmath1337 is a common eigenvector for @xmath1338 and @xmath565
however , @xmath1339 let @xmath1340 , and @xmath1341 . for arbitrary @xmath1342
let @xmath1343 by the above , for every @xmath1342 and @xmath1344 , the subspace @xmath1345 is invariant under the right action of @xmath9 . now let @xmath127 , and assume that @xmath13 is expansive .
adjusting @xmath13 by a power of @xmath68 if necessary , we may assume that @xmath13 has the form @xmath1346 , where each @xmath1347 $ ] with @xmath1348 and @xmath1349 .
the action of @xmath65 on @xmath1337 is then given by @xmath1350 if @xmath655 ,
then @xmath1351 , so is given by the @xmath273 matrix @xmath1352 $ ] .
if @xmath1353 , then the matrix of @xmath65 on @xmath1354 takes the following @xmath1355 circulant - like form , where for notational convenience we assume that @xmath1356 : by our expansiveness assumption @xmath65 is injective , and each @xmath1345 is @xmath65-invariant , hence @xmath1357 for all @xmath1342 and all @xmath1344 .
now suppose that @xmath1358 .
for convenience we will assume that both @xmath330 and @xmath814 are primes distinct from @xmath1304 .
then @xmath1359 is a basis for @xmath1332 .
note that if @xmath1353 , then @xmath1360 for @xmath1361 . since @xmath1362 by relative primeness of @xmath814 and @xmath1304 , we can parameterize the spaces @xmath1354 by @xmath1363 .
this gives the @xmath65-invariant decomposition @xmath1364 we now evaluate the limit of @xmath1365 as @xmath1366 using the decomposition . on each of the @xmath1367-dimensional spaces @xmath1368 in @xmath65 acts as multiplication by @xmath1369 .
hence the contribution to of the first large summand in is @xmath1370 by expansiveness , @xmath1369 never vanishes for @xmath1342 , so that @xmath1371 is continuous on @xmath579 . by convergence of the riemann sums to the integral , as @xmath1366 we have that @xmath1372 the additional factor of @xmath1304 in the denominator of shows that it converges to 0 as @xmath1366 . for the spaces
@xmath1354 with @xmath1353 , the expansiveness assumption shows that @xmath1373 never vanishes for @xmath1342 , so again @xmath1374 is continuous on @xmath579 .
hence for @xmath1353 , as @xmath1366 we have that @xmath1375 adding these up over @xmath1353 , and observing that @xmath1376 , we have shown the following .
[ thm : expansive - entropy ] let @xmath1377 be expansive .
then @xmath1378 where the matrices @xmath1379 are as given above . at first glance
the denominator @xmath1380 in appears puzzling .
the explanation is that one @xmath1304 comes from averaging over the @xmath1304-th roots of unity , while the other @xmath1304 comes from the size of the matrices @xmath1379 . from the point of view of von neumann
algebras , we should really be using the `` normalized determinant '' @xmath1381 corresponding to the normalized trace on @xmath1231 , and then the second @xmath1304 would not appear . for expansive polynomials in @xmath10 that are linear in @xmath68 the entropy formula in the preceding theorem can be simplified considerably .
[ thm : linear - entropy ] let @xmath1382 , where @xmath1383 and @xmath1384 are laurent polynomials in @xmath1385 $ ] .
if @xmath15 is expansive , then @xmath1386 where @xmath1387 and similarly for @xmath96 . let @xmath1388 , and assume that neither @xmath24 nor @xmath96 vanish anywhere on @xmath579 .
if @xmath15 is expansive , then @xmath1389 .
after the change of variables @xmath1390 , @xmath1391 , and @xmath1392 , we can apply theorem [ thm : linear - expansive ] to conclude that either @xmath1393 for all @xmath1394 , or @xmath1395 for all @xmath1394 .
the result then follows from by integrating over @xmath1394 .
the intuition behind is simple to explain .
the matrices @xmath1379 for @xmath1353 take the form @xmath1396 then @xmath1397 for fixed @xmath578 the first product behaves like @xmath1398 , and similarly for @xmath96 . whichever is larger will dominate , suggesting the formula . to make this intuition precise , we require several lemmas . for every @xmath1399 , @xmath508 , and @xmath309 ,
@xmath1400 if @xmath1401 , then @xmath1402 . if @xmath1403 , then @xmath1404 if @xmath1405 $ ] , then shows that @xmath1406 .
we will use the convention @xmath1407 . with the usual conventions about arithmetic and inequalities involving @xmath1408
, the results that follow will make sense and are true even if some of the polynomials are 0 .
suppose that @xmath1409 $ ] has degree @xmath1410 .
then for every @xmath508 and @xmath309 , @xmath1411 fix @xmath309 and set @xmath1412 , so that @xmath1413
. write @xmath1414 , where @xmath1415 and @xmath1416 .
factor @xmath1417 . then using the previous lemma , @xmath1418 [ lem : max - entropy ] let @xmath1419 $ ] each have degree @xmath1410 .
then for every @xmath1420 , @xmath1421 if @xmath1420 , then the previous lemma implies that for every @xmath508 we have that @xmath1422 and @xmath1423 .
hence @xmath1424 and the second inequality follows by integrating over @xmath508 .
for the first inequality , observe that @xmath1425 and similarly for @xmath119 .
we need one more property of mahler measure , proved by david boyd @xcite .
recall that @xmath1426 , and by convention @xmath1427 .
[ thm : boyd - continuity ] the map @xmath1428 given by @xmath1429 is continuous .
continuity is clear when the coefficients remain bounded away from 0 since the roots are continuous functions of the coefficients , but for example if @xmath1430 then continuity is more subtle .
boyd s proof , which also applies the polynomials in several variables with bounded degree , uses graeffe s root - squaring method , managing to sidestep various delicate issues , leading to a remarkably simple proof .
if @xmath1431 , then we have seen in that @xmath1432 by , for any complex numbers @xmath204 and @xmath205 , @xmath1433 hence @xmath1434 it follows that @xmath1435 let @xmath409 . writing @xmath1436 and @xmath1437 ,
note that these are laurent polynomials of uniformly bounded degree @xmath1410 in @xmath1438 $ ] , where the degree of a laurent polynomial is the length of its newton polygon . applying [ lem : max - entropy ] to @xmath1439 and @xmath1440 , we obtain that for @xmath1441 , @xmath1442 we must now consider the possibility that for some @xmath508 both @xmath1443 and @xmath1444 vanish as polynomials in @xmath1438 $ ] , and claim this never happens provided that @xmath15 is expansive .
write @xmath1445 and @xmath1446 , where @xmath1447 $ ] .
if @xmath1448 were a common zero for all the @xmath1449 and @xmath1450 , then its minimal polynomial over @xmath1260 would divide every @xmath1449 and @xmath1450 .
but then the nonzero point @xmath1451 would have @xmath573 , contradicting expansiveness .
thus the function @xmath1452 is continuous on @xmath429 . hence the sums @xmath1453 simultaneously approximate @xmath1235 by theorem [ thm : expansive - entropy ] , and are also riemann sums over @xmath1454 of the continuous function @xmath1455 , and so converge to the integral in , concluding the proof . our proof of theorem [ thm : linear - entropy ] made use of the expansiveness hypothesis on @xmath15 in order to use periodic points to approximate entropy .
but surely the entropy formula is valid more generally .
is the entropy formula for linear polynomials in @xmath10 valid for arbitrary @xmath1383 and @xmath1384 in @xmath1385 $ ] ?
it will follow from results in the next section that is valid if either @xmath1456 or @xmath1457 .
let us turn to the quadratic case @xmath1458 .
we start with a simple result about determinants .
let @xmath1459 , @xmath1460 , and @xmath1461 @xmath1462 be arbitrary complex numbers .
then @xmath1463 if @xmath1464 for every @xmath1465 , where subscripts are taken mod @xmath1304 , then the value of this determinant simplifies to @xmath1466 where @xmath692 and @xmath1278
. taking subscripts mod @xmath1304 , a permutation @xmath343 of @xmath1467 contributes a nonzero summand in the expansion of the determinant if and only if it has the form @xmath1468 , where @xmath1469 , 1 , or 2 .
the sequences @xmath1470 corresponding to permutations are precisely the closed paths of length @xmath1304 in the labeled shift of finite type depicted below . the paths @xmath1471 and @xmath1472 give the terms @xmath1473 and @xmath1474 , respectively , while it is easy to check that the golden mean shift of finite type produces middle term of the result .
if @xmath1475 for all @xmath1465 , then each occurrence of a block 20 in a closed path of length @xmath1304 can be replaced by the block 11 , changing the factor @xmath1476 to @xmath1477 together with an appropriate sign change .
the result of these substitutions is that every closed path of length @xmath1304 in the golden mean shift gives the same contribution @xmath1478 to the expansion , and there are @xmath1479^q=\tau^q+\sigma^q$ ] of them .
we use this to give the quadratic analogues of and .
let @xmath1480 .
suppose that @xmath1353 , and for @xmath1342 define @xmath1379 as above .
then @xmath1481 if @xmath1482 , then @xmath1483 motivated by theorem [ thm : expansive - entropy ] and the rigorous results from the expansive linear case , we can now formulate a reasonable conjecture about entropy in the quadratic case .
as in the linear case , for fixed irrational @xmath578 , the growth rate of the first and third terms in are given my @xmath1484 and @xmath1485 , respectively .
the growth rate for the second term should be the same for almost every choice of @xmath1342 , and if so denote this value by @xmath1486 .
for example , in case @xmath1482 , from we see that @xmath1487 .
let @xmath1480 . is @xmath1488 in particular , if @xmath1482 , is @xmath1489 fix @xmath1490 , and let @xmath1491 , @xmath1492 , and @xmath1493 .
these satisfy the conditions for , and this formula becomes @xmath1494 for example , letting @xmath1495 , for each of the three functions in there is a range of @xmath578 for which it is the maximum .
a similar analysis can be carried out for higher degree polynomials , but the evaluation of the relevant determinants now involves a finite family of more complicated ( and interesting ) shifts of finite type .
the methods of the previous section to compute entropy have some serious limitations because of the expansiveness assumptions .
there is a more geometric approach to entropy , using the theory of lyapunov exponents , which deninger used in @xcite to calculate fuglede - kadison determinants , or equivalently by theorem [ thm : li - thom ] , entropy , in much greater generality . to motivate this approach , first recall the linear example @xmath1496 . for @xmath647
defined in and @xmath1497 with @xmath573 , we have from that @xmath1498 c_0.\ ] ] for irrational @xmath578 , the products in have growth rate @xmath1499 and the limit is the same for almost every @xmath577 by ergodicity of irrational rotations . for toral automorphisms , entropy is the sum of the growth rates on various eigenspaces that are positive . by analogy , we would expect entropy here to be the integral of the positive growth rates , i.e. , that @xmath1500 this is a special case of , but with no assumptions on @xmath24 .
indeed , since the eigenspaces here are 1-dimensional , the techniques used in @xcite can be adapted to prove the validity of for every @xmath1501 $ ] . however , since this will be subsumed under deninger s results , there is no need to provide an independent proof here . to state the main result in @xcite , we need to give a little background . for each irrational @xmath508
there is the rotation algebra @xmath1502 , which is the von neumann algebra version of the twisted @xmath1503 algebras used in allan s local principle ( see @xcite for details ) .
there are also natural maps @xmath1504 .
as explained in @xcite , there is a faithful normalized trace function @xmath1505 on each @xmath1502 such that @xmath1506 for every @xmath1507 .
this implies that for determinants we have @xmath1508 hence we need a way of evaluating the integrands for @xmath1509 .
suppose that @xmath127 is monic in @xmath69 and of degree @xmath1510 , and so has the form @xmath1511 where the @xmath1512 $ ] and @xmath1513 .
calculations similar to example [ exam : y^2-xy-1 ] show that if @xmath1514 , then , for every @xmath323 , @xmath1515 put @xmath1516 and let @xmath1517 be the corresponding cocycle .
then the recurrence relation for the @xmath568 can be written as @xmath1518 [ thm : met ] let @xmath1051 be an invertible ergodic measure - preserving transformation on a measure space @xmath1520 , and @xmath1521 be a measurable map from @xmath1522 to the @xmath1519 complex matrices such that @xmath1523 .
then there is a @xmath1051-invariant measurable set @xmath1524 with @xmath1525 , an @xmath1526 , real numbers @xmath1527 and multiplicities @xmath1528 such that @xmath1529 , and measurable maps @xmath1530 from @xmath1531 to the space of @xmath1532-dimensional subspaces of @xmath1533 , such that for all @xmath1534 we have that to apply this result to our situation , fix an irrational @xmath508 .
let @xmath1539 be given by @xmath1540 .
let @xmath1541 . since the entries of @xmath1542 are continuous in @xmath577 , clearly @xmath1543 .
hence there are lyapunov exponents @xmath1544 and multiplicities @xmath1545 . with these in hand , we can now state deninger s main result from @xcite as it applies to the heisenberg case .
[ thm : deninger ] let @xmath1546 , where the @xmath1512 $ ] and @xmath1547 . for every irrational @xmath578 denote the lyapunov exponents for @xmath1548 and @xmath1549 as above by @xmath1544 with multiplicities @xmath1545 .
then @xmath1550 and hence by theorem [ thm : li - thom ] and , @xmath1551 let @xmath1552 . for each @xmath559
there is a nonzero vector @xmath1553 and a multiplier @xmath1554 with @xmath1555 such that @xmath1556 since the determinants of these matrices all have absolute value 1 , there is exactly one nonnegative lyapunov exponent , of multiplicity one . for each irrational @xmath508 its value
is given by @xmath1557 , and hence @xmath1558 a numerical calculation of the graph of @xmath1559 is shown in figure [ fig : surface ] , and indicates the complexity of these phenomena even in the quadratic case .
although lyapunov exponents are generally difficult to compute , there is a method to obtain rigorous lower bounds on the largest lyapunov exponent known as `` herman s subharmonic trick . ''
its use in our context was suggested to us by michael bjorklund .
let @xmath1546 , where the @xmath1560 $ ] , so that only nonnegative powers of @xmath68 are allowed , and @xmath1130 .
for every irrational @xmath508 let @xmath1561 denote the largest lyapunov exponent resulting from theorem [ thm : deninger ] .
then @xmath1562 , where @xmath1548 is the matrix given in , and @xmath1563 denotes the spectral radius of a complex matrix .
in particular , @xmath1564 fix an irrational @xmath508 .
put @xmath1565 , and let @xmath1566 . for a complex matrix
@xmath1567 $ ] define @xmath1568 .
theorem [ thm : met ] shows that for almost every @xmath588 we have that @xmath1569 expand @xmath1570,\ ] ] where the @xmath1571 are polynomials in @xmath68 with complex coefficients .
now each @xmath1572 is subharmonic for @xmath1399 , and hence @xmath1573 is also subharmonic for @xmath1399 .
thus @xmath1574 furthermore , @xmath1575 as @xmath945 . the entries in @xmath1548 are uniformly bounded above , and hence @xmath1576 is uniformly bounded for all @xmath309 and @xmath588 .
thus @xmath1577 observe that since only nonnegative powers of @xmath68 are allowed , this result is reminiscent of the face entropy inequality in corollary [ cor : face - entropy ] . it is stronger since it gives a lower bound for every irrational @xmath578 , but the integrated form is weaker since it uses only the top lyapunov exponent to give a lower bound for the entropy of the face corresponding to @xmath1578 . [
exam : big - example ] we finish by returning to example [ exam : y^2-xy-1 ] .
let @xmath646 . using the change of variables @xmath1390 , @xmath1391 and
@xmath1392 , @xmath13 becomes monic and linear in @xmath69 , hence by the previous theorem we can compute that @xmath1202 . on the other hand , treating @xmath13 as monic and quadratic in @xmath69 , we see that the lyapunov exponents must all be @xmath1579 . but the determinant has absolute value 1 , and so in fact the lyapunov exponents must vanish almost everywhere .
in other words , @xmath1580 by taking transposes to reverse the order of the product , we obtain from the introduction .
although this appears to be a simple result , we have not been able to obtain this as a consequence of any known results in random matrix theory . manfred einsiedler and klaus schmidt , _ markov partitions and homoclinic points of algebraic @xmath5-actions _ , in : dynamical systems and related topics , proc .
* 216 * ( 1997 ) , interperiodic publishing , moscow , 259279 .
douglas lind , klaus schmidt , and evgeny verbitskiy , _ entropy and growth rate of periodic points of algebraic @xmath5-actions _ , in : dynamical numbers : interplay between dynamical systems and number theory , ed .
s. kolyada , yu .
manin , m. mller , p. moree and t. ward , contemp .
532 , american mathematical society , providence , r.i . , 2010 .
s. a. yuzvinskii , _ metric properties of the endomorphisms of compact groups _ , izv .
nauk sssr ser
. mat . * 20 * ( 1965 ) , 12951328 ( russian ) ; translated in amer . math .
* 66 * ( 1968 ) , 6398 . | the study of actions of countable groups by automorphisms of compact abelian groups has recently undergone intensive development , revealing deep connections with operator algebras and other areas .
the discrete heisenberg group is the simplest noncommutative example , where dynamical phenomena related to its noncommutativity already illustrate many of these connections .
the explicit structure of this group means that these phenomena have concrete descriptions , which are not only instances of the general theory but are also testing grounds for further work .
we survey here what is known about such actions of the discrete heisenberg group , providing numerous examples and emphasizing many of the open problems that remain . |
the motivation for this paper is to solve exactly an eigenvalue problem that arises in @xcite and show that the exact scattering phase we obtain matches the string theory calculation .
this provides further confirmation of the duality between the nonsinglet sectors of the @xmath1 matrix model and time dependent long string configurations in @xmath0 string theory .
the problem is @xmath2 which can also be written as @xmath3 where @xmath4 .
it arises in general when one tries to compute the meson ( i.e. adjoint sector ) spectrum of the @xmath1 matrix model @xcite .
as opposed to the singlet sector , which for a general potential reduces to @xmath5 independent fermions , the non - singlet sector hamiltonian includes interactions between the different eigenvalues .
this interaction piece can be treated perturbatively in @xmath6 @xcite , and when , as in @xcite , one is interested in the double scaling limit around an inverted harmonic oscillator potential , the problem reduces to ( [ eq : first ] ) .
we obtain the exact eigenstate of the above problem for any @xmath7 by summing up a perturbation series .
the answer we get looks very simple , and there may be another , easier way of obtaining it .
we set up the problem following the notation of @xcite .
the eigenvalue equation ( [ eq : first ] ) can be written in fourier space as : @xmath8 to start we take @xmath9 . in @xmath10 space this corresponds to a wave with very negative energy bouncing off an essentially linear potential . in @xmath11 space
the picture is a wave with very positive energy which barrels through the potential very fast , so that its kinetic term is essentially linear , i.e. having a relativistic dispersion : @xmath12 the solution is @xmath13 @xmath14 for large @xmath10 we can fourier transform back and obtain ( up to overall phase ) in front of the @xmath15 , and also it is not taken into account that because of the @xmath16 in front of the term quadratic in @xmath11 in ( b.6 ) the determinants arising from the gaussian integrals have a relative factor of @xmath17 . ]
@xmath18 moving away from the strict limit @xmath9 we will have a scattering phase @xmath19 the idea is now to compute the phase @xmath20 by expanding the wave - function in powers of @xmath21 . in @xcite
the first correction is computed and at that order matches the conjectured exact result @xmath22 the perturbed wave - function that is obtained there has a singularity at @xmath23 .
this presumably reflects the fact that in @xmath10 space the wave - function is unstable to tunneling under the barrier .
when we sum up all the corrections the singularities will turn out to add up to a branch cut , and when we anti - symmetrize properly the branch cut will disappear .
we will then explicitly verify the solution .
for now , we write the purported solution : @xmath24and plug it in .
we could have a priori made the expansion a prefactor multiplying the exponential , but a little experimentation shows that it is easiest to do the expansion inside the exponent . as in @xcite
we obtain for the first correction : @xmath25 @xmath26 continuing and computing @xmath27 leads us to conjecture that the general correction will take the form @xmath28 where @xmath29 and @xmath30 are unknown coefficients . plugging into ( [ eq : eigen ] ) we obtain recursion relations that determine all the @xmath29 and @xmath30 .
the goal in the next subsection is to solve these .
first , it will be useful to note that @xmath31_k^{k-2i}\\ = \left(-{2 a_n \over n } \right ) { 1 \over ( e^{\pi k } - e^{-\pi k})^{2n}}. \end{split}\ ] ] and thus @xmath32 to keep expressions manageable , we define generating functions that encode the information in the @xmath29 and @xmath30 : @xmath33 we see that the combination @xmath34 appears a lot , so henceforth we set @xmath35 we will suppress the argument @xmath11 when no ambiguity can arise . to avoid messy equations ,
let us label by @xmath36 the derivative with respect to @xmath11 of the exponent in ( [ eq : purport ] ) : @xmath37 for conciseness we have omitted the dependence of @xmath36 on @xmath7 in the notation .
now , by expanding the @xmath38 in ( [ eq : eigen ] ) and using ( [ eq : diff ] ) we obtain the following @xmath39\\ = -a(k ) + { 2 \over 1 - x e^{-2f(x ) } } a(k ) + { 2 x e^{-2f(x ) } \over ( 1 - x e^{-2f(x)})^2 } \left({2i \over \tanh \pi k } - { 4i \over \tanh \pi k } x f'(x ) \right ) .
\end{split } \label{eq : long}\ ] ] we also note that @xmath40 so that ( [ eq : long ] ) becomes @xmath41\\ + { 4ixe^{-2f } \over \left(1-xe^{-2f}\right)^2 } { 1 \over \tanh \pi k } \left(1 - 2xf ' \right ) .
\label{eq : diffeq } \end{split}\ ] ] let us solve this equation .
first off , the terms proportional to @xmath42 must be equal , so that @xmath43 which can be reduced to the form @xmath44 this is solved by @xmath45 setting the remainder of the terms in ( [ eq : diffeq ] ) equal to each other we get @xmath46 now that we ve solved for the @xmath29 and @xmath30 we plug in to get the final form of the solution . integrating by parts we first obtain @xmath47 summing , @xmath48 after some easy algebraic manipulation of the terms above that involve @xmath49 , we get that the total exponent in ( [ eq : purport ] ) is @xmath50 so the purported solution is @xmath51 substituting in @xmath52 for @xmath53 , we obtain @xmath54 .
there are some problems with the above solution .
for one , the square root will have a branch cut , and we will have to decide where it goes .
the most naive choice of the branch cut ( which avoids the real @xmath11 axis altogether ) makes the exponent in ( [ eq : soln ] ) an even function of @xmath11 , and makes the solution odd in @xmath11 .
this means that for both large positive @xmath11 and large negative @xmath11 the wavefunction is outgoing ( or incoming , depending on conventions ) , and that we have two purported independent solutions ( for the two signs of the square root ) .
there is no local conservation so this might not be a problem , though by building wavepackets one might run into problems with unitarity ( have two packets that collide and disappear ) .
it should be noted , however , that this choice of branch cut makes the exponent in the perturbed solution ( [ eq : soln ] ) not converge to that of the unperturbed solution ( [ eq : pert ] ) as @xmath7 goes to minus infinity . instead , from the standpoint of the perturbation expansion it is more natural to choose the branch cut to bisect the real @xmath11 axis at @xmath23 .
also , when computing the first perturbative correction to the phase shift , @xcite integrates around the singularity at @xmath23 ; as successive more singular terms in the expansion pile up at @xmath23 , they fatten up into a branch cut that bisects the real @xmath11 axis at @xmath23 . with this choice of branch cut ,
the function one obtains is not smooth at @xmath23 , and is not odd in @xmath11 .
however , after anti - symmetrization , it becomes odd and smooth at @xmath23 .
in fact , it becomes analytic on the entire complex @xmath11 plane , a fact which we will need in verifying that it indeed satisfies ( [ eq : eigen ] ) . before we prove these things ,
let us summarize .
we choose the branch cut in the wavefunction to go through @xmath23 , in order to have a solution that approaches the unperturbed solution when @xmath7 goes to minus infinity . after anti - symmetrization
we will show that we obtain an analytic wavefunction . by plugging in ,
we independently verify in the next section that it does satisfy ( [ eq : eigen ] ) .
it is the only combination of the two branches that can be verified to satisfy ( [ eq : eigen ] ) using the method in the next section .
so even though there are other branches that give different wavefunctions that are smooth for real @xmath11 ( such as the ones discussed in the first paragraph of this section ) , they can not be verified using the methods of the next section .
we believe that these other branches in fact do not satisfy ( [ eq : eigen ] ) , though we do not prove this uniqueness statement .
let us now proceed systematically .
first , the integrand in the exponent of ( [ eq : soln ] ) has branch points whenever @xmath55 and defines a riemann surface which is a double cover of the complex @xmath11-plane .
what is not so obvious , but nevertheless true , is that the whole solution ( [ eq : soln ] ) is a well defined function on this riemann surface .
let us see why this is so .
first , we need a branch cut structure for @xmath56 - for definiteness , take the one discussed above ( the one that goes through @xmath23 ) . to continue
, it will be useful to relate @xmath57 to @xmath58 .
we have @xmath59 for definiteness take @xmath11 real and positive .
the integral in the exponent is @xmath60 the first term is easily evaluated to be @xmath61 .
the second one is trickier . letting @xmath62 and @xmath63 we see that it is equal to @xmath64 = 5.5 cm with the branch cut of @xmath65 going from @xmath66 to @xmath17 we can take that of the logarithm to run along the positive @xmath67 axis and deform the integration contour away to infinity as shown in figure [ w1 ] .
the two horizontal segments do not quite cancel because of the discontinuity in the logarithm , but together they yield an easily calculable contribution .
likewise , the contribution from large @xmath67 also does not vanish , but is easily calculable .
the integral , in the limit of @xmath68 going to infinity , is then @xmath69 = \\ 2 \pi i + 2 \pi \eh + 2 \log \left ( { \sinh \pi k \over e^{\pi \eh } } + \sqrt{1+{\sinh^2 \pi k \over e^{2 \pi \eh } } } \right ) \end{split}\ ] ] and so @xmath70 as one can straightforwardly verify , this equation relating @xmath71 to its translates by @xmath72 is the essential ingredient showing that ( [ eq : soln ] ) has the same branching structure as @xmath73 and so is well defined on the same riemann surface .
now it turns out that the function of @xmath11 obtained by summing ( [ eq : soln ] ) over the two preimages of each @xmath11 in the riemann surface is precisely the the antisymmetrization discussed at the beginning of this section . specifically , the function in question is @xmath74 with @xmath75where @xmath76 . from its construction as a sum over preimages we immediately see that @xmath77 is analytic everywhere ( the endpoints of the cuts are the only places where one must take a little care ) .
the fact that it solves ( [ eq : eigen ] ) will be checked presently .
we explicitly verify that @xmath77 is a solution .
we multiply ( [ eq : eigen ] ) through by @xmath78 to obtain @xmath79 now , the operator @xmath80 is just translation by @xmath81 when it acts on @xmath77 .
when it acts on @xmath82 it is not well defined because the branch cuts make the expansion in derivatives fail to converge .
nevertheless , we will prove the above equation is true for @xmath82 , where we will interpret @xmath80 to mean translation by @xmath81 .
this will show that the above is true for @xmath77 , because in that case @xmath80 is both well defined and translation by @xmath81 .
let s take @xmath83 for definiteness .
recall that @xmath84 .
we have @xmath85 recall also ( [ eq : f(k ) ] ) that @xmath86 acting on @xmath83 multiplies it by @xmath87 given these two facts , we compute that the right and left hand sides of ( [ eq : tocheck ] ) ( with @xmath77 replaced by @xmath83 ) are , respectively , @xmath88 and @xmath89 inserting the expression for @xmath90 and using elementary algebra shows that these are indeed equal .
first , we examine the large @xmath11 behavior of the exponent of , say , @xmath91 . we denote it by @xmath92 : @xmath93 where as before @xmath94 .
this integral can not be done in closed form , but by differentiating with respect to @xmath7 under the integral sign one does obtain one that can be done through simple integration by parts .
the result is @xmath95_{k_0 \left(\eh \right)}^k.\ ] ] this can be evaluated quite easily if one notices that it is simply half of the integral from @xmath96 to @xmath11 , with the branch cut chosen to go from @xmath97 to @xmath98 and the contour deformed above it .
thus @xmath99 taking the limit of @xmath11 going to infinity we see that @xmath92 approaches @xmath100 , which is starting to look like the conjectured phase shift ( [ eq : phase ] ) .
let us now figure out some asymptotics .
we spare the reader the easy integrals and state only that in the limit @xmath9 we have @xmath101 and in the limit @xmath102 @xmath103 because we know the derivative of @xmath104 with respect to @xmath7 , we can figure out the phase shift for any energy @xmath7 .
we obtain that for large @xmath11 , @xmath105\ ] ] this matches ( [ eq : phase ] ) .
i would like to thank a. frolov , j. maldacena , a. maloney , j. mcgreevy , and s. shenker for useful conversations .
this work was supported by nsf grant 0244728 .
j. maldacena , `` long strings in two dimensional string theory and non - singlets in the matrix model , '' arxiv : hep - th/0503112 .
g. marchesini and e. onofri , `` planar limit for su(n ) symmetric quantum dynamical systems , '' j. math .
* 21 * , 1103 ( 1980 ) . | we solve the non - local eigenvalue problem that arose from consideration of the adjoint sector of the @xmath0 matrix model in @xcite .
we obtain the exact wavefunction and a scattering phase that matches the string theory calculation . 0 |
electron - positron storage rings are widely used to investigate the behaviour of two - photon interactions via the process @xmath6 , where @xmath7 is a virtual photon .
the outgoing electron and positron carry nearly the full beam energy and their transverse momenta are usually so small that they are not detected .
this kind of event is characterised by an initial state @xmath8 , calculable by qed , and a low multiplicity final state .
this process is particularly useful in the study of the properties of hadron resonances .
the total cross section @xmath9 for the formation of a resonance r is given by : @xmath10 where @xmath11 is the differential luminosity function giving the flux of virtual photons .
for quasi - real photons @xmath12r ) is given by the breit - wigner formula : @xmath13 where @xmath14 is the invariant mass of the two - photon system , @xmath15 , @xmath16 , @xmath17 and @xmath18r ) are the mass , spin , two - photon partial width and total width of the resonance , respectively .
combining equations ( [ eq : stot ] ) and ( [ eq : sgg ] ) leads to the proportionality relation @xmath19 where the proportionality factor @xmath20 can be evaluated by a monte carlo integration .
equation ( [ eq : prop ] ) is used to determine the two - photon partial width of the resonance .
the quantum numbers of the resonance must be compatible with the initial state of the two quasi - real photons .
a neutral , unflavoured meson with even charge conjugation and helicity - zero or two can be formed . in order to decay into @xmath21
, the resonance must have @xmath22 . for the @xmath23 , 1@xmath24p@xmath2 tensor meson nonet , the f@xmath2(1270 ) , the a@xmath25(1320 ) and the f@xmath1(1525 )
can be formed . however , since these three states are close in mass , interferences must be taken into account . according to su(3 ) , the f@xmath2(1270 ) interferes constructively with the a@xmath25(1320 ) in the @xmath26 final state but destructively in the @xmath27 final state @xcite .
therefore , among the states of the tensor meson nonet , only the f@xmath1(1525 ) was observed by previous experiments in the @xmath21 final state @xcite .
since gluons do not couple to photons , a pure glueball couples to two photons only via a box diagram and its two photon width is therefore expected to be very small .
a state that can be formed in a gluon rich environment but not in two photon fusion has the typical signature of a glueball .
an analysis of the reaction @xmath28 is presented here , where only the @xmath29 decay is considered .
the data correspond to an integrated luminosity of 143 collected by the l3 detector at lep at @xmath30 gev and 52 at @xmath31 gev .
a study of the @xmath21 final state in two - photon collisions was already published by the l3 collaboration @xcite with a luminosity of 114 at @xmath30 gev .
the l3 experiment is described in detail elsewhere @xcite . in this analysis , the charged particle tracker is mainly used .
it is composed by a silicon microstrip vertex detector and a multiwire drift chamber .
the events are triggered by a low @xmath32 threshold charged - track trigger @xcite .
in order to select @xmath33 events , we require : * the total energy seen in the calorimeters must be smaller than 30 gev to exclude annihilation events .
* there must be exactly four good charged tracks in the tracking chamber with a net charge of zero . a good track requires more than 20 hits out of a maximum of 62 .
* the total momentum imbalance in the transverse plane must satisfy : + @xmath34 . *
events with photons are rejected .
a photon is defined as an isolated shower in the electromagnetic calorimeter with an energy larger than 100 mev .
the @xmath35 s are identified by requiring a secondary vertex distinct from the primary interaction point with a distance greater than 1 mm in the transverse plane . in order to select @xmath21 exclusive events ,
we require : * at least one of the two secondary vertices must be at a distance greater than 3 mm from the interaction point in the transverse plane . * the angle between the flight direction of each @xmath35 candidate ( taken as the line between the interaction point and the secondary vertex in the transverse plane ) and the total transverse momentum vector of the two outgoing tracks must be less than 0.3 rad . *
since the two @xmath35 s are produced back - to - back in the transverse plane , the angle between the flight directions of the two @xmath35 candidates in this plane is required to be @xmath36 rad .
+ fig .
[ fig : k ] shows the @xmath37 mass distribution with a 3 mm vertex cut for the @xmath35 .
fitting this distribution , a mass resolution of @xmath38 mev is found , consistent with the monte carlo simulation .
the invariant masses of the two @xmath35 candidates must be inside a circle of 40 mev radius around the @xmath21 point ( fig .
[ fig : kk ] ) . with these selection criteria ,
253 events are found in the data sample .
the background due to misidentified @xmath35 pairs is estimated to be negligible by a study of the @xmath35 sidebands .
the background due to the @xmath39 final state is determined to be negligible by a monte carlo simulation .
the beam - gas and beam - wall contributions are found negligible . the resulting @xmath21 invariant mass spectrum is shown in fig .
[ fig : kk_sp ] .
the spectrum is dominated by the f@xmath1(1525 ) resonance .
the f@xmath40a@xmath41 region shows the destructive f@xmath42a@xmath25 interference in the @xmath21 final state @xcite . a clear enhancement is visible in the 1750 mev region .
no excess is present around 2230 mev .
a maximum likelihood fit using three gaussians plus a constant is then performed on the @xmath21 mass spectrum .
the masses and widths of the gaussians are left free in the fit .
the fit is shown in fig .
[ fig : kk_sp ] and the results are summarised in table [ tab : li_fit ] .
ccc ' '' '' & f@xmath1(1525 ) & 1750 mev region + mass ( mev ) & 1520 @xmath43 7 & 1770 @xmath43 20 + width ( mev ) & 119 @xmath43 15 & 200 @xmath43 50 + nb . of events & 85 @xmath43 11 & 61 @xmath43 10 + [ tab : li_fit ] the f@xmath1(1525 ) statistical significance is about 8 standard deviations , the statistical significance of the enhancement around 1750 mev is about 6 standard deviations . from the fit 38@xmath437 events
are found in the f@xmath2-a@xmath2 region . in order to correct the data for the detector acceptance and efficiency
, a monte carlo procedure is used @xcite .
the nominal f@xmath1(1525 ) parameters @xcite are used for the generation .
the angular distribution of the two @xmath35 s in the two - photon center - of - mass system is generated according to phase space i.e. uniform in @xmath44 and in @xmath45 , where @xmath46 and @xmath45 are the polar and azimuthal angles taking the @xmath47 direction parallel to the electron beam . in order to take into account the helicity of a spin - two resonance , a weight is assigned to each generated event according to the weight functions : @xmath48 for the helicity - zero contribution and @xmath49 for the helicity - two contribution .
all the events are passed through the full detector simulation program and are reconstructed following the same procedure used for the data .
although the detector acceptance is rather high ( 15% for helicity - zero and 30% for helicity - two ) , the trigger efficiency ( 83% ) and the analysis cuts ( @xmath50 ) give a total efficiency of 3.9% for helicity - zero and 7.6% for helicity - two for the f@xmath1(1525 ) at @xmath30 gev . at @xmath31 gev the total efficiencies are 3.1% and 7.2% respectively .
+ in order to determine the helicity state , a study of the angular distribution of the two @xmath35 s from f@xmath1 decay in the two - photon center of mass is performed .
the experimental polar angle distribution is compared with the monte carlo in fig .
[ fig : ad_f2 ] for both the helicity - zero and helicity - two cases .
the monte carlo distributions are normalised to the same number of events as in the data and no background subtraction is done .
the @xmath51 values for the helicity - zero and helicity - two hypotheses are 81 and 7 for nine degrees of freedom respectively .
thus pure helicity - zero is excluded and helicity - two is largely dominant , in agreement with the theoretical predictions @xcite .
the total cross section @xmath52 times the branching ratio br into @xmath53 for the f@xmath1(1525 ) is measured using the formula @xmath54 where @xmath55 is the integrated luminosity and @xmath56 is the total efficiency .
the number of signal events @xmath57 is determined from the maximum likelihood fit . from our data only @xmath58br(f@xmath59 ) can be measured .
using the pdg @xcite value for br(@xmath29 ) and br(f@xmath60br(f@xmath61 ) from isospin symmerty , @xmath62br(f@xmath63 k@xmath64 ) can be determined .
the product @xmath65(f@xmath66 br(f@xmath67 is measured from the cross section using the formula @xmath68 where the proportionality factor @xmath20 is evaluated by monte carlo integration .
two separate measurements are performed for data collected at @xmath30 gev and @xmath31 gev .
the results are in very good agreement with the published value @xcite @xmath65(f@xmath69br(f@xmath70 under the hypothesis of a pure helicity - two state .
+ the enhancement around 1750 mev may be due to the formation of a radially excited f@xmath71 state , according to theoretical predictions @xcite .
however , the 1750 mev region is very interesting especially looking at the recent results by the crystal barrel collaboration @xcite .
they report the observation of three @xmath72=0@xmath73 states the a@xmath74(1450 ) , the f@xmath74(1370 ) and the f@xmath74(1500 ) . in the scenario presented in
@xcite the a@xmath74(1450 ) is the isovector member of the 0@xmath73 meson nonet while the f@xmath74(1370 ) and the f@xmath74(1500 ) can not both be isoscalar members of the nonet .
the f@xmath74(1370 ) is very likely mainly composed by u and d quarks while the f@xmath74(1500 ) is compatible with the ground state scalar glueball expected around 1500 mev .
this hypothesis includes the prediction of a further scalar state , the f@xmath75(1500 - 1800 ) , mainly composed by s quarks .
this state will couple strongly to k@xmath64 and its discovery is essential to support the f@xmath74(1500 ) glueball nature . in order to investigate the spin of the 1750 mev region ,
the angular distribution of the two @xmath35 s in the two - photon center of mass is studied .
the polar angle distribution is compared in fig .
[ fig : ad_18 ] with the monte carlo predictions for the spin two helicity - zero ( j=2 , @xmath76=0 ) , spin two helicity - two ( j=2 , @xmath76=2 ) and spin zero ( j=0 ) cases .
the monte carlo distributions are normalised to the same number of events as in the data and no background subtraction is done .
the @xmath51 values are 50 , 26 and 7 for nine degrees of freedom for j=2 @xmath76=0 , j=2 @xmath76=2 and j=0 hypotheses respectively .
j=0 assignment is found 3.2 times more probable than j=2 @xmath76=2 .
experimental indications for a scalar state around 1750 mev decaying into k@xmath77k@xmath78 @xcite and @xmath79 @xcite have been recently found .
the bes collaboration confirmed the previous observation by the mark iii collaboration of a resonance , the @xmath3(2230 ) @xcite , produced in radiative decays of the j/@xmath80 particle . due to its narrow width and its production in gluon rich environment
, this state is considered a glueball candidate .
its mass is consistent with the lattice qcd prediction for the ground state tensor glueball .
since gluons do not couple to photons , the two - photon width is expected to be small for a glueball . to make this statement more quantitative ,
a parameter called stickiness @xcite can be introduced for a state x : @xmath81 where @xmath82 is the mass of the state , @xmath83 is the energy of the photon from a radiative j/@xmath80 decay in the j/@xmath80 rest frame and @xmath84 is the angular momentum between the two gluons . for spin two states @xmath85 .
@xmath86 is a normalisation factor that can be calculated assuming by definition the stickiness of the f@xmath2(1270 ) tensor meson to be 1 . using the same method adopted for the f@xmath1(1525 ) , a monte carlo simulation is used to determine the detection efficiency for the @xmath3(2230 ) . for the simulation we use values for the mass and the width determined by combining the results by mark iii and bes
. a mass resolution of @xmath87= 60 mev is found by fitting a gaussian to the @xmath21 monte carlo spectrum .
the total detection efficiency is measured to be 16.5% at @xmath8891 gev and 9.3% at @xmath88183 gev under the hypothesis of a pure helicity - two contribution .
the signal region is chosen to be @xmath89 around the @xmath3(2230 ) mass . in order to evaluate the background two sidebands of @xmath90
are considered . at @xmath8891 gev and at @xmath88183 gev , 6 and 2 events are found in the signal region respectively . fitting a constant in the sideband region ,
the expected background are evaluated to be 4.9 and 7.9 events . using the standard method @xcite for extracting an upper limit for a poisson distribution with background
, we determine upper limits of 7.3 and 3.6 events respectively at 95% c.l . using the same method adopted for measuring the two - photon width of the f@xmath1(1525 ) , an upper limit of @xmath91br@xmath92
is derived combining the results from the two values of @xmath93 .
the combination is performed using a monte carlo technique .
we obtain @xmath91br@xmath94 ev at 95% c.l . under the hypothesis of a pure helicity - two state . combining the results reported by bes and mark iii for the @xmath95br@xmath96 and our upper limit on @xmath97br@xmath96 ,
we obtain a lower limit on the stickiness @xmath9833 at 95% c.l .
errors are taken into account by a monte carlo calculation .
this value is much larger than the values measured for all the established @xmath99 states .
the reaction @xmath100 is studied with the l3 detector at lep .
the spectrum is dominated by the f@xmath101(1525 ) tensor meson .
the angular distribution excludes pure helicity - zero and shows that helicity - two is largely dominant .
the @xmath102 is found to be consistent with the previously published value @xcite .
clear evidence for destructive f@xmath2-a@xmath2 interference is observed .
an enhancement of about 6 standard deviations is observed around 1750 mev .
it may be due to the formation of a radial recurrence of the f@xmath101(1525 ) tensor meson ( j=2 , @xmath1032 ) but the observed angular distribution favours the j=0 assignment .
therefore the enhancement may be due to the formation of the @xmath4 member of the @xmath5 meson nonet .
the latter case strongly supports the glueball nature of the f@xmath74(1500 ) @xcite .
no signal is present in the region of the @xmath3(2230 ) glueball candidate .
we obtain the upper limit @xmath91br@xmath94 ev at 95% c.l . under the hypothesis of a pure helicity - two state .
i would like to express my gratitude to the two - photon physics group of the l3 collaboration in particular to m.n .
focacci - kienzle , j.h .
field and b. monteleoni . | the k@xmath0k@xmath0 final state in two - photon collisions is studied with the l3 detector at lep using data at centre of mass energies from 91 gev to 183 gev .
the k@xmath0k@xmath0 mass spectrum is dominated by the formation of the f@xmath1(1525 ) tensor meson whose two - photon partial width is measured .
clear evidence for destructive f@xmath2-a@xmath2 interference is observed .
no signal is present in the region around 2.2 gev .
an upper limit for the two - photon partial width times the k@xmath0k@xmath0 branching ratio of the @xmath3(2230 ) glueball candidate is then derived .
an enhancement is observed around 1750 mev .
it may be due to the formation of a radial recurrence of the f@xmath1(1525 ) or to the @xmath4 member of the @xmath5 meson nonet . |
mode - locked trains of pulses , or frequency combs , have at the same time the coherence properties of c.w
. lasers and the high peak powers of pulsed lasers .
they are potentially perfect tools for generating non - classical states of light , as they are are at the same time high quality light sources , free of excess noise , and intense sources able to induce strong nonlinear effects , and therefore to generate strongly non - classical states of light , such as squeezed or quadrature - entangled states .
frequency combs have been used in many quantum optics experiments , and have efficiently produced non classical light , either in @xmath0 grangier , manipspopo6,serkland or @xmath1 watanabe , rosenbluh , friberg , schmitt , krylov , hirosawa , sharping , fiorentino media , but so far in a single - pass configuration in the nonlinear medium . in this configuration , one needs very high peak powers , and the system loses somehow its potential high quality in terms of pulse to pulse coherence and transverse profile . mode locked lasers have also been used to efficiently generate squeezed states in optical fibers @xcite .
the system has the drawback of generating non - minimal states states of light , because of the excess noise due to brillouin scattering in the fiber .
we have recently proposed @xcite to use synchronous optical cavities to recirculate the light in the nonlinear medium , thus enhancing further the nonlinear effects and imposing the cavity mode structure to the generated non - classical field .
this can be done by building synchronously pumped opos " or spopos . in a spopo
the cavity round - trip time is equal to the delay between successive pulses of the pumping mode - locked laser , so that the effect of the successive intense pump pulses add coherently , thus reducing considerably its oscillation threshold . in @xcite ,
a large squeezing effect was predicted in some supermodes " , which are well defined linear combinations of signal modes of different frequency , but not studied in detail . in the time domain these supermodes correspond to trains of pulses of different waveforms , orthogonal each other .
the purpose of this paper is to precise the quantum model used to predict the effects and to investigate in a detailed way , through analytical or numerical methods , the potentialities of the system in realistic situations spopos have already been implemented as efficient sources of tunable ultra - short pulses manipspopo0,manipspopo1,manipspopo2,manipspopo3,manipspopo4,manipspopo5 their temporal properties have been theoretically investigated theoryspopo1,theoryspopo3,theoryspopo4 , and actively mode - locking of opos has been recently achieved @xcite .
the decomposition of a pulsed field in terms of a basis of normal modes , similar to the supermodes we consider in this paper , has been introduced in different contexts for a complete quantum characterization of either the pulsed squeezed light generated by parametric down conversion @xcite or solitons in optical fibers @xcite .
such approaches are strongly connected with the schmidt decomposition of two - photon states for the characterization of pairwise entanglement huang1996,law2000 and the bloch - messiah reduction of any optical circuit characterized by a linear input - output relation @xcite . in this context menicucci _
@xcite proposed optical frequency combs as scalable resources for quantum computation . the article is organized as follows : we present first the model that we will use .
the system turns out to be characterized by a real and symmetric matrix @xmath2 , which contains all the information about the effective nonlinear interaction .
the eigensystem of @xmath2 is thus of special relevance and is studied in section iii , where the spopo threshold and several general properties of the spectrum of @xmath2 are addressed .
an analytical approximation to the diagonalization is also given that allows a better insight into the general trends as parameters are varied .
in section iv it is shown that the introduced eigenmodes or supermodes are squeezed , the corresponding eigenvalues determining the amount of squeezing , which can be measured in a balanced homodyne detection experiment that uses as the local oscillator ( lo ) a field with the same spectrum as the desired supermode . in section
v one then studies the squeezing properties of spopos in two realistic cases , corresponding to bibo and knbo@xmath3 crystals , using an appropriate scaling property of the diagonalization problem .
finally , an appendix details the case of the singly resonant spopo .
we consider quasi - degenerate collinear type i interaction , by means of which the pumping frequency comb , at frequencies around @xmath4 , is converted by a nonlinear @xmath5 crystal into multimode signal radiation at frequencies around @xmath6 , and vice - versa , where @xmath7 and @xmath6 are the two frequencies at which perfect phase matching occurs .
this implies that one has @xmath8 , where @xmath9 is the crystal refractive index .
the nonlinear crystal is placed inside a high finesse optical cavity of length @xmath10 , which is assumed to be dispersion compensated by intracavity dispersive elements , so that all cavity modes around the frequency @xmath6 are equally spaced by a common free spectral range @xmath11 , which is made equal to that of the pumping laser , thus warranting the synchronization of the pump to the opo cavity .
this ensures that the pulse - to - pulse delay of the pump beam coincides with the cavity round - trip time and successive pump and signal pulses superpose in time , thus maximizing the strength of the interaction .
hence the external pump mean field , which is a phase - locked multimode coherent field , can be written as @xmath12@xmath13 is the average laser irradiance ( power per unit area ) , @xmath14 is the normalized ( @xmath15 ) complex spectral component of longitudinal mode labeled by the integer index @xmath16 , and @xmath17 corresponds to the phase - matched mode . as we will be concerned with femtosecond lasers with pulse durations around @xmath18fs , the number of pump modes will be typically on the order of @xmath19 .
two possibilities for pumping can be used : either ( i ) the pump also resonates inside the cavity ( doubly resonant case ) , which requires in addition dispersion compensation at the pump spectral region , or ( ii ) the cavity is transparent for the pump ( singly resonant case ) , a case which is free from the previous restriction and thus more amenable for experimentation , at the expense of a higher threshold , as we will see .
we detail more the latter case in the appendix at the end of this paper .
we will limit here our analysis to non - chirped pumps as chirping requires a more general treatment which will be presented elsewhere . as the finesse of the cavity
is assumed to be high , the intracavity signal field operator @xmath20 can be written as a superposition of cavity modes . inside the @xmath5 crystal , which extends from @xmath21 to @xmath22 , one can write @xmath23where @xmath24 , @xmath25 is the annihilation operator for the @xmath16-th signal mode in the interaction picture , verifying standard boson commutation relations@xmath26 = \delta _ { m , n } , \label{boson_s}\]]@xmath27 is the spatial profile of mode @xmath16 , equal to @xmath28 in the case of ring cavities , while for linear cavities it is equal to @xmath29 $ ] , where @xmath30is the corresponding wavenumber .
finally @xmath31 is the single photon field amplitude , whose value depends on the type of cavity . for a ring cavity
@xmath32where @xmath33 is the transverse area of the signal field , while for a linear cavity @xmath34note that we are writing the field as a superposition of plane waves , but the treatment is approximately valid for gaussian beams provided that the crystal is placed at the beam waist and the rayleigh length is much longer than the crystal length @xmath35 . in this case
@xmath36 with @xmath37 the beam radius .
similarly we have @xmath38 for the pump transverse mode .
the interaction hamiltonian @xmath39 describing the parametric interaction in the nonlinear crystal is given as usual by : @xmath40 , \label{hidef}\]]where @xmath41 and @xmath42 are the nonlinear electric polarization at signal and pump frequencies , and @xmath43 accounts for the effective area of interaction corresponding to the three - mode overlapping integral across the transverse plane and , for gaussian beams , it is given by @xmath44 .
the calculation of the hamiltonian depends on the type of configuration ( singly- or doubly resonant ) . here
we consider the simpler case of a doubly resonant spopo and leave the details of the singly resonant case for the appendix . in this case
an expression for the intracavity pump field operator @xmath45 analogous to ( [ defes ] ) , now centered around @xmath4 , can be used and the following expression for @xmath39 in the rotating wave approximation is obtained : @xmath46where @xmath47 is the relevant nonlinear susceptibility , @xmath48 and @xmath49 are pump boson operators ( @xmath17 denotes the phase matched mode ) verifying @xmath50 = \delta _ { m , n}$ ] , and @xmath51 is as @xmath31 with the substitutions @xmath52 and @xmath53 .
the phase - mismatching factor of the crystal , @xmath54 , is given by : @xmath55@xmath56 being the phase - mismatch angle : @xmath57 making use of the standard input - output formalism of optical cavities the following set of heisenberg equations for the signal annihilation operators @xmath58 is derived straightforwardly : @xmath59where the cavity damping rate @xmath60 , or cavity linewidth , is equal to @xmath61 , @xmath62 is the transmission factor of the single cavity mirror at which losses are assumed to be concentrated , the coupling constant @xmath63 is given by @xmath64and @xmath65 is a factor depending on the geometry of the cavity that amounts to @xmath66 for the ring cavity and to @xmath67 for the linear cavity .
analogously , the evolution equations for the pump annihilation operators @xmath68 are : @xmath69where @xmath70 is the cavity damping coefficient evaluated at pump frequencies . to get these simple equations , we have assumed that @xmath71 @xmath72 , and neglected the dispersion of the nonlinear susceptibility , which is a very good approximation as far as the pulses bandwidth is not too large @xcite .
the " operators correspond to quantum fields entering the cavity through the coupling mirror .
we consider the case where the input signal field is the vacuum , and the input pump field a coherent state .
we have therefore @xmath73 , @xmath74 , and the following correlations @xmath75with the notation @xmath76 , the rest of correlations being null .
the mean input field @xmath77 is related to the @xmath78 and @xmath13 coefficients introduced in by : @xmath79 as detailed in the appendix , in the case where only the signal field is resonating into the cavity the evolution of the signal field annihilation operators is given by an expression identical to ( [ dsdr ] ) , but now the pump annihilation operators are given by @xmath80 we see that these operators contain two contributions .
the first one corresponds to the free - field part that is described by means of two independent boson operators @xmath81 associated to the fields impinging the cavity from both the directions labeled with the superscripts @xmath82 .
we consider the unidirectional pumping case where the input pump field propagating from left to right ( labeled with @xmath83 ) is a coherent state with a mean value of @xmath84 , while the input pump field propagating form right to left ( and labeled with @xmath85 ) is the vacuum @xmath86 .
the mean input field , @xmath87 , is still given by eq .
( [ pext ] ) , and the only non - null correlations are : @xmath88 ^{\dag } \right\rangle = \delta _ { m , n}\,\delta \left ( t - t^{\prime } \right ) .\ ] ] below threshold signal modes have a zero mean value , whereas the pump field is characterized by a huge amplitude .
one can therefore use a linearization procedure for the quantum fluctuations , which amounts to setting @xmath89 in ( [ dsdr ] ) . in the doubly resonant case @xmath90 , as given by ( [ pump ] ) , while in the singly resonant case @xmath91 , as given by ( [ srpump ] ) . in both cases
the final equations for the signal field anihilation operators are identical : @xmath92where @xmath93is a pump amplitude parameter , and @xmath94 is an important scaling parameter for the pump , which can be shown to be the threshold value for the pump in the c.w . regime ( single mode pump configuration ) . in the optimized configuration for the pump focussing ( @xmath95 ) ,
it is equal to : @xmath96 .coefficient for retrieving the cw threshold @xmath94 in the different experimental situations considered . [
cols="^,^,^",options="header " , ] the fact that they are close to degeneracy ( see fig . [ caseceigenvalues ] ) is reflected in an almost equal reduction of noise variances below the standard quantum limit . despite the differences reported between the variances evaluated both by means of a perfectly mode matched and a gauss - hermite lo , a still significative degree of squeezing can be detected in realistic situations thus suggesting that the shaping of the lo is not a critical issue for the experimental configuration considered in this section
actually , there is a larger set of supermodes the variances of which are all close to the value of the critical one . in particular , there are about @xmath97 supermodes that have variances comprised in @xmath98 db , between @xmath99 db and @xmath100
furthermore , by considering the number of supermodes that present a noise reduction bigger than @xmath101 db , one discovers that , this time , their number amounts to about @xmath102 .
we have seen that even if the cases a and b do not verify at the same time the condition , the analytical solution obtained in the gaussian approximation works quite well in both cases .
nevertheless , in spite of the fact that this condition gives an approximately good idea of the reliability of gaussian approximation , it is interesting to study the passage from a perfectly gaussian case to a non - gaussian one trough a gray " region where the differences between the two cases are not big .
let us consider eq . in the limit of very large @xmath35 . in such case
, since from eq .
@xmath103 , then @xmath104 asymptotically converges to : @xmath105 is constant for values of @xmath35 compatible with a non - gaussian regime . in figure lambda0 l vs l ,
then , we report the values of this product as a function of the crystal thickness . for @xmath106 mm the analytical solution , as expected , is in good agreement with the numerical one , while for greater thicknesses the discrepancy is significative .
also this result is expected , since the analytical solution for the critical eigenvalue has not validity when the condition is violated . on the other side ,
the fact that also the analytical solution reaches asymptotically , for increasing @xmath35 , a plateau suggests a @xmath107-like behavior of @xmath108 .
the existence of such a plateau can be explained from the point of view of the evolution of the pump and signal pulses in the time domain . as discussed in the previous section , when the condition is violated , the walk - off between pump and signal is bigger than the pump temporal width . as a consequence ,
the exchange of energy between the two fields is disadvantaged till a point where the threshold can not change anymore even increasing the crystal length . since , from eqs . and ,
@xmath109 , for large values of @xmath35 , then , also the product @xmath110 reach a constant value , thus explaining the plateau in fig .
[ lambda0 l vs l ] . versus the crystal length.,scaledwidth=40.0% ] in the same way
, one can explain the discrepancy observed between the two plateaux in fig .
[ lambda0 l vs l ] .
as already discussed in case c , in non - gaussian configurations the pump bandwidth is larger than the phase - matching one , thus not all the pump modes are phase matched and not all of them contribute to the final value of the threshold . therefore the number of pump modes actually involved is smaller than the nominal value @xmath111 that should , then , be corrected . from a quantum point of view
, we have detailed in the previous sections the noise properties of the supermodes connected to the first four highest @xmath112 ( see tables [ vars09a ] , [ vars09b ] and [ vars09c ] ) and we calculated the number of supermodes showing a squeezing better than @xmath101 db for getting a qualitative indication about the multimodicity " of the system prepared in a specific experimental configuration .
these results can be appreciated in fig .
[ vkmin_vs_l ] where the noise variances of the supermodes that satisfy this criterium have been traced for the three cases previously discussed .
the curve in the middle corresponds to case a ( @xmath113 mm ) , a gaussian configuration . as the thickness of the crystal
is increased to @xmath114 mm ( case b ) the condition is violated but the gaussian approximation is still good .
this means that , even if the value of @xmath115 is decreasing ( because we are reducing the number of pump modes that are phase - matched ) the spectrum is still a geometric progression but with a smaller , in absolute value , common ratio ( see @xmath116 in eq . ) thus causing an overall decrease of the spectrum with respect to the case a. the consequence is a reduction of the number of supermodes with a squeezing greater than @xmath101 db as it results from the upper curve in fig .
[ vkmin_vs_l ] ( red circles ) . finally ,
when the crystal length is further increased to @xmath117 mm ( case c ) , we pass to a completely non - gaussian configuration where the decrease of the critical eigenvalue @xmath115 ( from @xmath118 for @xmath113 mm to @xmath119 for @xmath117 mm ) causes a significative deformation of the spectrum and a non null set of eigenvalues flatten around @xmath108 . in this case , since the degree of squeezing for each supermode depends on the ratio @xmath120 ( see eq . ) , the amount of squeezing is globally increased as the lower curve ( green diamonds ) in fig . [ vkmin_vs_l ] shows .
these results not only confirm that a spopo is a highly multi - mode device but also show another important quality : the malleability for controlling its multimodicity " .
we have seen , in fact , that one can just increase the thickness of the nonlinear crystal in order to improve the number of supermodes that play an important role from a quantum point of view . in this paper
we have presented a study of the eigenvalues and eigenvectors of the matrix @xmath2 in function of the crystal length @xmath121 . however , since both the pump pulse width and the crystal length are present in eq . ,
notice that effects similar to those discussed in this section can be observed by playing with @xmath122 and keeping constant @xmath35 .
in fact the passage from a gaussian to a non - gaussian configuration takes place when the pump bandwidth @xmath123 becomes smaller than the phase - matching one @xmath124 and , clearly , this can be obtained keeping fixed the latter and increasing @xmath122 ( see fig .
[ legenda ] ) .
eventually one could even fix @xmath35 and @xmath125 and play with the group velocity mismatch by choosing different types of nonlinearities , but , experimentally , this could result in a stiffer malleability of the device .
in this paper , we have shown that both the dynamical and quantum noise properties of the spopo depend on the spectrum of eigenvalues @xmath126 of the linear coupling matrix @xmath2 .
we have studied in detail this spectrum in various experimentally feasible configurations , using either an analytical approach in some simple limit cases , or a numerical approach in the general case .
it turns out that among the roughly 100,000 eigenvalues , 99,900 or so are zero , and about a hundred are significantly different from zero and contribute to the quantum dynamics of the system .
spopos are therefore devices which produce simultaneously many highly squeezed vacuum modes .
this property can be used to improve the performances of metrological methods using frequency combs , for example to perform ultra - accurate time transfer between remote clocks beyond the shot noise limit @xcite .
in addition , it is well known that if one mixes by one way or another different squeezed modes , one gets strongly entangled states @xcite .
this is also the case here : we will show in a forthcoming publication that spopos are indeed likely to generate various pairs of strongly entangled supermodes , as well as multipartite entangled states .
more generally , this paper is an example of the fact that , by using appropriately chosen pump spectra and phase matching curves , one can reach various eigenvalue spectra , and therefore tailor at will the quantum properties of the light generated by the optical system , which can be useful for example to generate interesting states for multidimensional quantum information processing .
we detail in this appendix the derivation of the heisenberg equations in the case of a singly resonant degenerate type i spopo , i.e. when the signal field resonates inside the cavity , but not the pump field .
we consider here the case of the linear fabry - perot cavity .
the treatment of this case is more complex than the doubly resonant case usually considered in theoretical approaches as the pump field can not be quantized inside the cavity .
the signal field inside the nonlinear crystal , which extends from @xmath21 to @xmath22 , is written as @xmath127where @xmath128 , and @xmath129 .
on the contrary , the pump field is not affected by the cavity and hence it is given by a continuum of modes .
we shall use the common approach of quantizing the pump field in a line of length @xmath130 with periodic boundary conditions and , in the end of the calculations , we will make @xmath131 .
we thus write @xmath132 + \nonumber\\ & + & \mathrm{h.c.},\end{aligned}\]]where the superscripts @xmath133 label the propagation direction .
@xmath134 are single photon field amplitudes . in the limit
@xmath135 the frequencies @xmath136 are given by @xmath137 , @xmath138 , and the wavenumbers @xmath139 , with @xmath140 the crystal refractive index .
note that we are writing the fields as a superposition of plane waves , but the treatment is still approximately valid for gaussian beams as far as the thin crystal is placed at the ( common ) beam waist of pump and signal , and the rayleigh lengths are much longer than the crystal length @xmath35 . in such case @xmath141 with @xmath142 the corresponding beam radius .
the interaction hamiltonian @xmath39 is calculated as usual as in .
inserting the expressions of the second order nonlinear electric polarizations , one obtains the following form of the interaction hamiltonian , in which @xmath143 is the second order nonlinear susceptibility ( whose dispersion is neglected ) : @xmath144 \times \notag \\ & \times & e^{i\left ( \omega_{\mathrm{s},m}+\omega_{\mathrm{s}% , q}-\nu_{j}\right)t } e^{-i\left(k_{\mathrm{s},m}+k_{\mathrm{s},q}\right)l/2}+ \notag \\ & + & \mathrm{h.c.,}\end{aligned}\ ] ] where we defined the phase - mismatch factor@xmath145 } { \left ( \kappa_{j}-k_{\mathrm{s},m}-k_{% \mathrm{s},q}\right ) \frac{l}{2}}.\ ] ] in ( [ hilinear ] ) we dropped highly phase mismatched terms , as usual .
if we introduce new signal boson operators @xmath146 the interaction hamiltonian becomes as ( [ hilinear ] ) but without the exponential @xmath147 . in the following we will use the new operators but omit the superscript @xmath148 " for simplicity . from the previous expression of the hamiltonian , one can derive following heisenberg equations governing the time evolution of the pump and signal operators : @xmath149\times \notag \\ & \times & e^{i\left(\omega_{\mathrm{s},m}+\omega_{\mathrm{s}% , q}-\nu_{j}\right)t } .
\label{s}\end{aligned}\ ] ] using the usual wigner - weisskopf approach , valid because the nonlinear interaction is assumed to be instantaneous , and using the approximation @xmath152 we obtain a value of @xmath153 that one can then insert in eq . :
@xmath154 we can now pass to the continuum limit . for that we define continuum pump operators in the following way : @xmath155 which verify @xmath156 ^{\dag}\right ] = \frac{l_{\mathrm{p}}}{2\pi c}\delta_{j , k}\underset{% l_{\mathrm{p}}\rightarrow\infty}{\longrightarrow}\delta\left ( \nu _ { j}-\nu_{k}\right ) .\ ] ] transforming sums into integrals , we obtain the following equation for the signal modes : @xmath157 where @xmath158 } { \left(k\left(\nu\right)-k_{% \mathrm{s},m}-k_{\mathrm{s},q}\right)\frac{l}{2 } } \times \notag \\ & \times & \left [ \hat{p}_{\mathrm{free}}^{\left(+\right)}\left(\nu\right)+% \hat{p}_{\mathrm{free}}^{\left(-\right)}\left(\nu\right ) \right ] e^{i\left(\omega_{\mathrm{s},m}+\omega_{\mathrm{s},q}-\nu\right)t } , \\ i_{m , q}^{\left(2\right)}&= & \int\mathrm{d}\nu\delta \left(\omega_{\mathrm{s}% , n}+\omega_{\mathrm{s},r}-\nu\right)\frac{\nu}{n\left(\nu\right ) } \times \notag \\ & \times & \frac{\sin\left[\left(k\left(\nu\right)-k_{\mathrm{s},m}-k_{\mathrm{% s},q}\right)\frac{l}{2}\right ] } { \left(k\left ( \nu\right)-k_{\mathrm{s}% , m}-k_{\mathrm{s},q}\right)\frac{l}{2 } } \times \notag \\ & \times & \frac{\sin\left[\left(k\left(\nu\right)-k_{\mathrm{s},n}-k_{\mathrm{% s},r}\right)\frac{l}{2}\right ] } { \left(k\left(\nu\right)-k_{\mathrm{s},n}-k_{% \mathrm{s},r}\right)\frac{l}{2 } } \times \notag \\ & \times & e^{i\left(\omega_{\mathrm{s},m}+\omega_{\mathrm{s},q}-\nu\right)t}.\end{aligned}\ ] ] in order to calculate the first integral , we write it as a sum over frequency intervals of width @xmath159 and centered at frequencies @xmath160 , @xmath161 .
thus @xmath162 becomes @xmath163 } { \left(k_{\mathrm{p},r}-k_{\mathrm{s},m}-k_{% \mathrm{s},q}\right)\frac{l}{2 } } \times \notag \\ & \times & \int\nolimits_{r}\mathrm{d}\nu \left[\hat{p}_{\mathrm{free}% } ^{\left(+\right)}\left(\nu\right)+\hat{p}_{\mathrm{free}}^{\left(-\right)}% \left(\nu\right)\right ] e^{-i\nu t}\end{aligned}\ ] ] where @xmath164 .
we now define new pump operators @xmath165 which can be shown to verify the following property : @xmath166 ^{\dag}\right\rangle \simeq\delta _ { r_{1},r_{2}}\delta\left ( t_{1}-t_{2}\right ) .
\label{corr - p}\ ] ] @xmath167 now becomes @xmath168% } { \left(k_{\mathrm{p},r}-k_{\mathrm{s},m}-k_{\mathrm{s},q}\right)\frac{l}{2}% } \times \notag \\ & \times & e^{i\left(\omega_{\mathrm{s},m}+\omega_{\mathrm{s},q}-\omega_{% \mathrm{p},r}\right)t } \left [ \hat{p}_{\mathrm{in},r}^{\left(+\right)}% \left(t\right)+\hat{p}_{\mathrm{in},r}^{\left(-\right)}\left(t\right ) \right]% .\end{aligned}\ ] ] retaining only slowly varying terms in the evolution , and including the losses of the optical cavity at rate @xmath169 , one finally gets : @xmath170 + \notag \\ & - & g^{2}\sum_{n}\sum\limits_{q}f_{m , q}f_{n , m+q - n}\times \notag \\ & \times & \hat{s}_{q}^{\dag}\left(t\right)\hat{s}_{n}\left(t\right)\hat{s}% _
{ m+q - n}\left(t\right ) , \label{dsmlan}\end{aligned}\ ] ] where the coupling constant @xmath171 is given by @xmath172 and @xmath173 corresponds to the field at signal frequencies entering the cavity through the coupling mirror .
when that input is coherent or vacuum , the case we consider , those in " operators verify the following correlation@xmath174 and thus behave as @xmath175 ( see eq([corr - p ] ) ) .
let us now consider the regime below the oscillation threshold : the signal modes are almost not excited and the double sum in ( [ dsmlan ] ) can be neglected .
also , the pump in " fields can be approximated by their mean values as their fluctuation part gives rise to smaller terms , which are neglected for the same reasons as before .
hence , we have for a unidirectional pumping : @xmath176 @xmath13 being the average power per unit area of the modelocked pump laser and @xmath177 .
( [ pinres ] ) is obtained by demanding that the pump field corresponding to the set @xmath178 equals the external pump field given by eq .
( [ eext ] ) inside the crystal .
the linearized equations for the spopo below threshold finally become@xmath179 where @xmath180 in conclusion , we have shown that the linearized equations ( [ dsdtsingle ] ) formally coincide with those of a doubly resonant spopo ( eq.(dsdtdouble ) ) , the only difference being the exact value of @xmath94 .
laboratoire kastler brossel , of the ecole normale suprieure and the universit pierre et marie curie
paris6 , is umr8552 of the centre national de la recherche scientifique .
this work was partially supported by the french - spanish programme partenariats hubert curien``-programa de acciones integradas '' ( projet picasso 13663pc - accin integrada hf2006 - 0018 ) .
gp , nt and cf acknowledge the financial support of the future and emerging technologies ( fet ) programme within the seventh framework programme for research of the european commission , under the fet - open grant agreement hideas , number fp7-ict-221906 .
gjdev acknowledges financial support by grant best/2007/161 of the generalitat valenciana and by projects fis2005 - 07931-c03 - 01 and fis2008 - 06024-c03 - 01 of the spanish ministerio de educacin y ciencia and the european union feder .
r. m. shelby , m. d. levenson , s. h. perlmutter , r. g. devoe , and d. f. walls , phys .
lett . * 57 * , 691 ( 1986 ) .
g. j. de valcrcel , g. patera , n. treps , and c. fabre , phys .
a * 74 * , 061801(r ) ( 2006 ) . | quantum models for synchronously pumped type i optical parametric oscillators ( spopo ) are presented .
the study of the dynamics of spopos , which typically involves millions of coupled signal longitudinal modes , is significantly simplified when one considers the supermodes " , which are independent linear superpositions of all the signal modes diagonalizing the parametric interaction . in terms of these supermodes
the spopo dynamics becomes that of about a hundred of independent , single mode degenerate opos , each of them being a squeezer .
one derives a general expression for the squeezing spectrum measured in a balanced homodyne detection experiment , valid for any temporal shape of the local oscillator .
realistic cases are then studied using both analytical and numerical methods : the oscillation threshold is derived , and the spectral and temporal shapes of the squeezed supermodes are characterized . |
layered compounds of transition - metal elements frequently show interesting and novel electrical and magnetic properties and have been studied extensively . among them , la@xmath0cuo@xmath2 @xcite and lafeaso @xcite as the representatives are widely known because of their superconductivity with high critical temperatures after doping or under high pressures , in which the cuo@xmath0 square planar layers and the feas tetrahedral layers play a substantial role respectively .
now another class of layered compounds built from alternatively stacking special octahedral layers m@xmath0q@xmath0o ( m = fe or ti ; q = as , sb , s , or se ) with otherwise layers are attracting much attention.@xcite here an m@xmath0q@xmath0o layer includes an anti - cuo@xmath0-type m@xmath0o square planar layer with two q atoms respectively located above and below the center of each m@xmath0o square unit . as shown in fig .
[ figa ] , the m@xmath0q@xmath0o layer is in an octahedral layer structure , bridging or a combination of the cuo@xmath0 square planar layer and the feas tetrahedral layer structures , which is thus expected to show distinct electronic and magnetic properties and possibly superconductivity by doping or under pressure .
fe@xmath0q@xmath0o layer - based ( q = s or se ) oxychalcogenides were first synthesized in 1992 @xcite and recently characterized to be antiferromagnetic insulators with a magnetic moment of @xmath3 on each fe ion experimentally@xcite and with strong correlation effects suggested theoretically.@xcite to our knowledge , there has been no any structural distortion found experimentally in these compounds .
ti@xmath0q@xmath0o layer - based ( q = as or sb ) oxypnictides have been also studied for a quite long time .
compounds na@xmath0ti@xmath0q@xmath0o were first synthesized in 1990,@xcite which subsequently have been intensively investigated both experimentally and theoretically.@xcite the temperature - dependent powder neutron diffraction measurement showed that there is a structural distortion in the ti@xmath0sb@xmath0o layer at a temperature of about 120k,@xcite which corresponds well to the observed anomalies in the temperature - dependent magnetic susceptibility and resistivity measurements.@xcite however , no neutron scattering due to magnetic spin ordering was detected .
meanwhile , the measurement further found that the bond distance ratio of o - ti - o / sb - ti - sb increases below the transition temperature , but the crystal space group symmetry remains unchanged after the structural distortion.@xcite this is different from the tetragonal - orthorhombic structural distortions observed in the iron pnictides.@xcite recently another new ti@xmath0as@xmath0o layer - based oxypnictide bati@xmath0as@xmath0o was successfully synthesized,@xcite which shows the similar characteristics as na@xmath0ti@xmath0q@xmath0o , namely there are anomalies observed as well in the temperature - dependent magnetic susceptibility , resistivity , and heat capacity measurements at a temperature of about 200k.@xcite considering that the similar anomalies are observed in the iron pnictides , in which the superconductivity takes place when the anomaly is suppressed , one thus speculates that the ti@xmath0q@xmath0o layer - based oxypnictides may also be another parent compounds for the high t@xmath4 superconductivity . in the iron pnictides , those anomalies are ascribed to the tetragonal - orthorhombic structural transitions accompanied by the collinear antiferromagnetic ordering.@xcite the underlying mechanism is still in debate , either the as - bridged antiferromagnetic superexchange interactions between fluctuating fe local moments @xcite or spin density wave instability induced by the fermi surface nesting.@xcite there are now more and more evidences in favor of the fluctuating fe local moment picture .
especially , the neutron inelastic scattering experiments have shown that the low - energy magnetic excitations can be well described by the spin waves based on the quantum heisenberg model.@xcite likewise , the scenario in ti@xmath0q@xmath0o layer - based oxypnictides is very confusing .
the experimentally observed anomalies suggest that there is an antiferromagnetic transition , but no magnetic ordering has been detected .
theoretically , no ordered magnetic state has been either found in the first - principles electronic structure calculations,@xcite which however show that there is a fermi surface nesting .
it was thus suggested that the observed anomalies would be due to charge density wave or spin density wave instability induced by such a fermi surface nesting.@xcite in order to clarify the issue , we performed the thorough first - principles electronic structure calculations upon compounds na@xmath0ti@xmath0q@xmath0o ( q = as , sb ) .
we find that the ground state of na@xmath0ti@xmath0as@xmath0o is a blocked checkerboard antiferromagnetic semiconductor , in contrast , the ground state of na@xmath0ti@xmath0sb@xmath0o is a bi - collinear antiferromagnetic semimetal . and
we verify that these two antiferromagnetic states both are indeed induced by the fermi surface nesting rather than based on local magnetic moments .
in our calculations the plane wave basis method was used.@xcite we used the pw91-type@xcite generalized gradient approximation for the exchange - correlation functionals .
the ultrasoft pseudopotentials @xcite were used to model the electron - ion interactions . after the full convergence test , the kinetic energy cut - off and the charge density cut - off of the plane wave basis were chosen to be 600ev and 4800ev , respectively .
the gaussian broadening technique was used and a mesh of @xmath5 k - points were sampled for the brillouin - zone integration in nonmagnetic state .
the lattice parameters with the atomic positions were fully optimized by the energy minimization .
it was found that the calculated lattice parameters have an excellent agreement with the experimental ones ( less than @xmath6 ) .
ti@xmath0as@xmath0o : ( a ) a conventional unit cell composed of two formula cells ; ( b ) a primitive unit cell composed of one formula cell.,width=226 ] compound na@xmath0ti@xmath0pn@xmath0o ( pn = as or sb ) crystallizes in tetragonal ( i4/mmm ) symmetry and has a layered structure , as shown in fig . [ figa](a ) .
its conventional tetragonal unit cell consists of two formula unit cells .
however , its primitive unit cell is constructed by considering na@xmath0ti@xmath0pn@xmath0o as a triclinic crystal , in which only one formula unit cell is included as shown in fig . [ figa](b ) .
on the other hand , na@xmath0ti@xmath0pn@xmath0o can be considered geometrically with an anti - k@xmath0nif@xmath2 type structure , where a ti atom is located between two oxygen atoms in a plane to form a square planar layer of ti@xmath0o , which is an anti - configuration to the cuo@xmath0 layer in the high @xmath7 cuperates .
meanwhile a ti atom is also four - coordinated by as atoms , forming a ti@xmath0as@xmath0o layer where two as atoms are respectively located above and below the center of a ti@xmath0o square unit .
the crystal is formed by stacking such ti@xmath0pn@xmath0o layers in a body - centered manner along the @xmath8 axis with the separation by a double - layer of na atoms ( fig . [ figa](a ) ) .
we first studied the nonmagnetic state of compound na@xmath0ti@xmath0pn@xmath0o ( pn = as or sb ) , which would describe the high temperature phase of these materials .
the electronic band structure of this nonmagnetic state also provides a reference to further studying of the possible low temperature magnetic phases , by analyzing which we can better understand the underlying mechanism or interactions that drive possible magnetic phase transitions and the related structural transitions . in the calculations
, we adopted a primitive unit cell including only one formula cell ( fig .
[ figa](b ) ) .
ti@xmath0as@xmath0o in the nonmagnetic state : ( a ) electronic band structure ; ( b ) hole - type fermi surface sheets ; ( c ) electron - type fermi surface sheets .
the fermi energy is set to zero.,width=264 ] after the full structural optimization , we find the tetragonal crystal lattice parameters of @xmath9 and @xmath10 for na@xmath0ti@xmath0as@xmath0o , in good agreement with the experimental values of @xmath11 and @xmath12 measured at a temperature of 310k .
@xcite the calculated electronic band structure and fermi surface in the primitive unit cell are presented in fig .
as we see , there are three fermi surface sheets , contributed from the two bands crossing the fermi energy level ( fig .
[ figb](a ) ) . among them
, the one sheet is of the hole - type , forming an approximate square - shaped pipe centered at each @xmath13 point in the brillouin zone ( fig .
[ figb](b ) ) ; the other two are of the electron - type , one of them is a @xmath14-centered box with the windows at its top and bottom , the other is a @xmath15-centered perfectly fluted pipe around the corner of the brillouin zone ( fig . [ figb](c ) ) .
the volumes enclosed by these fermi surface sheets are 0.186 electrons / cell and 0.186 holes / cell respectively .
the electron carrier concentration is the same as the hole carrier concentration .
both are equal to @xmath16 .
compound na@xmath0ti@xmath0as@xmath0o is thus a semimetal with a low carrier concentration between the ones in normal metals and semiconductors , similar to the recent iron pnictides , for example , lafeaso and bafe@xmath0as@xmath0.@xcite the density of states at the fermi energy is 3.84 states per ev per formula unit .
the corresponding electronic specific heat coefficient is @xmath17 = @xmath18 and the pauli paramagnetic susceptibility is @xmath19=@xmath20 ( 1 mole formula unit ) , in reasonable agreement with the measured value of @xmath21.@xcite ti@xmath0as@xmath0o per formula cell in the nonmagnetic state .
the fermi energy is set to zero.,width=264 ] inspection of the distribution of the density of states will help to clarify the atomic bonding character .
figure [ figc ] plots the calculated total and atomic orbital - resolved partial density of states of na@xmath0ti@xmath0as@xmath0o in the nonmagnetic state respectively . as expected ,
the calculations show that most electrons of the na @xmath22-orbitals transfer into the o @xmath23-orbitals so that the occupied o @xmath23-orbitals are far below the fermi energy level while the empty na @xmath22-orbitals are much above the fermi energy level ( centered at + 5.0 ev ) .
although the density of states at the fermi level is dominated by the ti @xmath24-orbitals , the density of states around the fermi energy ( from -3.0 to 2.0ev ) consist mainly of both the ti @xmath24-orbitals and as @xmath25-orbitals .
especially , the two peaks in the distribution of the as @xmath25-orbitals coincide with those of the ti @xmath24-orbitals , as specifically shown in fig .
[ figd ] , which indicates that there is the hybridization between @xmath26 ( @xmath27 ) orbitals and @xmath28 ( @xmath29 ) orbitals , namely there is covalent bond formed between them . figure [ figd ] also specifies that it is the remaining three @xmath24-orbitals ( @xmath29 , @xmath30 , and @xmath31 ) that dominate in the density of states at the fermi energy .
ti@xmath0as@xmath0o in the nonmagnetic state projected onto ti-@xmath24 atomic orbitals ( upper panel ) , as-@xmath25 atomic orbitals and o-@xmath23 atomic orbitals ( lower panel ) , respectively.,width=264 ] ti@xmath0sb@xmath0o in the nonmagnetic state : ( a ) electronic band structure ; ( b ) hole - type fermi surface sheets ; ( c ) electron - type fermi surface sheets .
the fermi energy is set to zero.,width=264 ] likewise , we performed the calculations on na@xmath0ti@xmath0sb@xmath0o .
the optimized tetragonal crystal lattice parameters are found to be of @xmath32 and @xmath33 , also in good agreement with the experimental values of @xmath34 and @xmath35 measured at a temperature of 150k.@xcite figure [ figf ] shows the electronic band structure and fermi surface calculated in a primitive unit cell , similar to those of na@xmath0ti@xmath0as@xmath0o . from the volumes enclosed by these fermi sheets , we determine that the electron and hole concentrations are 0.140 electrons / cell and 0.140 holes / cell respectively .
the corresponding electron ( or hole ) carrier density is about @xmath36 .
the density of states at the fermi energy is 4.40 states per ev per formula unit .
the corresponding electronic specific heat coefficient and pauli paramagnetic susceptibility are @xmath17 = @xmath37 and @xmath19=@xmath38 ( 1 mole formula unit ) , respectively .
the latter is in excellent agreement with the experimental measurement on a high - quality single crystal ( @xmath39).@xcite ti@xmath0sb@xmath0o per formula cell in the nonmagnetic state .
the fermi energy is set to zero.,width=264 ] in fig .
[ figg ] , we plot the calculated total and atomic orbital - resolved partial density of states for na@xmath0ti@xmath0sb@xmath0o in the nonmagnetic state , which are overall similar to the ones of na@xmath0ti@xmath0as@xmath0o ( fig .
[ figc ] ) except that the sb @xmath40-orbitals have more contribution to the density of states at the fermi energy than the as @xmath25-orbitals do .
this is easily understandable because the sb @xmath40-orbitals are more diffusive and extended than the as @xmath25-orbitals .
nevertheless , the atomic bonding character of na@xmath0ti@xmath0sb@xmath0o is basically the same as the one of na@xmath0ti@xmath0as@xmath0o , i.e. o and na atoms are in ionic state while there are covalent bonds formed between ti and sb atoms . in order to well analyze the fermi surface symmetry of compounds na@xmath0ti@xmath0pn@xmath0o ( pn = as , sb ) in connection with the crystal symmetry , we replot the calculated fermi surfaces in the tetragonal conventional brillouin zone as shown in fig . [ fige ] , which clearly reflects the two dimensional character of the electronic structures by the nearly @xmath41-dispersionless fermi surface sheets , being consistent with the layered crystal structure . as we see , the fermi surfaces in these two compounds are very similar in shape to each other .
there are now six fermi surface sheets for each fermi surface , among which there are the two nearly degenerate square - box - like fermi - surface sheets centered around x contributed from the hole - type bands in conjugation with another two similar fermi surface sheets centered around m but from the electron - type bands .
obviously these two sets of fermi surface sheets between x and m are strongly nested with each other .
the corresponding nesting vector is @xmath42 or @xmath43 .
the early studies reported in refs . and first suggested that such a fermi surface nesting in na@xmath0ti@xmath0sb@xmath0o would likely lead to a charge- or spin - density wave instability in association with the observed anomalies in the temperature - dependent susceptibility and resistivity.@xcite ti@xmath0as@xmath0o in the nonmagnetic state : ( a ) and ( b ) are respectively the hole - type and electron - type fermi surface sheets , and ( c ) is the top view of the whole fermi surface in the brillouin zone denoted by the dotted square ; na@xmath0ti@xmath0sb@xmath0o in the nonmagnetic state : ( d ) and ( e ) are respectively the hole - type and electron - type fermi surface sheets , and ( f ) is the top view of the whole fermi surface in the brillouin zone denoted by the dotted square.,width=283 ] the powder neutron diffractions have not yet detected any magnetic spin ordering in compounds na@xmath0ti@xmath0pn@xmath0o ( pn = as , sb ) even though the observed anomalous transitions in the temperature - dependent magnetic susceptibility and electric resistivity strongly suggested an antiferromagnetic transition in the compounds .
nevertheless the neutron diffractions indeed found a commensurate structural distortion but without the space group symmetry breakdown .
right now we still can not exclude any weak magnetic ordering just based on the powder neutron diffraction because of its low resolution .
the issue that we meet here is still whether or not there exists a magnetic long - range order related to the structural distortion , based on either local moments formed around ti atoms or spin - density wave induced by the fermi surface nesting ; otherwise , whether or not there exists a spin - unpolarized charge - density wave related to the structural distortion , induced still by the fermi surface nesting . in order to clarify this issue ,
we have carried out the systematical and extensive calculations on compounds na@xmath0ti@xmath0pn@xmath0o by constructing a wide variety of unit cells and magnetic orders .
we first performed the spin - unpolarized calculations with different unit cells and possible distortions , especially by elaborately constructing new unit cells in consideration of the nesting vector between x and m points in the brillouin zone .
it turns out that all the distortions and structures different from the original one are unstable .
we thus exclude spin - unpolarized charge - density waves in compounds na@xmath0ti@xmath0pn@xmath0o .
we then investigated whether or not any magnetic order exists , based on local moments around ti atoms .
if there are local moments formed around ti atoms , the exchange interactions between these local moments would be either pn - bridged or o - bridged superexchange interaction with the direct exchange interaction between the nearest neighbor ti atoms .
there would be a variety of stable or metastable magnetic orders resulting from the competition among these exchange interactions , such as the ferromagnetic , checkerboard antiferromagnetic , and collinear antiferromagnetic orders , all of which were found to be stable or metatable in fe - based superconducting parent compounds.@xcite our spin - polarized calculations show that none of these magnetic orders is stable or metastable in compounds na@xmath0ti@xmath0pn@xmath0o , similar to the previous calculations that show no magnetic ordering found in na@xmath0ti@xmath0sb@xmath0o.@xcite we thus exclude magnetic orders based on local moments . to the end
, we came to study possible magnetic orders based on spin - density waves induced by the fermi surface nesting .
as presented above , there are two independent and orthogonal but equivalent nesting vectors @xmath44 and @xmath45 , respectively .
a possible spin density wave will oscillate as @xmath46 on the ti - ti square lattice ( @xmath47 being a lattice site vector and @xmath48 magnetization vector ) , with a wave vector @xmath49 equal to a commensurate linear combination of @xmath50 and @xmath51 , namely @xmath52 ( @xmath53 and @xmath54 being integers ) .
we elaborately constructed several kinds of magnetic orders with magnetic unit cells based on such wave vectors @xmath49 . through the calculations , we eventually find two stable peculiar magnetic orders with wave vectors @xmath55 and @xmath50 ( or @xmath51 ) respectively , called as blocked checkerboard antiferromagnetic order and bi - collinear antiferromagnetic order respectively .
pn@xmath0o ( pn = as , sb ) layer .
the large blue circles represent the projected sites of pn atom into the ti@xmath0o plane , the dark olive circles represent the o atoms , the small black circles with arrows represent ti atoms and the green and red arrows indicate the two opposite directions of ti moments .
the blocked checkerboard antiferromagnetic order ( a ) and the bi - collinear antiferromagnetic order ( b ) are shown . here a small blue dashed square represents an @xmath56 unit in ( a ) and ( b ) .
the large blue dashed square @xmath57 unit in ( a ) , and the blue dashed rectangle @xmath58 unit in ( b ) correspond to a magnetic unit cell in the blocked checkerboard antiferromagnetic order and bi - collinear antiferromagnetic order respectively . for the blocked checkerboard antiferromagnetic order ,
the ti@xmath0pn@xmath0o layer is divided into square blocks denoted by gray dash - dotted lines .
the four ti moments in a block are in the same direction and those between two nearest neighbor blocks have opposite directions.,width=283 ] figure [ fig8-two - order](a ) schematically shows the blocked checkerboard antiferromagnetic order , in which the ti@xmath0pn@xmath0o ( pn = as , sb ) layer is divided into @xmath59-squares represented by the dash - dotted thin lines .
each square is a magnetic block with the moments of four ti atoms being parallel . between the nearest neighbor blocks , the block moments are in anti - parallel .
but the magnetic unit cell is a @xmath60 square .
it should be emphasized that in each magnetic block the four ti moments are centered around a pn atom rather than o atom . for the latter case ,
the calculations show that it is unstable or energetically unfavorable .
thus in the blocked checkerboard anitferromagnetic order any pair of the next nearest neighbor ti moments bridged by an o atom is always in anti - parallel .
this is also consistent with the o - bridged superexchange effect being more stronger than the pn - bridged superexchange effect , even though this magnetic order is induced by the fermi surface nesting . in fig .
[ fig8-two - order](b ) , we schematically show the bi - collinear antiferromagnetic order , in which the ti moments align in parallel along a diagonal direction and in anti - parallel along the other diagonal direction on the ti - ti square lattice .
in other words , if the ti - ti square lattice is divided into two square sublattices @xmath61 and @xmath62 , the ti moments on each sublattice take their own collinear antiferromagnetic order . here
the magnetic unit cell is a @xmath63 rectangle . unlike in the blocked checkerboard antiferromagnetic order , here half of pairs of the next nearest neighbor ti moments bridged by o atoms are in parallel while the other half are in anti - parallel , likewise with pn atoms . for compounds na@xmath0ti@xmath0pn@xmath0o the moment around each ti atom
is found to be about @xmath64 in both blocked checkerboard and bi - collinear antiferromagnetic orders .
this is consistent with the spin density wave mechanism , in which the moment should be small since its formation is only attributed to those states nearby the fermi energy .
considering there is a large amount of antiferromagnetic quantum fluctuation in low dimensional systems , the effective ordering moment around each ti atom will be further much reduced .
this is likely the reason why no magnetic ordering in compounds na@xmath0ti@xmath0pn@xmath0o has so far been detected by the powder neutron diffraction.@xcite a bi - collinear antiferromagnetic order was first introduced in the study of fe - based superconductors to describe the ground state of @xmath65-fete,@xcite very interestingly as a strong evidence to against the view of spin density waves in the fe - based superconducting parent compounds , since there is no any fermi surface nesting vector to correspond to such a magnetic order , moreover there is a large magnetic moment of about 2.5@xmath1 formed on each fe atom .
the substantial difference between the two cases lies in the fact that there is only one electron in 3@xmath66-orbital for a ti@xmath67 ion , in contrast there are five electrons in 3@xmath66-orbital for an fe@xmath68 ion .
accordingly it is not expected that there is strong electronic correlation effect in na@xmath0ti@xmath0pn@xmath0o .
we have performed a gga+@xmath69 calculation .
we find that when @xmath69 is larger than 2 ev , the ferromagnetic state becomes the lowest energy state for both compounds .
this contradicts the experimental observation that there is an antiferromagnetic - like transition around 300 k for na@xmath0ti@xmath0as@xmath0o and around 120 k for na@xmath0ti@xmath0sb@xmath0o.@xcite as aforementioned , the calculations show that any magnetic state , except the blocked checkerboard and bi - collinear antiferromagnetic states , is unstable and will relax into a nonmagnetic state for compound na@xmath0ti@xmath0as@xmath0o .
ti@xmath0as@xmath0o in the blocked checkerboard antiferromagnetic state : ( a ) band structure ; ( b ) magnetic unit cell ; ( c ) brillouin zone with selected high symmetric points .
here the top of the valence band sets to zero.,width=283 ] to calculate the blocked checkerboard antiferromagnetic ( afm ) order in na@xmath0ti@xmath0as@xmath0o , we construct such a crystal unit cell consisting of four formula cells , which is triclinic with three base vectors of ( 2@xmath70 , 0 , 0 ) , ( 0 , 2@xmath70 , 0 ) , and ( 0.5@xmath70 , 0.5@xmath70 , 0.5@xmath8 ) respectively , as shown in fig .
[ fig9](b ) . for the bi - collinear antiferromagnetic order ,
the crystal unit cell is also constructed as a triclinic cell but with three base vectors of ( 2@xmath70 , 0 , 0 ) , ( 0 , @xmath70 , 0 ) , and ( 0.5@xmath70 , 0.5@xmath70 ,
0.5@xmath8 ) respectively ( see fig . [
fig11](b ) ) .
ti@xmath0as@xmath0o in the bi - collinear antiferromagnetic state : ( a ) band structure ; ( b ) magnetic unit cell ; ( c ) brillouin zone with selected high symmetric points . here the top of the valence band sets to zero.,width=283 ] after the full structural optimization , we find that compound na@xmath0ti@xmath0as@xmath0o in the blocked checkerboard afm order is energetically lower by about 4.5 mev / ti than in the bi - collinear afm order .
figures [ fig9 ] and [ fig11 ] show the electronic band structures of compound na@xmath0ti@xmath0as@xmath0o in the blocked checkerboard afm and bi - collinear afm orders respectively . in both afm orders
, na@xmath0ti@xmath0as@xmath0o is transformed into a semiconductor from a semimetal , with a small band gap of about 0.15 and 0.05 ev respectively ; moreover the moment around each ti atom is found to be 0.56 and 0.53 @xmath1 respectively .
these calculated results are summarized in table [ table1 ] .
.calculated energies and magnetic moments of the nonmagnetic , ferromagnetic , checkerboard , collinear , bi - collinear , and blocked checkerboard antiferromagnetic states respectively , for na@xmath0ti@xmath0as@xmath0o and na@xmath0ti@xmath0sb@xmath0o .
the energy in the nonmagnetic state is set to zero .
the bar " in table means that the corresponding magnetic state is not stable and will relax into a nonmagnetic state .
[ cols="<,^,^,^,^",options="header " , ]
we have presented the first - principles calculations of the electronic structures of layered pnictide - oxide na@xmath0ti@xmath0pn@xmath0o ( pn = as , sb ) .
we find that the ground state of na@xmath0ti@xmath0as@xmath0o is a blocked checkerboard antiferromagnetic semiconductor with a small gap of about 0.15 ev , while the ground state of na@xmath0ti@xmath0sb@xmath0o is a bi - collinear antiferromagnetic semimetal ; and both have a small moment of about 0.5@xmath1 around each ti atom .
moreover we find that there is a tetramer structural distortion in the blocked checkerboard antiferromagnetic state , in which the four ti atoms at each magnetic block gather towards their center of an as atom , in good agreement with the experimental observation of a commensurate structural distortion but the space group symmetry retained after the anomaly happening .
we confirm that there is a strong fermi surface nesting in na@xmath0ti@xmath0pn@xmath0o .
and we verify that the two antiferromagnetic states both are induced by the fermi surface nesting rather than based on local magnetic moments .
actually these two antiferromagnetic states are close in energy for each of na@xmath0ti@xmath0pn@xmath0o , in which one is the ground state , then the other is the metastable state .
we may consider layered pnictide - oxides na@xmath0ti@xmath0pn@xmath0o as a paradigm for spin density waves .
we wish to thank professor xian - hui chen for bringing the layered pnictide - oxides to our notice .
this work is supported by national program for basic research of most of china ( grant no .
2011cba00112 ) and national natural science foundation of china ( grant nos .
11190024 and 91121008 )
. _ note added . _ while in the preparation of this manuscript , we learnt of a paper by david singh ( arxiv1209.4668 ) , which reports that the instability at @xmath15 point associated with the fermi surface nesting leads to a double striped antiferromagnetic ( namely bi - collinear afm ) state in compound bati@xmath0sb@xmath0o .
we have subsequently examined a variety of magnetic states for compounds bati@xmath0pn@xmath0o ( pn = as , sb ) and find that none of them is stable except the blocked checkerboard and bi - collinear antiferromagnetic states , which are however nearly degenerated with the nonmagnetic state with a lower energy of no more than 0.5 mev / ti and the moment around each ti atom is just about 0.2 @xmath1 .
the fermi surfaces that we obtained for bati@xmath0pn@xmath0o are similar to those shown in figs . 6 and 7
reported in the arxiv1209.4668 paper , in which the fermi surface nesting clearly is dramatically weaker than those in compound na@xmath0ti@xmath0pn@xmath0o reported in the present paper ( see fig .
[ fige ] ) .
j.d . bednorz and k.a .
muller , z. phys .
b : condensed matter * 64 * , 189 ( 1986 ) ; m.k .
ashburn , c.j .
torng , p.h .
hor , p.l .
meng , l. gao , z.j .
huang , y.q .
wang , and c.w .
chu , phys .
lett , * 58 * , 908 ( 1987 ) .
j. zhao , d. x. yao , s. li , t. hong , y. chen , s. chang , w. ratcliff , ii , j. w. lynn , h. a. mook , g. f. chen , j. l. luo , n. l. wang , e. w. carlson , j. hu , and p. dai , phys .
* 101 * , 167203 ( 2008 ) . | from the first - principles calculations , we have studied the electronic and magnetic structures of compound na@xmath0ti@xmath0pn@xmath0o ( pn = as or sb ) .
we find that in the ground state na@xmath0ti@xmath0as@xmath0o is a blocked checkerboard antiferromagnetic semiconductor with a small band gap of about 0.15 ev , in contrast , na@xmath0ti@xmath0sb@xmath0o is a bi - collinear antiferromagnetic semimetal , both with a small moment of about 0.5@xmath1 around each ti atom . we show that there is a strong fermi surface nesting in na@xmath0ti@xmath0pn@xmath0o .
and we verify that the blocked checkerboard and bi - collinear antiferromagnetic states both are the spin density waves induced by the fermi surface nesting .
a tetramer structural distortion is found in company with the formation of a blocked checkerboard antiferromagnetic order , in good agreement with the experimentally observed commensurate structural distortion but with space group symmetry retained after the anomaly happening .
further analysis and discussion in connection with experimental observations are given as well . |
in this paper , we study spectral properties of the schrdinger operator @xmath3 defined for @xmath4 in the half - line @xmath5 $ ] . here
@xmath6 is the semiclassical parameter and @xmath7 is a piecewise continuous real - valued potential supported in @xmath8 $ ] .
the operator @xmath9 with the neumann boundary condition at @xmath10 is self - adjoint on @xmath11 ; therefore , its resolvent @xmath12 is a bounded operator from @xmath13 to @xmath14 for @xmath15 not in the spectrum of @xmath9 .
this resolvent can be extended meromorphically as an operator @xmath16 to @xmath17 with isolated poles of finite rank ; these poles are called * resonances*. ( the reader is referred to @xcite for details . ) to each resonance @xmath18 corresponds a * resonant state * ; that is , a nonzero @xmath19 solving the equation @xmath20 with the neumann boundary condition at the right endpoint and with the following * outgoing condition * at @xmath21 : @xmath22 ( note that for @xmath23 , @xmath24 solves the free equation @xmath25 , so it must be a linear combination of @xmath26 . ) for @xmath27 , the outgoing condition implies that @xmath24 is exponentially decreasing on the negative half - line and thus @xmath28 ; therefore , @xmath18 is a pole ( of the resolvent ) lying in the upper half - plane if and only if @xmath15 is an eigenvalue of @xmath9 on @xmath13 . since @xmath9 is self - adjoint ,
all poles in the upper half - plane have to lie on the imaginary axis . there may be poles @xmath18 with @xmath29 and @xmath30 ; however , we will restrict our attention to purely imaginary resonances : a positive number @xmath31 is called a * bound state * if @xmath32 is a pole of the resolvent @xmath33 , and an * antibound state * if @xmath34 is a pole .
we see from above that @xmath31 is an ( anti)bound state if and only if there exists a nonzero solution @xmath24 of the problem @xmath35 the plus sign in ( [ e : leftcondo ] ) corresponds to an antibound state and the minus sign corresponds to a bound state .
we will also study neumann eigenvalues of @xmath9 on @xmath8 $ ] ; i.e. , those @xmath31 for which there exists a nonzero solution to ( [ e : equation ] ) with boundary conditions ( [ e : rightcond ] ) and @xmath36 since the space of solutions to ( [ e : equation ] ) and ( [ e : rightcond ] ) is always one dimensional , * bound states , antibound states , and neumann eigenvalues never coincide*. however , bindel and zworski proved in @xcite that bound and antibound states located away from zero coincide modulo errors of order @xmath37 for some @xmath38 , if the potential satisfies the following conditions : @xmath39,\\ \exists \varepsilon>0:v(x)=0\text { for all } x\in ( a , a+\varepsilon ) .
\end{gathered}\ ] ] in this paper , we prove a similar result with more general assumptions on the potential : [ l : main ] suppose that @xmath1 is a piecewise continuous real - valued potential supported in @xmath8 $ ] and satisfying the following * bump condition * : @xmath40.\ ] ] fix two constants @xmath41
. then there exist constants @xmath42 such that for @xmath43 and any @xmath44 $ ] : \1 . if @xmath31 is a neumann eigenvalue , then there exist a bound state @xmath45 and an antibound state @xmath46 such that @xmath47 .
if @xmath31 is a bound or an antibound state , then there exists a neumann eigenvalue @xmath48 such that @xmath49 . ,
[ -2 , -1.5 , -1 , 0 , 1 , 1.5 , 2 ] ) ` and ` splinepot([0 , 0.2 , -1 , -0.2 , -1 , 0.2 , 0 ] , [ -2 , 1.5 , -1 , 0 , 1 , 1.5 , 2 ] ) ` ) ] the bump condition ( [ e : bump ] ) can not be disposed of completely , as illustrated by the numerical experiments performed using @xcite .
figure 1 shows two potentials on the whole line , each supported in @xmath50 $ ] , and the corresponding bound states ( denoted by squares ) and antibound states ( denoted by circles ) .
the vertical coordinate of each ( anti)bound state on the picture corresponds to its value @xmath31 ; the horizontal coordinate corresponds to the value of @xmath51 used .
we see that the conclusion of the theorem does not appear to hold for the potential on the left , which does not satify the bump condition ; at the same time , it is true for the potential on the right .
theorem 1 , formulated for the half - line case , applies to these numerical experiments on the whole line since for even potentials , the set of their ( anti)bound states is composed of these states for the positive half - line with dirichlet condition and the same states for the neumann condition ; the theorem above can be applied with dirichlet condition in place of ( [ e : rightcond ] ) .
( however , condition ( [ e : leftcondn ] ) can not be replaced by dirichlet condition in the theorem . )
the study of resonances in one dimension has a long tradition going back to origins of quantum mechanics , see for instance @xcite .
one of the first studies of their distribution was conducted by regge @xcite ; since then , there have been many mathematical results on the topic , including @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , @xcite , and @xcite .
concerning antibound states , hitrik has shown in @xcite that for a positive compactly supported potential , there are no antibound states in the semiclassical limit .
this agrees with our result since there are no bound states in this case .
simon proved in @xcite that between any two bound states , there must be an odd number of antibound states ; the following corollary of this result follows almost immediately using the methods we develop to prove theorem 1 : [ l : simon ] consider the half - line problem with a bounded compactly supported potential @xmath1 ( which does not need to satisfy any positivity condition ) .
then for each two bound states @xmath52 , the interval @xmath53 contains at least one antibound state . in particular , if there are @xmath54 bound states in some subinterval of @xmath55 , then there are at least @xmath56 antibound states in the same subinterval
. the proof of theorem 1 works as follows : we study the evolution ( in @xmath4 ) of the vectors @xmath57 for the three solutions of ( [ e : equation ] ) with initial data at @xmath58 satisfying the conditions ( [ e : leftcondo ] ) and ( [ e : leftcondn ] ) .
the idea is to look at these three vectors at @xmath59 . since @xmath60 on the interval @xmath61 , the transition matrix for the considered vectors from @xmath58 to @xmath59 will have an expanding and a contracting direction .
( in fact , if we introduce rescaling @xmath62 , then the behavior of the original system for small @xmath63 is similar to the behavior of the rescaled system for large @xmath64 , and the latter will be similar to the behavior of the geodesic flow on a two - dimensional manifold of negative curvature . )
it turns out that our three vectors lie in a certain angle between the expanding and the contracting directions , from which it follows that they will stay in this angle for later times ( lemma [ l : estimate ] ) ; what is more , their polar angles will get exponentially close to each other ( lemma [ l : close ] ) . finally , we can study how the polar angles of the considered vectors change with @xmath31 ( lemma [ l : angle ] ) : it follows ( lemma [ l : nondegenerate ] ) that the polar angle for the solution with neumann initial data at @xmath58 will strictly increase in @xmath31 and the polar angle for the solution with the same data at @xmath65 will decrease in @xmath31 .
the proof is then completed by a pertrubation argument ( lemma [ l : perturb ] ) .
the detailed proofs of theorems [ l : main ] and [ l : simon ] are given in section 3 .
both are elementary and use certain properties of ordinary differential equations presented in section 2 .
the authors would like to thank maciej zworski for suggesting the problem and for many illuminating discussions .
throughout this section , @xmath66 is an interval in @xmath67 , @xmath68 , @xmath69 , @xmath6 , and @xmath70 .
any solution to the equation @xmath71 is determined by the vector @xmath57 at any @xmath4 ; we will sometimes view this vector in polar coordinates : [ d : langle ] define the * length * @xmath72 and the * polar angle * @xmath73 by the equations @xmath74 here @xmath73 lies in the circle @xmath75 .
[ l : wronskian ] define the * wronskian * @xmath76 by @xmath77 then @xmath78 note that the @xmath76 is just the oriented area of the parallelogram spanned by the vectors @xmath57 and @xmath79 .
the next lemma tells us that if the vector @xmath57 falls inside a certain angle in the plane at the initial time , then it will stay inside that angle for all later times : [ l : estimate ] suppose that @xmath80 for all @xmath81 and some constants @xmath82 .
let @xmath24 be a solution to @xmath71 and define @xmath83 let @xmath84 be a point in @xmath66 and assume that @xmath85 at @xmath84 .
then for @xmath86 , the functions @xmath87 are increasing in @xmath4 and @xmath88 we have @xmath89 therefore , @xmath85 yields @xmath90 and thus ( [ e : ugeq ] ) .
next , @xmath91 using ( [ e : wronskian ] ) , we see that @xmath92 as long as @xmath93 .
it remains to prove that @xmath94 for @xmath86 .
suppose this is false and let @xmath95 .
then @xmath24 is not identically zero ; since it solves a second order linear ode , @xmath96 everywhere . but
@xmath93 on @xmath97 $ ] , so @xmath98 are increasing on this interval .
in particular , @xmath99 at @xmath100 and thus ( [ e : ugeq ] ) holds at this point .
however , by the choice of @xmath100 we have @xmath101 , which contradicts @xmath96 . in the next section
, we will use the following crude estimate on how fast the solutions of an ode can grow : [ l : crude ] assume that @xmath102 for @xmath81 and that @xmath24 is a solution to @xmath71 .
take @xmath103 ; then @xmath104 without loss of generality we may assume that @xmath105 .
we have @xmath106 ; thus @xmath107 and the lemma is proven by gronwall s inequality .
[ l : angle ] assume that @xmath108 is a family of solutions to @xmath109 , @xmath103 , and @xmath110 and @xmath111 are independent of @xmath31 .
let @xmath112 , @xmath113 .
then @xmath114 we have @xmath115 .
( to see that , differentiate the formulas in definition [ d : langle ] in @xmath31 and use the definition of the wronskian . )
now , we differentiate the equation @xmath109 in @xmath31 to get @xmath116 .
it remains to apply ( [ e : wronskian ] ) together with @xmath117 .
[ l : perturb ] assume that @xmath118 is a @xmath119 map from the interval @xmath120 $ ] to the circle @xmath121 and @xmath122 for all @xmath123 .
suppose that @xmath124 is a continuous map such that @xmath125 for all @xmath31 .
put @xmath126 and @xmath127 $ ] . then : \1 .
if @xmath128 has @xmath129 , then there exists @xmath130 with @xmath131 and @xmath132 .
if @xmath133 has @xmath131 , then there exists @xmath134 with @xmath129 and @xmath132
. we can lift @xmath118 and @xmath135 to continuous maps @xmath136 ; then @xmath137 and @xmath138 for @xmath139 .
we have @xmath140 for some @xmath141 .
then @xmath142 and @xmath143 ; it remains to apply the intermediate value theorem .
similar to the previous statement , we have @xmath144 for some @xmath141 and @xmath145 .
[ l : calc ] assume that @xmath118 is a @xmath119 map from some interval @xmath66 to the circle @xmath146 with @xmath147 for all @xmath148 .
let @xmath124 be a continuous map such that @xmath149 for all @xmath148 .
if @xmath150 are two roots of the equation @xmath151 , then the interval @xmath53 contains at least one root of the equation @xmath152 . as in the previous lemma , lift @xmath118 and @xmath135 to maps @xmath136 ; we can make @xmath153 for all @xmath148 . since @xmath154 everywhere
, we have @xmath155 , where @xmath156 are some integers .
therefore , @xmath157 and @xmath158 ; it remains to apply the intermediate value theorem .
we assume in this section that @xmath159 for some constants @xmath160 and @xmath161 ; the constants in our estimates will depend on @xmath162 and @xmath163 .
( we need to expand the interval @xmath164 $ ] a little bit to be able to apply lemma [ l : perturb ] . )
consider the solutions @xmath165 to the equation ( [ e : equation ] ) in @xmath8 $ ] with the initial data @xmath166 define @xmath167 , @xmath168 , and @xmath169 to be the polar angles of vectors @xmath57 at @xmath59 for @xmath170
. then @xmath171 is * a neumann eigenvalue if @xmath172 and @xmath173 are linearly dependent ; that is , ( recalling that they solve the same second order ode ) if @xmath174 ; * a bound state if @xmath175 ; * an antibound state if @xmath176 .
we first prove lemma [ l : close ] .
put @xmath186}v(x)$ ] , @xmath187 , @xmath188 , @xmath189 , and consider the wronskians @xmath190 these are nonnegative for @xmath191 at @xmath58 . then by lemma [ l : estimate ] , all these six functions are nonnegative and increasing in @xmath4 for @xmath192 .
our first goal is to get an exponential lower bound on the length @xmath72 for @xmath191 at @xmath59 . for @xmath172 , note that by ( [ e : wrondef ] ) @xmath193 same applies to @xmath194 .
however , @xmath195 needs more careful analysis since @xmath196 at @xmath58 . for that , take @xmath197 and put @xmath198}v(x)>0 $ ] , @xmath199 , @xmath200 , and @xmath201 .
first , we have by lemma [ l : crude ] @xmath202 next , @xmath203 and @xmath204 , so by ( [ e : ugeq ] ) @xmath205 finally , we apply lemma [ l : estimate ] on the interval @xmath206 $ ] to get @xmath207 for @xmath208 small enough and all @xmath31 , @xmath209 , so we have @xmath210 the next step is to use that @xmath172 and @xmath211 solve the same equation ( [ e : equation ] ) and thus @xmath212 is constant in @xmath4 .
therefore , at @xmath59 we have by ( [ e : wrondef ] ) @xmath213 that finishes the proof of lemma [ l : close ] . to prove lemma [ l : nondegenerate ] , first note that by lemma [ l : angle ] , @xmath214 and @xmath215 by ( [ e : ugeq ] ) , @xmath216 .
also , by lemma [ l : crude ] , @xmath217 for @xmath192 ; thus @xmath218 and lemma [ l : nondegenerate ] is proven , which finishes the proof of theorem [ l : main ] . to prove theorem [ l : simon ] , let @xmath219 ; a bound state corresponds to @xmath220 and an antibound state corresponds to @xmath221 .
since @xmath222 is increasing with @xmath31 , by an argument similar to the proof of lemma [ l : angle ] we get @xmath223 for all @xmath31 .
moreover , @xmath224 is never zero , as this would correspond to @xmath194 and @xmath195 being linearly dependent .
we may now apply lemma [ l : calc ] with @xmath225 and @xmath226 . | we consider the schrdinger operator @xmath0 on a half - line , where @xmath1 is a compactly supported potential which is positive near the endpoint of its support .
we prove that the eigenvalues and the purely imaginary resonances are symmetric up to an error @xmath2 . |
a record is an entry in a series of events that exceeds all previous entries . in recent years
there has been a surge of interest in the statistics of record - breaking events , both from the theoretical point of view as well as in multiple applications .
the occurrence of record - breaking events has been studied for instance in sports @xcite , in evolution models in biology @xcite , in the theory of spin - glasses @xcite and in models of growing networks@xcite .
recently there has been some progress in understanding the phenomenon of global warming via studying the occurrence of record - breaking temperatures @xcite .
more precisely , let us consider a sequence or a discrete - time series of random variables @xmath19 with @xmath20 entries .
this sequence may represent for example the daily maximum temperature in a city or the daily maximum price of a stock .
a record is said to happen at step @xmath21 if the @xmath21-th member of the sequence is bigger than all previous members , i.e. , if @xmath22 for all @xmath23 .
let @xmath24 denote the number of records in this sequence of @xmath20 entries .
clearly @xmath24 is a random variable whose statistics depends on the joint distribution of @xmath25 of the members of the sequence .
when the members of the sequence are independent and identically distributed ( i.i.d ) random variables each drawn from a distribution @xmath26 , i.e. , the joint distribution factorizes , @xmath27 , the record statistics is well understood from classical theories @xcite . in particular , when @xmath26 is a continuous distribution , it is known that the distribution of record number @xmath28 is universal for all @xmath2 , i.e. , independent of the parent distribution @xmath26 .
the average number of records up to step @xmath2 , @xmath29 for all @xmath2 and the universal distribution , for large @xmath2 , converges to a gaussian distribution with mean @xmath30 and variance @xmath31 .
while the statistical properties of records for i.i.d random variables ( rv s ) are thus well understood for many years , numerous questions remain open for more realistic systems with time - dependent or correlated rv s . in principle
there are many different ways to generalize the simple i.i.d .
rv scenario described above .
for instance , one can consider time series of rv s that are independent , but not identically distributed .
one example for this case is the so called linear drift model with rv s from probability distributions with identical shape , but with a mean value that increases in time .
this model was first proposed in the 1980 s @xcite and was recently thoroughly analyzed in refs . @xcite . in 2007
krug also considered the case of uncorrelated rv s from distributions with increasing variance @xcite .
another possible generalization is the one where rv s are correlated .
perhaps , the simplest and the most natural model of correlated rv s is an @xmath2-step one dimensional discrete - time random walk with entries @xmath32 where the position @xmath33 of the walker at discrete time @xmath21 evolves via the markov jump process @xmath34 with @xmath35 and @xmath36 represents the random jump at step @xmath21 .
the noise variables @xmath36 s are assumed to be i.i.d variables , each drawn from a symmetric distribution @xmath37 .
for instance , it may include lvy flights where @xmath38 which has a divergent second moment .
even though this model represents a very simple markov chain , statistical properties of certain observables associated with such a walk may be quite nontrivial to compute , depending on which observable one is studying @xcite .
for instance , in recent years there has been a lot of interest in the extremal properties of such random walks .
these include the statistics of the maximal displacement of the walk up to @xmath2 steps with several applications @xcite and the order statistics , i.e. , the statistics of the ordered maxima @xcite as well as the universal distribution of gaps between successive ordered maxima of a random walk @xcite .
the statistics of the number of record - breaking events in the discrete - time random walk process in eq .
( [ markov.1 ] ) has also been studied in a number of recent works with several interesting results @xcite . in 2008 ,
majumdar and ziff computed exactly the full distribution @xmath28 of the record number up to @xmath2 steps and found that when the jump distribution @xmath37 is continuous and symmetric , the record number distribution @xmath28 is completely universal for all @xmath2 , i.e. , independent of the details of the jump distribution @xcite .
in particular , for instance , the lvy flight with index @xmath12 ( thus with a divergent second moment of the jump distribution @xmath37 ) has the same record number distribution as for a gaussian walk ( with a finite second moment of @xmath37 ) .
this is a rather amazing result and the deep reason for this universality is rooted @xcite in the so called sparre andersen theorem @xcite . in particular , for large @xmath2 , @xmath39 where the scaling function @xmath40 is universal @xcite .
the mean number of records @xmath41 for large @xmath2 @xcite .
in contrast , this universal result does not hold for symmetric but discontinuous @xmath37 .
for example , if @xmath42 , then @xmath43 represents the position of a random walker at step @xmath21 on a @xmath44-d lattice with lattice spacing @xmath44 . in this case
, the mean number of records still grows as @xmath45 for large @xmath2 but with a smaller prefactor , @xmath46 @xcite .
these results were later generalized to several interesting cases , for instance , to the record statistics of one dimensional random walk in presence of an external drift @xcite and one dimensional continuous - time random walk with a waiting - time distribution between successive jumps @xcite .
the record statistics of the distance traveled by a random walker in higher dimensions with and without drift has been studied numerically in the context of contamination spread in porous medium @xcite . in @xcite
, it was also found that the record statistics of stock markets is very similar to the ones of biased random walks . while in refs .
@xcite the record statistics of a single discrete - time random walker was studied , the purpose of this article is to generalize these results to the case where one has @xmath1 independent one dimensional discrete - time random walks . in this @xmath1-walker process , a record happens at an instant when the maximum position of all the walkers at that instant exceeds all its previous values .
we will see that despite the fact that the walkers are independent , the record statistics is rather rich , universal and nontrivial even in this relatively simple model .
let us first summarize our main results .
we derive asymptotic results for the mean of the record number @xmath47 up to a time @xmath2 and also discuss its full distribution .
it turns out that for @xmath48 , while the full universality with respect to the jump distribution found for @xmath49 case is no longer valid , there still remains a vestige of universality of a different sort . in our analysis , it is important to distinguish two cases : case ( i ) where the jump distribution @xmath37 has a finite variance @xmath50 and case ( ii ) where @xmath4 is divergent as in the case of lvy flights with lvy index @xmath12 . in both cases , we find that the mean record number @xmath6 grows _ universally _ as @xmath7 for large @xmath2
. however , the @xmath1 dependence of the prefactor @xmath8 , in particular for large @xmath1 , turns out to be rather different in the two cases @xmath51 { } \left\{\begin{array}{rl}2 \sqrt{\log n } & \textrm{in case i}\quad ( \textrm{independent of}\ , \sigma^2 ) \\ \vspace{1.5mm}\\ 4/\sqrt{\pi } & \textrm{in case ii}\quad ( \textrm{independent of}\ , \mu ) \end{array}\right.\ ] ] in addition , we also study the distribution of the record number @xmath0 . for finite @xmath4
we argue and confirm numerically that the distribution of the random variable @xmath13 converges to the gumbel law asymptotically for large @xmath2 and @xmath1 ( see section ii for details ) .
in contrast , in case ii , we find numerically that the distribution of @xmath16 converges , for large @xmath2 and @xmath1 , to a nontrivial distribution independent of the value of @xmath12 ( see section ii for details ) .
we were however unable to compute this asymptotic distribution analytically and it remains a challenging open problem .
finally , we discuss the applications of our results to the study of the record statistics of 366 daily stock prices from the standard & poors 500 index @xcite .
we analyze the evolution of the record number in subsets of @xmath1 stocks that were randomly chosen from this index and compare the results to our analytical findings .
while the strong correlations between the individual stocks seem to play an important effect in the record statistics , the dependence of the record number on @xmath1 still seems to be the same as in the case of the @xmath1 independent random walkers .
the rest of the paper is organized as follows . in section
ii , we define the @xmath1-walker model precisely and summarize the main results obtained in the paper . in section iii , we present the analytical calculation of the mean number of records for multiple random walkers , in both cases where @xmath4 is finite ( case i ) and @xmath4 is infinite ( case ii ) .
section iv is devoted to an analytic study of the distribution of the record number in the case where @xmath4 is finite . in section
v we present a thorough numerical study of the record statistics of multiple random walks , and in section vi we discuss the application of our results to the record statistics of stock prices . finally , we conclude in section vii and present the technical details of some of the analytical computations concerning the computation of the mean number of records and the distribution of the record number for lattice random walks in the three appendices a , b and c ..
here we consider the statistics of records of @xmath1 independent random walkers all starting at the origin @xmath3 .
the position @xmath52 of the @xmath53-th walker at discrete time step @xmath21 evolves via the markov evolution rule @xmath54 where @xmath55 for all @xmath56 and the noise @xmath57 s are i.i.d variables ( independent from step to step and from walker to walker ) , each drawn from a symmetric distribution @xmath37 .
we are interested in the record statistics of the composite process .
more precisely , consider at each step @xmath21 , the maximum position of all @xmath1 random walkers @xmath58 .
\label{max.1}\ ] ] a record is said to happen at step @xmath21 if this maximum position at step @xmath21 is bigger than all previous maximum positions , i.e. if @xmath59 for all @xmath60 ( see fig . [ multi1.fig ] ) . in other words , we are interested in the record statistics of the stochastic discrete - time series @xmath61 , with the convention that the initial position @xmath62 is counted as a record .
note that even though the position of each walker evolves via the simple independent markovian rule in eq .
( [ markov.2 ] ) , the evolution of the maximum process @xmath61 is highly non - markovian and hence is nontrivial
. random walkers .
each walker starts at the origin and evolves via the markov jump process in eq .
( [ markov.2 ] ) .
a record happens at step @xmath21 if the maximum position at step @xmath21 @xmath59 for all @xmath63 .
the record values are shown by filled circles.,scaledwidth=80.0% ] let @xmath0 denote the number of records up to step @xmath2 for this composite @xmath1-walker process .
clearly @xmath0 is a random variable and we are interested in its statistics . for a single walker @xmath49
, we have already mentioned that the probability distribution of the record number @xmath64 is completely universal , i.e. , independent of the jump distribution @xmath37 as long as @xmath37 is symmetric and continuous @xcite . in particular , for example , the record number distribution is the same for simple gaussian walkers as well for lvy flights with index @xmath12 .
here we are interested in the opposite limit when @xmath65 .
we find that while the complete universality of the record statistics is no longer true for @xmath48 , a different type of universal behavior emerges in the @xmath65 limit . in this large @xmath1 limit , there are two universal asymptotic behaviors of the record statistics depending on whether the second moment @xmath66 of the jump distribution is finite or divergent . for example , for gaussian ,
exponential , uniform jump distributions @xmath4 is finite .
in contrast , for lvy flights
where latexmath:[$f(\eta)\sim @xmath12 , the second moment @xmath4 is divergent . in these two cases , we find the following behaviors for the record statistics .
0.3 cm * case i ( @xmath4 finite ) : * in this case , we consider jump distributions @xmath37 that are symmetric with a finite second moment @xmath68 . in this case , the fourier transform of the jump distribution @xmath69 behaves , for small @xmath70 , as @xmath71 examples include the gaussian jump distribution , @xmath72 , exponential jump distribution @xmath73 $ ] , uniform jump distribution over @xmath74 $ ] etc . for such jump distributions , we find that for large number of walkers @xmath1 , the mean number of records grows asymptotically for large @xmath2 as @xmath75{n\to \infty } 2\ , \sqrt{\ln n}\ , \sqrt{n } \;.
\label{meanrecord1}\ ] ] note that this asymptotic behavior is universal in the sense that it does not depend explicitly on @xmath76 as long as @xmath76 is finite .
moreover , we argue that for large @xmath1 and large @xmath2 , the scaled random variable @xmath16 converges , in distribution , to the gumbel form , i.e , @xmath77 \xrightarrow[n\to \infty]{n\to \infty } f_{1}\left[\left(x-2\,\sqrt{\ln n}\right)\,\sqrt{\ln n}\right ] \ ; , \quad { \rm where } \quad f_1(z)=\exp\left[-\exp[-z]\right ] .
\label{gumbel.1}\ ] ] indeed , for large @xmath1 and large @xmath2 , the scaled variable @xmath16 converges , in distribution , to the maximum of @xmath1 independent random variables @xmath78{n\to \infty } m_n\ , \quad { \rm where}\quad m_n= { \rm max}(y_1,y_2,\ldots , y_n ) \label{gumbel.2}\ ] ] where @xmath79 s are i.i.d non - negative random variables each drawn from distribution @xmath80 for @xmath81 and @xmath82 for @xmath83
. 0.3 cm * case ii ( @xmath4 divergent ) :* in this case we consider jump distributions @xmath37 such that the second moment @xmath4 is divergent . in this case , the fourier transform @xmath84 of the noise distribution behaves , for all @xmath70 , as @xmath85 where @xmath12 .
examples include lvy flights where @xmath86 with the lvy index @xmath87 . for the noise distribution in eq .
( [ fourier2 ] ) , we find , quite amazingly , that in the large @xmath1 and large @xmath2 limit , the record statistics is ( i ) completely universal , i.e. , independent of @xmath18 and @xmath88 ( ii ) more surprisingly and unlike in case - i , the record statistics also becomes independent of @xmath1 as @xmath65 .
for example , we prove that for large @xmath1 , the mean number of records grows asymptotically with @xmath2 as @xmath75{n\to \infty } \frac{4}{\sqrt{\pi}}\ , \sqrt{n } \ ; , \label{meanrecord2}\ ] ] which is exactly twice that of one walker , i.e. , @xmath89 for large @xmath2 .
similarly , we find that the scaled variable @xmath16 , for large @xmath2 and large @xmath1 , converges to a universal distribution @xmath77 \xrightarrow[n\to
\infty]{n\to \infty } f_2(x ) \ ; , \label{distri2}\ ] ] which is independent of the lvy index @xmath18 as well as of the scale @xmath88 in eq .
( [ fourier2 ] ) .
while we have computed this universal distribution @xmath90 numerically rather accurately , we were not able to compute its analytical form .
let @xmath0 be the number of records up to step @xmath2 for @xmath1 random walkers , i.e. , for the maximum process @xmath91 .
let us write @xmath92 where @xmath93 is a binary random variable taking values @xmath3 or @xmath44 .
the variable @xmath94 if a record happens at step @xmath21 and @xmath95 otherwise .
clearly , the total number of records up to step @xmath2 is @xmath96 so , the mean number of records up to step @xmath2 is @xmath97 where @xmath98 is just the record rate , i.e. , the probability that a record happens at step @xmath21 . to compute the mean number of records , we will first evaluate the record rate @xmath99 and then sum over @xmath21 . to compute @xmath99 at step @xmath21 , we need to sum the probabilities of all trajectories that lead to a record event at step @xmath21 .
suppose that a record happens at step @xmath21 with the record value @xmath100 ( see fig .
[ recrate1.fig ] ) .
this corresponds to the event that one of the @xmath1 walkers ( say the dashed trajectory in fig .
[ recrate1.fig ] ) , starting at the origin at step @xmath3 , has reached the level @xmath100 for the first time at step @xmath21 , while the rest of the @xmath101 walkers , starting at the origin at step @xmath3 , have all stayed below the level @xmath100 till the step @xmath21 .
also , the walker that actually reaches @xmath100 at step @xmath21 can be any of the @xmath1 walkers .
finally this event can take place at any level @xmath102 and one needs to integrate over the record value @xmath100 . using the independence of @xmath1 walkers and taking into account the event detailed above
, one can then write @xmath103^{n-1}\ , dx \ ; , \label{rate.1}\ ] ] where @xmath104 denotes the probability that a single walker , starting at the origin , stays below the level @xmath100 up to step @xmath21 and @xmath105 is the probability density that a single walker reaches the level @xmath100 for the first time at step @xmath21 , starting at the origin at step @xmath3 . with record value @xmath100 for @xmath106 walkers , all starting at the origin .
this event corresponds to one walker ( the dashed line ) reaching the level @xmath100 for the first time at step @xmath21 while the other walkers stay below the level @xmath100 up to step @xmath21.,scaledwidth=80.0% ] the two quantities @xmath105 and @xmath104 can be reinterpreted in terms of slightly more familiar objects via the following observation .
note that by shifting the origin to the level @xmath100 and using the time - reversal property of the trajectory of a single random walker , it is easy to see that @xmath105 is just the probability density that a single walker , starting at the origin at step @xmath3 , reaches @xmath100 at step @xmath21 while staying positive at all intermediate steps . by a similar shift of the origin to level @xmath100 and using the reflection symmetry of the trajectories around the origin , it is clear that @xmath104 can be interpreted as the probability that a single walker , starting at an initial position @xmath102 at step @xmath3 , stays positive ( i.e. , does not cross the origin ) up to step @xmath21 .
this is then the familiar persistence or the survival probability of a single random walker @xcite .
in fact , both these quantities @xmath105 and @xmath104 can be regarded as special cases of the more general restricted green s function in the following sense .
consider a single random walker starting at position @xmath100 at step @xmath3 and evolving its position via successive uncorrelated jumps as in eq .
( [ markov.1 ] ) .
let @xmath107 denote the probability density that the walker reaches @xmath108 at step @xmath21 , starting at @xmath102 at step @xmath3 , while staying positive at all intermediate steps .
the subscript @xmath109 denotes that it is indeed the restricted green s function counting only the trajectories that reaches @xmath110 at step @xmath21 without crossing the origin in between .
it is then clear from our discussion above that @xmath111 in the second line , the survival probability @xmath104 is obtained from the restricted green s function by integrating over all possible final positions of the walker .
note also , from eqs .
( [ pm1 ] ) and ( [ qm1 ] ) , that the survival probability starting exactly at the origin is @xmath112 hence , if we know the restricted green s function @xmath113 , we can in principle compute the two required quantities @xmath105 and @xmath104 . using the markov evolution rule in eq .
( [ markov.1 ] ) , it is easy to see that the restricted green s function @xmath113 satisfies an integral equation in the semi - infinite domain @xcite @xmath114 starting from the initial condition , @xmath115 .
such integral equations over the semi - infinite domain are called wiener - hopf equations and are notoriously difficult to solve for arbitrary kernel @xmath116 . fortunately , for the case when @xmath116 represents a continuous and symmetric probability density as in our case , one can obtain a closed form solution for the following generating function ( rather its laplace transform ) @xcite @xmath117= { \tilde g}(\lambda,\lambda_0,s)= \frac{\phi(s,\lambda)\,\phi(s,\lambda_0)}{\lambda+\lambda_0}\ ; , \label{ivanov.1}\ ] ] where @xmath118}{\lambda^2+k^2}\ , dk\right ] \quad { \rm and}\quad { \hat f}(k)= \int_{-\infty}^{\infty } f(x)\ , e^{i\,k\,x}\ , dx \;. \label{ivanov.2}\ ] ] while the formula in eq .
( [ ivanov.1 ] ) is explicit , it is rather cumbersome and one needs further work to extract the asymptotic behavior of @xmath105 and @xmath104 from this general expression . to make progress
, one can first make a change of variable on the left hand side ( lhs ) @xmath119 and then take the @xmath120 limit . using @xmath121 and the definition @xmath122 , and replacing @xmath110 by @xmath100
we then obtain the following relation @xmath123 where @xmath124 is given in eq .
( [ ivanov.2 ] ) .
similarly , putting @xmath125 on the lhs of eq .
( [ ivanov.1 ] ) , using the definition @xmath126 and replacing @xmath127 by @xmath128 , it is easy to see that @xmath129 the formula in eq .
( [ qm.2 ] ) is known in the literature as the celebrated pollaczek - spitzer formula @xcite and has been used in a number of works to derive exact results on the maximum of a random jump process @xcite .
interestingly , this formula has also been useful to compute the asymptotic behavior of the flux of particles to a spherical trap in three dimensions @xcite .
let us also remark that by making a change of variable @xmath130 on the lhs of eq .
( [ qm.2 ] ) and taking @xmath131 , one obtains the rather amazing universal result for all @xmath21 @xmath132 which is known as the sparre andersen theorem @xcite . in particular , for large @xmath21 , @xmath133 .
note that for the case of a single walker @xmath49 , it follows from eq .
( [ rate.1 ] ) that the record rate at step @xmath21 is simply given by @xmath134 where we have used eq .
( [ q0 ] ) and the sparre andersen theorem ( [ sa1 ] ) .
thus , one obtains the rather surprising universal result for the @xmath49 case : for all continuous and symmetric jump distributions , the mean number of records up to step @xmath2 , @xmath135 is universal for all @xmath2 and grows as @xmath136 for large @xmath2 @xcite .
the universality in this case can thus be traced back to sparre andersen theorem .
in contrast , for @xmath48 , we need the full functions @xmath105 and @xmath104 to compute the record rate in eq .
( [ rate.1 ] ) .
this is hard to compute explicitly for all @xmath21 .
however , one can make progress in computing the asymptotic behavior of the record rate @xmath99 for large @xmath21 and large @xmath1 , as we show below .
in turns out that for large @xmath21 , the integral in eq .
( [ rate.1 ] ) is dominated by the asymptotic scaling behavior of the two functions @xmath105 and @xmath104 for large @xmath21 and large @xmath100 . to extract the scaling behavior of @xmath105 and @xmath104
, our starting point would be the two equations ( [ pm.2 ] ) and ( [ qm.2 ] ) .
the next step is to use these asymptotic expressions in the main formula in eq .
( [ rate.1 ] ) to determine the record rate @xmath99 at step @xmath21 for large @xmath21 and large @xmath1 .
the procedure to extract the asymptotics is somewhat subtle and algebraically cumbersome . to facilitate an easy reading of the paper , we relegate this algebraic procedure in the appendices . here
we just use the main results from these appendices and proceed to derive the results announced in eqs .
( [ meanrecord1 ] ) and ( [ meanrecord2 ] ) .
the asymptotic behavior of @xmath105 and @xmath104 depend on whether @xmath137 is finite or divergent .
this gives rise to the two cases mentioned in section ii .
0.3 cm * case i(@xmath4 finite ) : * in this case , we show in appendix a that in the scaling limit @xmath138 , @xmath139 but keeping the ration @xmath140 fixed , @xmath105 and @xmath104 approach the following scaling behavior @xmath141 where @xmath142 . note that @xmath143 . 0.3 cm * case ii ( @xmath4 divergent ) : * for the case when the fourier transform of the jump distribution @xmath84 has the small @xmath70 behavior as in eq .
( [ fourier2 ] ) , we show in appendix b that in the scaling limit when @xmath138 , @xmath139 , but keeping the ratio @xmath144 fixed , @xmath145 while it is hard to obtain explicitly the full scaling functions @xmath146 and @xmath147 for all @xmath148 , one can compute the large @xmath148 asymptotic behavior and obtain @xmath149 where the two amplitudes are @xmath150 with the constant @xmath151 having different expressions for @xmath152 and @xmath153 @xmath154 the expressions above ( [ betamu1 ] , [ betamu2 ] ) can be written in a unified way for any @xmath155 as @xmath156 where , in the last equality , we have used @xmath157 .
we recall that here we are considering discrete time random walks ( [ markov.1 ] ) . in the continuous time
random walk framework , with an exponential waiting time between jumps , the quantity @xmath146 was studied in ref .
@xcite . by performing an asymptotic analysis similar to the one presented in appendix b
, the authors showed that @xmath146 behaves , for large @xmath148 , like in eq .
( [ g2largez ] ) with the same exponent albeit with a different amplitude . on the other hand , the exact asymptotic result ( [ g2largez ] ) , together with eq .
( [ unif_beta ] ) can also be used to study the normalized pdf @xmath158 of the position after @xmath21 steps , with the condition that the walker stays positive at all intermediate steps , which was recently studied in ref .
it reads @xmath159 where we have used the sparre - andersen theorem @xmath160 for large @xmath21 . from eq .
( [ g2largez ] ) , one obtains the large @xmath148 behavior of @xmath161 as @xmath162 where we have used @xmath163 . on the other hand , if one considers the probability density function @xmath164 of the position of a free lvy random walk after @xmath21 steps , with a jump distribution as in eq .
( [ fourier2 ] ) after @xmath21 steps , it assumes the scaling form , valid for large @xmath21 , @xmath165 where the asymptotic behavior is given by @xmath166 therefore the above result ( [ asympt_constrained_propag ] ) establishes that @xmath167 : this result was recently obtained analytically in perturbation theory for @xmath18 close to @xmath168 , @xmath169 , and conjectured to hold for any @xmath18 , on the basis of thorough numerical simulations @xcite . here this result is established exactly for any @xmath170 .
while the large @xmath148 behavior of @xmath146 is the most relevant one for our study , we mention , for completeness , that its small @xmath148 behavior was also studied in ref .
@xcite , yielding @xmath171 .
finally we remark that the asymptotic behavior of @xmath147 for large @xmath148 has been computed in great detail recently in ref .
@xcite , only the first two leading terms are presented in eq .
( [ h2largez ] ) here .
we are now ready to use these asymptotic behavior of @xmath105 and @xmath104 in eq .
( [ rate.1 ] ) to deduce the large @xmath21 behavior of the record rate .
noting that for large @xmath21 , the integral is dominated by the scaling regime , we substitute in eq .
( [ rate.1 ] ) the scaling forms of @xmath105 and @xmath104 found in eqs .
( [ pm1scaling ] ) , ( [ qm1scaling ] ) , ( [ pm2scaling ] ) and ( [ qm2scaling ] ) .
we then get , for large @xmath21 , @xmath172^{n-1}\ , dz \ ; , \label{rate.2}\ ] ] where @xmath173 and @xmath174 depending on the two cases .
so , we notice that in all cases the record rate decreases as @xmath175 for large @xmath21 , albeit with different @xmath1-dependent prefactors in the two cases .
hence , the mean number of records @xmath176 up to step @xmath2 grows , for large @xmath2 , as @xmath177^{n-1}\ , dz \;.
\label{avgrec.1}\ ] ] next we estimate the constant @xmath8 for large @xmath1 .
we first note that @xmath8 in eq .
( [ avgrec.1 ] ) can be expressed as @xmath178^n\}\ , dz \ ; , \label{alphan}\ ] ] where @xmath179 .
noticing that @xmath180 is an increasing function of @xmath148 approaching @xmath44 as @xmath181 , the dominant contribution to the integral for large @xmath1 comes from the large @xmath148 regime .
hence , we need to estimate how the ratio @xmath182 behaves for large @xmath148 .
let us consider the two cases separately .
0.3 cm * case i ( @xmath4 finite ) : * in this case , we have explicit expressions for @xmath183 and @xmath184 respectively in eqs .
( [ pm1scaling ] ) and ( [ qm1scaling ] ) .
hence we get @xmath185^n \\ & = & \int_0^{\infty } dy\ , y\ , \frac{d}{dy } [ { \rm erf}(y/2)]^{n}. \label{alphan1.case1}\end{aligned}\ ] ] the rhs of eq .
( [ alphan1.case1 ] ) has a nice interpretation .
consider @xmath1 i.i.d positive random variables @xmath186 , each drawn from the distribution : @xmath187 for @xmath81 and @xmath82 for @xmath83 .
let @xmath188 denote their maximum .
then the cdf of the maximum is given by @xmath189= \left[\int_0^{y } p(y')\,dy'\right]^n= [ { { \rm erf}(y/2}]^n\ , . \label{maximum.1}\ ] ] the probability density of the maximum is then given by : @xmath190^n$ ] . hence , the rhs of eq .
( [ alphan1.case1 ] ) is just the average value @xmath191 of the maximum .
this gives us an identity for all @xmath1 @xmath192 from the standard extreme value analysis of i.i.d variables @xcite , it is easy to show that to leading order for large @xmath1 , @xmath193 which then gives , via eq .
( [ avgrec.1 ] ) , the leading asymptotic behavior of the mean record number @xmath194{n\to \infty } 2 \sqrt{\ln n}\ , \sqrt{n } \;.
\label{avgrec.case1}\ ] ] 0.3 cm * case ii ( @xmath4 divergent ) : * to evaluate @xmath8 in eq .
( [ alphan ] ) , we note that when @xmath4 is divergent , unlike in case i , we do not have the full explicit form of the scaling functions @xmath146 and @xmath147 .
hence evaluation of @xmath8 for all @xmath1 is difficult .
however , we can make progress for large @xmath1 . as mentioned before , for large @xmath1 , the dominant contribution to the integral in eq .
( [ alphan ] ) comes from large @xmath148 . for large @xmath148 , using the asymptotic expressions in eqs .
( [ g2largez ] ) and ( [ h2largez ] ) , we get @xmath195 where we have used eqs .
( [ amu1 ] ) and ( [ bmu1 ] ) for the expressions of @xmath196 and @xmath197 .
we next substitute this asymptotic constant for the ratio @xmath198 in the integral on the rhs of eq .
( [ alphan ] ) .
the integral can then be performed trivially and we get , for large @xmath1 , @xmath199 from eq .
( [ avgrec.1 ] ) we then get for the mean record number @xmath194{n\to \infty } \frac{4}{\sqrt{\pi}}\ , \sqrt{n } \ , .
\label{avgrec.case2}\ ] ] in contrast to case i in eq .
( [ avgrec.case1 ] ) , here the mean record number becomes independent of @xmath1 for large @xmath1 .
in the previous section , we performed a very precise study of the mean number of records @xmath200 up to step @xmath2 , in both cases where @xmath4 is finite and divergent . in the present section ,
we investigate the full probability distribution function ( pdf ) of the record number @xmath0 .
however , we have been able to make analytical progress for the record number distribution only in case i where @xmath4 is finite to which we restrict ourselves below .
the clue that leads to an analytical computation of the record number distribution is actually already contained in the exact expression of the mean record number in eqs .
( [ avgrec.1 ] ) and ( [ alphan1.case1 ] ) .
this result suggests that there perhaps is a relation between the record number @xmath0 and the stochastic variable @xmath201 defined as @xmath202 .
\label{ydef}\ ] ] note that @xmath201 simply denotes the maximum position of all the walkers _ up to _ step @xmath2 . in this section
, we will see that for case i where @xmath4 is finite , indeed there is a close relation between the two random variables @xmath0 and @xmath201 . to uncover this relation , it is actually instructive to consider first the case of @xmath1 independent lattice random walks defined by eq .
( [ markov.2 ] ) where the noise @xmath57 s are i.i.d .
random variables with a distribution @xmath203 .
consider now the time evolution of the two random processes @xmath0 and @xmath201 . at the next time
step @xmath204 , if a new site on the positive axis is visited by any of the walkers for the first time , the process @xmath201 increases by @xmath44 , otherwise its value remains unchanged .
whenever this event happens , i.e. , a new site on the positive side is visited for the first time , one also has a record event , i.e. , the process @xmath0 also increases by @xmath44 .
otherwise @xmath0 remains unchanged .
thus , the two random processes @xmath201 and @xmath0 are completely locked with each other at all steps : whenever one of them increases by unity at a given step the other does the same simultaneously and when one of them does not change , the other also remains unchanged .
in other words , for every realization , we have , @xmath205 .
now , initially all walkers start at the origin indicating @xmath206 while @xmath207 since the initial point is counted as a record by convention .
this allows us to write the following identity for all @xmath2 and @xmath1 @xmath208 + 1 \;.\end{aligned}\ ] ] we can now take advantage of this identity to compute the probability @xmath209 as the distribution of @xmath201 , i.e. , the maximum of @xmath1 independent lattice walkers up to @xmath2 steps can be computed using the standard method of images .
one obtains , after some manipulations left in appendix [ app_lattice ] , for @xmath210 @xmath211 \right)^n - \frac{1}{2^{n\,n } } \left(\sum_{k=0}^{\lfloor \frac{n+m-1}{2 } \rfloor } \left[{n \choose
k } - { n \choose k - m+1}\right ] \right)^n \ ; , \end{aligned}\ ] ] where @xmath212 is the largest integer not greater than @xmath100 . for instance , for @xmath49 one gets from eq .
( [ exact_bernoulli ] ) @xmath213 where @xmath214 is the smallest integer not less than @xmath100 .
we have checked that this formula for @xmath49 ( [ exact_bernoulli_n1 ] ) yields back the result for the first moment @xmath215 as obtained in ref .
@xcite . from the above formula ( [ exact_bernoulli ] ) one can also compute @xmath216 , for instance with mathematica ,
although obtaining a simple closed form formula for it for @xmath48 seems rather difficult . in table
[ first_values ] we have reported the first few values of @xmath200 for @xmath49 to @xmath217 .
.first values of @xmath200 obtained from eq .
( [ exact_bernoulli ] ) . [ cols="^,^,^,^,^,^",options="header " , ] using the identity ( [ identity ] ) , one can also obtain the large @xmath2 behavior of @xmath0 .
indeed , in this limit , each rescaled ordinary random walk @xmath218 converges , when @xmath14 , to a brownian motion @xmath219 with a diffusion coefficient @xmath220 , on the unit time interval , @xmath221 $ ] .
therefore from the above identity ( [ identity ] ) one gets , in the limit @xmath14 @xmath222 \;.\end{aligned}\ ] ] now , the distribution of @xmath223 , i.e. , the maximum of a single brownian motion ( with diffusion constant @xmath220 ) over a unit interval is well known ( see e.g. , in @xcite ) @xmath224 = \sqrt{\frac{2}{{\pi } } } \int_0^m\exp\left(-\frac{x^2}{2 } \right ) \ ; dx = { \rm erf}\left(\frac{m}{\sqrt{2}}\right ) \;,\end{aligned}\ ] ] eq .
( [ large_n_discrete ] ) demonstrates that in this case , @xmath16 is distributed like the maximum of a collection @xmath1 i.i.d positive random variables @xmath225 , each drawn from the distribution : @xmath226 for @xmath227 and @xmath228 for @xmath229 . from eq .
( [ discrete_n1 ] ) one obtains also that @xmath230 , for @xmath231 , as obtained in ref .
@xcite , using a different method .
more generally for any @xmath1 one has @xmath232 = \left ( { \rm proba.}\,[\tilde m_{1}\leq m ] \right)^n
= \left [ { \rm erf } \left(\frac{m}{\sqrt{2}}\right ) \right]^n \;.\end{aligned}\ ] ] having discussed the lattice random walk , let us now return to the case where the jump distribution is continuous in space but with a finite @xmath4 . in this case
, we do not have an identity between @xmath0 and @xmath201 analogous to eq .
( [ identity ] ) for lattice random walks .
nevertheless , we conjecture below and later provide numerical evidence in section va , that for large @xmath2 , the scaled random variable @xmath16 converges , in distribution , to the scaled variable @xmath233 up to a prefactor @xmath234 , i.e. , @xmath235 where @xmath236 indicates that the random variables on both sides of eq .
( [ converge1 ] ) have the same probability distribution . on the other hand , using central limit theorem ,
it is easy to see that the position of each rescaled walker @xmath237 converges in distribution , as @xmath238 , to a continuous - time brownian motion @xmath239 with a diffusion coefficient @xmath240 , over the unit interval @xmath241 $ ] .
thus the conjecture in eq .
( [ converge1 ] ) is equivalent to @xmath242 \ , .
\label{large_n_continuous}\ ] ] the argument leading to this conjecture in eq .
( [ large_n_continuous ] ) can be framed as follows .
consider first the case for @xmath49 . in this case , the full distribution of @xmath64 was computed in ref .
@xcite for all @xmath2 and in particular , for large @xmath2 , it was shown that @xcite that @xmath243 \ ; , \end{aligned}\ ] ] where @xmath244 is a brownian motion with diffusion coefficient @xmath240 on the unit time interval starting from @xmath245 . this result ( [ continuous_n1 ] ) , for continuous jump distribution , is very similar to the one obtained for a lattice random walk in eq .
( [ large_n_discrete ] ) for @xmath49 , except that the diffusion coefficient of the brownian motion involved in the discrete case is @xmath220 while it is @xmath240 , independently of @xmath4 , for continuous jump distributions .
in particular , from eq .
( [ continuous_n1 ] ) one obtains @xmath246 { } { \rm erf}\left(\frac{x}{2}\right ) \;.\end{aligned}\ ] ] hence , at least for @xmath49 , we know that for large @xmath2 , @xmath247 for the continuous jump distribution in eq .
( [ continuous_n1 ] ) behaves in a statistically similar way as that for lattice walks , the only difference is that the effective diffusion constant of the underlying brownian motion is @xmath240 in the continuous case ( eq .
( [ large_n_continuous ] ) ) , while it is @xmath220 for the lattice case ( eq .
( [ large_n_discrete ] ) ) . based on this exact relation for @xmath49 , it is then natural to presume that this asymptotic equality in law between record numbers for continuous jump process and lattice walks holds even for @xmath48 , thus leading to the conjecture in eq .
( [ large_n_continuous ] ) . as a first check of the validity of this conjecture
, we note that the result for the first moment of the record number in eqs .
( [ avgrec.1 ] ) and ( [ maximum.2 ] ) is fully consistent with the conjecture in eq .
( [ large_n_continuous ] ) . as announced in the introduction ( [ gumbel.2 ] ) , the conjecture in eq .
( [ large_n_continuous ] ) is equivalent to say that @xmath16 converges , in distribution , to the maximum of @xmath1 independent random variables @xmath248 , where @xmath79 s are i.i.d non - negative random variables each drawn from distribution @xmath80 for @xmath81 and @xmath82 for @xmath83 .
in particular , the cdf of @xmath16 is given by @xmath249 { } \left [ { \rm erf}\left(\frac{x}{2}\right ) \right]^n \;.\end{aligned}\ ] ] in section vb we will demonstrate that this conjecture is well supported by our numerical simulations . from eq .
( [ large_n_continuous ] ) , one can then use standard results for the extreme statistics of independent random variables @xcite to obtain that for large @xmath1 and large @xmath2 , the scaled random variable @xmath16 ( properly shifted and scaled ) converges , in distribution , to the gumbel distribution as announced in eq .
( [ gumbel.1 ] ) .
although we have not found a rigorous proof of the above result ( [ large_n_continuous ] ) , the fact that both formulae for random walk with discrete ( [ large_n_discrete ] ) and continuous ( [ large_n_continuous ] ) jump distribution differ essentially by a factor of @xmath250 is reminiscent of a similar difference , by the same factor of @xmath250 , for the survival probability @xmath251 corresponding to both random walks ( starting from the origin ) .
this quantity @xmath251 plays indeed a crucial role in the computation of the record statistics of a random walk @xcite . for the lattice random walk
, one has indeed @xmath252 while for the continuous random walk one has , from sparre - andersen s theorem , @xmath253 , independently of @xmath4 .
this fact certainly deserves further investigations . in the case of divergent @xmath4 ,
the two random variables @xmath0 and @xmath201 do not seem to be related in any simple way . this is already evident from the result for the mean record number @xmath254 in eq .
( [ avgrec.case2 ] ) for @xmath255 . as @xmath238 , @xmath256 where the prefactor does not depend on @xmath1 for large @xmath1 .
in contrast , using standard extreme value statistics @xcite , it is easy to show that the mean value @xmath257 for large @xmath2 and @xmath1 with @xmath12 .
this rather different asymptotic behavior of the mean thus already rules out any relationship between @xmath0 and @xmath201 for case ii .
so , for this case , our result for the distribution of @xmath0 is only restricted to numerical simulations that are presented in the next section .
in this section we present the results of our numerical simulations of @xmath1 independent random walks both with a finite @xmath4 ( case i ) and with a divergent @xmath4 ( case ii ) and compare them with our analytical results .
in the first subsection [ subsection_distributions ] we study the statistics of the record numbers ( both its mean value and its full distribution ) . since , at least in case i , the mean record number is strongly related to the expected maximum of the process we will then analyze the evolution of the largest of the @xmath1 random walkers .
this will be done in section [ subsection_max ] .
we find that the statistics of the maximum significantly differs between the cases i and ii . finally , in section [ subsection_correlations ] we will consider the correlations between individual record events in the two different cases and show that , at least asymptotically , these correlations are not different from the case of only one single random walker .
* case i ( @xmath4 finite)*. in fig .
( [ rec_no.fig ] ) , we show our numerical results for @xmath6 for @xmath4 finite , which were obtained by a direct simulation of the jump process in eq .
( [ markov.1 ] ) with @xmath258 steps , with a gaussian distribution of the jump variables @xmath57 s ( mean @xmath3 and @xmath259 ) .
these results have been obtained by averaging over @xmath260 different realizations of the random walks .
these data , on fig .
( [ rec_no.fig ] ) are indexed by the label gaussian. our numerical data show a very nice agreement with our analytical result obtained in eq .
( [ avgrec.1 ] ) yielding @xmath261 .
the large @xmath1 behavior of @xmath8 can be easily obtained by a saddle point analysis , yielding : @xmath262 \;.\end{aligned}\ ] ] it turns out that for @xmath263 , the sub - leading corrections ( [ sub_leading ] ) are still sizeable .
we have also computed numerically the distribution of the ( scaled ) record numbers @xmath16 and compared it to our conjecture in eq .
( [ large_n_continuous ] ) .
the results of this comparison , for different values of @xmath264 are shown in fig .
[ check_conjecture ] , where the data were obtained by averaging over @xmath265 realizations of independent random walks of @xmath266 steps .
the data , for two different continuous jump distributions ( exponential and uniform ) show a very nice agreement with our analytical prediction in eq .
( [ conjecture ] ) , which we argue to be an exact result . as mentioned above
, one expects that this distribution will converge , for @xmath15 to a gumbel distribution ( [ gumbel.1 ] ) , albeit with strong finite @xmath1 effects .
* case ii ( @xmath4 divergent)*. in fig . ( [ rec_no.fig ] ) , we show our numerical results for @xmath6 for @xmath4 divergent , obtained by a direct simulation of the random walk ( [ markov.1 ] ) where the distribution of @xmath57 s has a power law tail @xmath267 with @xmath268 , and different values of @xmath18 .
the data presented there were also obtained by averaging over @xmath260 different realizations of random walks , with @xmath269 steps .
these data show that , in this case , @xmath270 approaches a constant value for fixed ( and large ) @xmath2 and @xmath271 , which is fully consistent with the value of @xmath272 obtained analytically in eq .
( [ avgrec.case2 ] ) .
the simulations also show how the speed of this convergence is modified when @xmath273 is varied .
while for small @xmath274 , @xmath275 approaches the universal limit value of @xmath272 very quickly , we find a slower convergence for @xmath276 .
the numerical computation of the distribution of the ( scaled ) record number @xmath16 is of special interest because an analytical study of it , beyond the first moment , is still lacking .
the two plots on fig .
( [ distributions.fig ] ) show our numerical results for this distribution , which were obtained by averaging over @xmath269 independent random walks of length @xmath266 . the left panel in fig .
( [ distributions.fig ] ) shows the rescaled distribution of @xmath16 at a fixed time step @xmath277 and @xmath278 .
apparently all curves for @xmath279 and @xmath269 collapse on one line . in the inset of the left figure we kept @xmath277 and @xmath280 fixed and varied @xmath18 : one can see that all the cdf s collapse .
these results suggest that @xmath77 \xrightarrow[n\to \infty]{n\to \infty } f_2(x ) \ ; , \label{distri2.1}\ ] ] where the limiting distribution function @xmath281 is independent of @xmath273 .
we tried to guess the analytic form of @xmath281 by comparing with several known continuous distributions that are defined for positive real numbers .
we are certain that @xmath281 is not a gaussian distribution . by far the best results
were obtained by fitting with a weibull distribution @xmath282 with two free real parameters @xmath283 .
fitting with the least - squares method gives values of @xmath284 and @xmath285 .
the right panel in fig .
( [ distributions.fig ] ) compares this fit with the distribution obtained from a simulation of @xmath286 random walks of length @xmath266 . while the agreement , both for the cdf and the probability density function ( pdf ) is quite good ,
there are still some deviations between the two , particularly for small values close to zero .
we are not sure , if this difference is a finite @xmath1 effect or if the real limit distribution of @xmath16 for @xmath271 effectively differs from a weibull distribution . in the case of jump distributions with a finite second moment @xmath4 ( case i ) ,
we have shown that the mean @xmath287 of the maximum of all walkers up to step @xmath2 and the mean record number @xmath200 are proportional to each other in the limit @xmath288 , both grow with @xmath289 .
in contrast , for lvy walks with index @xmath273 ( case ii ) the relation between these two observables does not hold any more and the mean record number grows much slower than the maximum ( see the discussion at the end of section iv ) . to illustrate the similarities and differences in the growth rate of @xmath0 and @xmath201 in the two cases ( i and ii )
, we compare their respective time evolution for @xmath290 different samples in fig .
( [ max.fig ] ) . on the left panel
, we consider the gaussian jump distribution with zero mean and @xmath259 and we see that the process @xmath0 and @xmath291 become identical very quickly .
moreover , the two processes evolve almost in a deterministic fashion with growing @xmath2 and hardly fluctuate from one sample to another .
in contrast , on the right panel where we plot the two processes for @xmath278 , their behavior change drastically .
first of all , the two processes @xmath0 and @xmath201 do not seem to have relation to each other . while @xmath0 again evolves almost deterministically and in a self - averaging way , the trajectories of the process
@xmath201 differ strongly from one sample to another and @xmath201 is clearly non self - averaging .
in particular , the process @xmath201 can , like in a single lvy flight , perform very large jumps exceeding its previous value by several orders of magnitude .
an important feature of the record statistics of a single random walk ( @xmath49 ) is its renewal property , which leads to the fact , that each time after a record event , the record statistics is , in some sense , _ reseted_. after a record event the process evolves as a new process with the record value as its new origin .
therefore it is very simple to give the pairwise correlations between record events . in fact , from the above argument , we have @xmath292 = \textrm{prob}\left[\textrm{rec .
at}\;n - k\right ] \times \textrm{prob}\left[\textrm{rec .
at}\;k\right ] = r_{n - k,1}r_{k,1 } \;.\ ] ] using the results from @xcite this gives the following ( exact ) result for @xmath293 $ ] : @xmath292 = \left(2\left(n - k\right ) \atop n - k\right ) \left(2k\atop k\right ) 2^{-2n } \;.\ ] ] in the special case of @xmath294 we find @xmath295 = \frac{1}{2 } r_{n-1}$ ] . with this
we find that the conditional probability of a second record directly following a record that just occurred is always given by @xmath296 = \frac{1}{2 } \;.\ ] ] in our efforts to understand and compute the distribution of the record number @xmath0 for lvy flights ( case ii ) , we considered the correlations between successive record events also for @xmath297 random walks .
if the correlations between successive record events in the large @xmath1 limit would vanish , one could assume that the asymptotic distribution of @xmath0 approaches a gaussian .
however , we found that this is not the case .
( [ correlations.fig ] ) gives the behavior of @xmath293 $ ] for the @xmath49 case , as well as for @xmath280 for random walks of the two cases i and ii with lvy indices @xmath298 and @xmath278 . in all three cases @xmath295 $ ] approaches @xmath299 = \frac{1}{2}r_{n , n}$ ] for large @xmath2 , proving that for @xmath288 the probability for a second record after an occurred one is just @xmath300 .
the inset of fig .
( [ correlations.fig ] ) also shows this behavior . here , while for @xmath49 the conditional probability for a second record is always @xmath300 , this value is only approached for larger @xmath2 in the case of @xmath297 . for small @xmath2
the conditional probability is larger .
the oldest application of the random walk model , which was already proposed by le bachelier @xcite in 1900 , is the one to stock data @xcite . in his model
the stock prices perform a so - called geometric random walk and trends in the stocks are modeled by a linear drift in the logarithms of the stock prices . in @xcite the record statistics of stocks in the standard and poors 500 @xcite ( s&p 500 ) index were compared to the records in a random walk with a drift .
the authors could show that on average , on a time interval of @xmath301 trading days , the statistics of upper records in individually detrended stocks are in good agreement with the same statistics of a random walk with a symmetric jump distribution .
the lower records however showed a significant deviation from this model . here , we want to extend this analysis to the record statistics of @xmath1 stocks .
the question is , to what degree , the record statistics of @xmath1 normalized and randomly chosen stocks from the s&p 500 can be compared to the record statistics of @xmath1 independent random walks . as in @xcite , the observational data we used consisted of @xmath302 stocks that remained in the s&p 500 index for the entire time - span from january 1990 to march 2009 .
overall , we had data for @xmath303 consecutive trading days for each stock at our disposal .
in @xcite we found that it is useful to analyze this data over smaller intervals , on which one can then detrend the measurements .
we decided to split up the @xmath303 trading days into @xmath304 consecutive intervals of each @xmath305 trading days , which is roughly one calendar year . in each of these intervals we considered the logarithms @xmath306 of the stock prices @xmath307 , where @xmath308 is the first trading day .
the random walk model then suggest that these logarithms @xmath309 perform a biased random walk that starts at the origin ( @xmath310 ) . since our analytical theory presented in this paper only works for symmetric random walks we had to detrend the stocks .
we subtracted an index - averaged linear trend obtained by linear regression from the @xmath309 s in order to obtain symmetric random walkers .
finally , in order to make the stocks comparable to our model of @xmath1 random walkers of the same jump distribution , we had to normalize the @xmath309 s by dividing through the standard deviations of the respective individual jump distributions .
after this detrending and normalization we can assume that the jump distributions in the individual time series have at least the same mean and the same variance , which should then be given by 0 and 1 .
for a fixed @xmath1 , we randomly selected subsets of size @xmath1 from the @xmath302 detrended and normalized stocks for each of the @xmath304 intervals of length @xmath311 and computed the evolution of the record number @xmath0 in these subsets . to get reliable statistics we average @xmath0 over @xmath269 different subsets with a fixed @xmath1 and also averaged over the @xmath304 consecutive intervals .
the resulting @xmath200 s for the upper and the lower record number and @xmath1 between @xmath44 and @xmath312 are given in fig .
( [ stocks.fig ] ) .
we find that both the curve for the upper and the curve for the lower mean record number are not in agreement with our theoretical prediction for the case of @xmath1 gaussian random walks given by @xmath313
. however , fig .
( [ stocks.fig ] ) shows , that the @xmath200 for the stocks increase with @xmath1
. we also considered subsets of size @xmath314 and found that for @xmath1 closer to the maximal value of @xmath302 , @xmath200 gets almost constant in @xmath1 .
while the increase of @xmath200 for smaller @xmath1 indicates that the statistics behave like @xmath1 independent gaussian random walks , the fact that it saturates for large @xmath1 could indicate that they behave like a lvy flight with lvy index @xmath273 .
we know however , that the tail exponent of the daily returns @xmath315 in the stock data is much larger than @xmath298 and that they definitely do not perform a lvy flight @xcite .
much more likely is that the correlations between the individual stocks play an important role .
in addition , we observed that at least for @xmath316 , @xmath200 grows proportional to @xmath317 .
one way to interpret this finding is the following : when we assume that in @xmath1 stocks only a smaller number of @xmath318 ( with @xmath319 and @xmath320 ) is effectively independent and that only these @xmath318 stocks contribute to the record statistics , the mean record number should be given by @xmath321 and saturate if the value of @xmath322 is achieved , where @xmath323 is the total number of stocks . in fig .
( [ stocks.fig ] ) we fitted curves of the form @xmath324 with @xmath325 and @xmath326 to the development of the upper and lower @xmath200 .
the good agreement with the fitted curves and the data confirms our assumption . apparently , the record statistics of @xmath1 detrended and normalized stocks is the same as the one of @xmath318 independent gaussian random walks
this finding is also confirmed by the inset in fig .
( [ stocks.fig ] ) .
there we plotted @xmath327 for different interval length @xmath2 and some different subset sizes @xmath1 .
the fact that all the lines collapse tells us that @xmath327 is independent of @xmath1 and therefore @xmath328
in conclusion we have presented a thorough analysis of the record statistics of @xmath1 independent random walkers with continuous and symmetric jump distributions . for @xmath9 ,
we have found two distinct cases : the case where the variance of the jump distribution @xmath4 is finite and the case where @xmath4 does not exist ( case ii ) as in the case of lvy random walkers with index @xmath155 . in the first case
we have found that the mean record number behaves like @xmath329 for @xmath330 while in case ii , @xmath331 for @xmath330 .
we have then argued that , in the first case , the full distribution of the scaled number of records @xmath16 is given by the distribution of the maximum of @xmath1 independent brownian motions with diffusion coefficient @xmath240 .
this statement was suggested by an exact result for lattice random walks and it was corroborated ( i ) by our exact calculation of the first moment @xmath270 valid for any value of @xmath1 and ( ii ) by our numerical simulations . of course it would be very interesting to obtain a proof of this result . from this connection with extreme value statistics , one
thus expects that the distribution of @xmath16 converges , for @xmath15 , to a gumbel form ( [ gumbel.1 ] ) .
this connection between record statistics and extreme value statistics could also be useful to compute the record statistics in other models discussed in the introduction , for instance , in the linear drift model @xcite . in the case of lvy
random walkers , we have shown numerically that the full distribution of @xmath270 converges , when @xmath15 , to a limiting distribution @xmath90 which is independent of @xmath18 . the exact computation of this universal distribution remains a challenging problem .
other interesting questions concern the extension of these results to include a linear drift @xcite or to the case of constrained lvy walks , like lvy bridges which were recently studied in the context of real space condensation phenomena @xcite .
finally the applications of our results to the record statistics of stock prices from the standard & poors 500 index suggest that , among a set of @xmath1 stocks , only a smaller number , which scales like @xmath332 , with @xmath333 , are effectively independent .
the record statistics of these @xmath332 stocks is then very similar to the statistics of @xmath332 independent random walkers .
this idea might be useful for future investigations of the fluctuations of such ensemble of stock prices .
0.3 cm * acknowledgments : * we acknowledge support by anr grant 2011-bs04 - 013 - 01 walkmat and the department of business administration and finance at the university of cologne for providing access to the stock data from the s&p 500 .
gw is grateful for the kind hospitality of the laboratoire de physique thorique et modles statistiques during the completion of this work and for the financial support provided by dfg within the bonn cologne graduate school of physics and astronomy .
we start from eqs . ( [ pm.2 ] ) and ( [ qm.2 ] ) . when @xmath4 is finite , by central limit theorem , the typical position of a walker after @xmath21 steps scales as @xmath334 for large @xmath21 .
hence the natural scaling variable is @xmath335 .
consider first eq . ( [ pm.2 ] )
satisfied by @xmath105 . to extract the leading scaling function in the scaling limit @xmath138 , @xmath139 but keeping @xmath335 fixed , we need to investigate @xmath124 , given explicitly in eq .
( [ ivanov.2 ] ) , in the limit when @xmath336 , @xmath337 but keeping the ratio @xmath338 fixed . to extract the behavior of @xmath124 in this scaling limit
, it is advantageous to work with an alternative expression of @xmath124 derived in ref . @xcite for finite @xmath4 @xmath339}\,\exp\left[-\frac{\lambda}{\pi}\ , \int_0^{\infty } \frac{dk}{\lambda^2+k^2}\ , \ln\left[\frac{1-s { \hat f}(k)}{1-s+ s\sigma^2k^2/2}\right]\right ] .
\label{phisl1}\ ] ] this expression is more suitable for extracting the scaling limit . in the limit @xmath336 and @xmath337 , the expression inside the exponential in eq .
( [ phisl1 ] ) tends to @xmath3 and hence , to leading order , we have @xmath340}\ , . \label{phisl2}\ ] ] inverting the laplace transform with respect to @xmath128 ( it has a simple pole at @xmath341 ) one gets from eq .
( [ pm.2 ] ) @xmath342 setting @xmath343 with @xmath344 in the scaling limit , the sum on the lhs of eq .
( [ genpm.1 ] ) can be approximated , to leading order , by a continuous integral : @xmath345 and we have @xmath346 next we need to invert the laplace transform with respect to @xmath347 .
we use the explicit inversion formula , @xmath348= \frac{b}{2\,\sqrt{\pi}\,m^{3/2}}\ , \exp[-b^2/4m]$ ] . applying this to eq .
( [ genpm.2 ] ) gives , to leading order , in the scaling limit @xmath349 \ : , \label{pms.1}\ ] ] which can be reorganized in the scaling form @xmath350 next we consider @xmath104 given in eq .
( [ qm.2 ] ) .
following exactly the same procedure as in the case of @xmath105 we find , in the scaling limit , @xmath351 .
\label{qms.1}\ ] ] inverting the laplace transform with respect to @xmath347 upon using the explicit inversion formula , @xmath352={\rm erfc}(b/\sqrt{4m})$ ] , we get , to leading order in the scaling limit @xmath353 which proves the result in eq .
( [ qm1scaling ] ) .
we consider jump distribution @xmath37 such that its fourier transform behaves , for small @xmath70 , as @xmath354 with @xmath12 . in this case
, the position of the walker after @xmath21 steps , grows as @xmath355 for large @xmath21 @xcite .
hence the natural scaling limit is @xmath138 , @xmath356 with the ratio @xmath144 fixed . for @xmath105 , we expect a scaling form @xmath357 . the power of @xmath21 outside the scaling function is chosen to ensure that @xmath358 .
this is needed since we know from eq .
( [ q0 ] ) and the sparre andersen theorem in eq .
( [ sa1 ] ) that @xmath359 for large @xmath21 .
similarly , for @xmath104 , we expect a scaling form @xmath360 in the scaling limit . to extract the leading scaling functions @xmath146 and @xmath147 respectively from eqs .
( [ pm.2 ] ) and ( [ qm.2 ] ) , we need to investigate the function @xmath124 in eq .
( [ ivanov.2 ] ) in the corresponding scaling limit @xmath336 , @xmath361 but keeping the ratio @xmath362 fixed .
fortunately , this was already done in ref .
@xcite in a different context . setting @xmath343 with @xmath344 , the leading behavior of @xmath124 in the scaling limit
is given by ( see eqs .
( 43)-(47 ) of ref .
@xcite ) @xmath363\right ] .
\label{phisl3}\ ] ] let us first consider the function @xmath105 in eq .
( [ pm.2 ] ) .
we substitute the anticipated scaling form @xmath364 on the lhs of eq .
( [ pm.2 ] ) . as before , setting @xmath365 , we can replace , in the scaling limit , the sum over @xmath21 by a continuous integral over @xmath21 @xmath366 we then make a change of variable @xmath367 and @xmath368 to get @xmath369 we next substitute eq . ( [ lhs.2 ] ) on the lhs of eq .
( [ pm.2 ] ) and eq .
( [ phisl3 ] ) on the rhs of eq .
( [ pm.2 ] ) .
writing the scaled variable as @xmath370 and comparing lhs with the rhs , we see that the @xmath371 cancels from both sides leaving us with @xmath372\right]\equiv j_\mu(w ) \;.
\label{g2s.1}\ ] ] similarly , by substituting the anticipated scaling form @xmath373 on the lhs of eq .
( [ qm.2 ] ) and doing exactly the same series of manipulations , we get @xmath374 where @xmath375 is defined in eq .
( [ g2s.1 ] ) . for later purposes , it is further convenient to define a pair of laplace transforms @xmath376 in terms of which eqs .
( [ g2s.1 ] ) and ( [ h2s.1 ] ) read @xmath377 the equations ( [ g2s.1 ] ) and ( [ h2s.1 ] ) determine , in principle , the two scalings functions @xmath146 and @xmath147 for all @xmath148 . in practice , it is hard to invert these two equations to obtain @xmath146 and @xmath147 for all @xmath148 .
however , it is possible to extract the large @xmath148 asymptotics of these two functions by analyzing the leading singular behavior of @xmath375 in eq .
( [ g2s.1 ] ) as @xmath378 .
clearly , it follows from the definition of @xmath375 in eq .
( [ g2s.1 ] ) that @xmath379 .
we are , however , interested in the leading singular correction term in @xmath375 as @xmath378 which , as it turns out , depends on whether @xmath152 , @xmath380 or @xmath278 .
below , we consider the three cases separately .
we consider @xmath375 in eq .
( [ g2s.1 ] ) and compute the derivative @xmath381 as @xmath378 .
simple computation shows that @xmath382 note that the integral defining @xmath383 is convergent as @xmath384 for @xmath152 . integrating over @xmath385 and using @xmath379 we get the leading correction term as @xmath378 @xmath386 where @xmath383 is given in eq .
( [ bmudef.1 ] ) . substituting this small @xmath385 behavior of @xmath375 on the rhs of eq .
( [ g21.lap ] ) , it follows that to match the powers of @xmath385 on both sides , the laplace transform @xmath387 must have the following small @xmath388 behavior @xmath389 using the classical tauberian theorem ( for a simple derivation see the appendix a.2 of ref .
@xcite ) , one immediately gets the following large @xmath148 behavior of @xmath146 @xmath390 with the amplitude @xmath391 similarly , substituting the small @xmath385 behavior of @xmath375 on the rhs of eq .
( [ h21.lap ] ) and matching powers of @xmath385 on both sides , we get @xmath392 once again , using the tauberian theorem of inversion , we get @xmath393 with the amplitude @xmath394 where @xmath395 is given in eq .
( [ amudef ] ) .
finally , note that the ratio @xmath396 is universal in the sense that it is independent of @xmath397 as well as on the scale factor @xmath88 . unlike in the previous case
, one finds that the first derivative of @xmath375 at @xmath399 is finite when @xmath400 $ ] and is given by @xmath401 note that for @xmath380 , the integral in eq .
( [ jmuder.1 ] ) is convergent as @xmath402 .
thus , as @xmath378 , @xmath403 . to obtain the leading non - analytic singular term , we need to compute the next term . by taking two derivatives with respect to @xmath385 near @xmath399 and then re - integrating back , we find the following leading singular behavior of @xmath375 near @xmath399 @xmath404 substituting this small @xmath385 behavior of @xmath375 on the rhs of eq .
( [ g21.lap ] ) and matching powers of @xmath385 on both sides we get @xmath405 where @xmath406 and @xmath407 are given respectively in eqs .
( [ jmuder.1 ] ) and ( [ jmu12 ] ) .
again , inverting via the tauberian theorem ( see ref .
@xcite ) , we get @xmath408 with the amplitude @xmath409 exactly in a similar way , we substitute the small @xmath385 behavior of @xmath375 on the rhs of eq .
( [ h21.lap ] ) , match powers of @xmath385 on both sides and find that @xmath410 where @xmath406 and @xmath407 are defined in eqs .
( [ jmuder.1 ] ) and ( [ jmu12 ] ) .
inverting via the tauberian theorem gives the desired result @xmath411 with the amplitude @xmath412 where @xmath395 is given in eq .
( [ amudef12 ] ) . in this case , also we note that the ratio @xmath413 is universal and does not depend explicitly on @xmath380 and @xmath88 . in this case , from eq .
( [ g2s.1 ] ) @xmath414 \quad { \rm where}\quad i_1(w)= \frac{1}{\pi}\int_0^{\infty } \frac{du}{1+u^2}\ , \ln ( 1+a\,w\,u ) \;. \label{j1w.1}\ ] ] let us first derive the leading singular behavior of @xmath415 as @xmath378 . making a change of variable @xmath416 in the integral we get @xmath417 next
, we divide the integration range @xmath418 into two parts @xmath419 $ ] and @xmath420 and write @xmath421 . the second part @xmath422 , i.e. , the integral over @xmath423 $ ] is a completely analytic function as @xmath378 .
thus the leading singular behavior of @xmath415 as @xmath378 is contained only in the first part @xmath424 in this integral , we can now safely expand @xmath425 and perform the integral term by term .
the leading singularity comes from the first term of this expansion @xmath426\underset{w\to 0}{\sim}- \frac{a}{\pi}\,w\ , \ln w \label{z1w.2}\ ] ] which indicates , from eq .
( [ j1w.1 ] ) , that @xmath427 substituting this small @xmath385 behavior of @xmath428 on the rhs of eq .
( [ g21.lap ] ) and matching the leading behavior of @xmath385 on both sides indicates that @xmath429 this indicates , using tauberian inversion theorem ( see ref .
@xcite ) , @xmath430 similarly , substituting the small @xmath385 behavior of @xmath375 on the rhs of eq .
( [ h21.lap ] ) and matching leading behavior of @xmath385 on both sides we get @xmath431 which , when inverted , provides the following large @xmath148 behavior @xmath432 finally , we notice that even for this marginal @xmath278 case , the ratio @xmath433 has the same value as in the other two cases , namely for @xmath152 and @xmath380 .
let us remark that if one puts @xmath278 in the expression of @xmath395 in eq .
( [ betamu2 ] ) , we get @xmath434 .
correspondingly @xmath435 from eq . ( [ amu1 ] ) and @xmath436 from eq . ( [ bmu1 ] )
, we find that they are consistent respectively with @xmath437 in ( [ g22.largez1 ] ) and and @xmath438 in ( [ h22.largez1 ] ) .
in other words , the final asymptotic results for @xmath146 and @xmath147 in the marginal case @xmath278 are included in the range @xmath439 $ ] , even though the details for @xmath278 are quite different , as it has logarithmic singularities .
in this appendix we consider @xmath1 lattice random walks ( rw ) starting at @xmath440 , for @xmath441 and evolving as @xmath442 where the noise @xmath57 s are i.i.d .
random variables with a distribution @xmath203 .
the aim is to show the result in eq .
( [ exact_bernoulli ] ) , taking advantage of the relation ( [ identity ] ) .
we first consider a single random walk , @xmath49 , and denote by @xmath443 the number of lattice rw starting at @xmath444 and ending in @xmath445 after @xmath2 steps .
one has @xmath446 to compute the cumulative distribution function ( cdf ) of the maximal displacement of @xmath1 walkers we need to compute the number of walks , for a single walker @xmath49 , which stay strictly below a given value @xmath447 .
we thus denote , for @xmath49 , by @xmath448 the number of walks which stays strictly below an integer @xmath447 and end up in @xmath445 after @xmath2 steps .
to do this we use the reflection principle , e. g. the method of images : @xmath448 can be obtained by subtracting to @xmath443 the number of free walks which start in @xmath449 and end in @xmath445 after @xmath2 steps .
this yields : @xmath450 the total number of walks @xmath451 which start at @xmath444 and stay strictly below @xmath447 after @xmath2 steps are obtained by summing @xmath443 in eq .
( [ reflection_1 ] ) over the endpoint @xmath452 .
this yields @xmath453 \;,\end{aligned}\ ] ] where @xmath212 is the largest integer not greater than @xmath100 .
therefore one has @xmath454 = \frac{w_m(n)}{2^n } = \frac{1}{2^n } \sum_{k=0}^{\lfloor \frac{n+m}{2}\rfloor } \left[{n \choose k } - { n \choose k - m}\right ] \;.\end{aligned}\ ] ] we can now write the cdf of the maximal displacement of @xmath1 independent walkers as @xmath455 = \left(\frac{w_m(n)}{2^n}\right)^n = \left(\frac{1}{2^n } \sum_{k=0}^{\lfloor \frac{n+m}{2}\rfloor } \left [ { n \choose k } - { n \choose k - m}\right]\right)^n \;,\end{aligned}\ ] ] where @xmath456 , from which one gets @xmath457 = \frac{1}{2^{nn } } \left([w_{m+1}(n)]^n - [ w_m(n)]^n \right ) \;.\end{aligned}\ ] ] finally , using the identity ( [ identity ] ) , one obtains the result given in the text in eq .
( [ exact_bernoulli ] ) . | we study the statistics of the number of records @xmath0 for @xmath1 identical and independent symmetric discrete - time random walks of @xmath2 steps in one dimension , all starting at the origin at step @xmath3 . at each time step
, each walker jumps by a random length drawn independently from a symmetric and continuous distribution .
we consider two cases : ( i ) when the variance @xmath4 of the jump distribution is finite and ( ii ) when @xmath4 is divergent as in the case of lvy flights with index @xmath5 . in both cases
we find that the mean record number @xmath6 grows universally as @xmath7 for large @xmath2 , but with a very different behavior of the amplitude @xmath8 for @xmath9 in the two cases .
we find that for large @xmath1 , @xmath10 independently of @xmath4 in case i. in contrast , in case ii , the amplitude approaches to an @xmath1-independent constant for large @xmath1 , @xmath11 , independently of @xmath12 . for finite @xmath4
we argue , and this is confirmed by our numerical simulations , that the full distribution of @xmath13 converges to a gumbel law as @xmath14 and @xmath15 . in case ii , our numerical simulations indicate that the distribution of @xmath16 converges , for @xmath14 and @xmath17 , to a universal nontrivial distribution , independently of @xmath18 .
we discuss the applications of our results to the study of the record statistics of 366 daily stock prices from the standard & poors 500 index . |
charm sensitivities have increased exponentially over the last two decades .
current experiments aim to reconstruct @xmath4 events , and the @xmath5 factories and compass facility@xcite could achieve @xmath6-event sensitivity .
we are designing an experiment for the tevatron s c0 area which could reconstruct @xmath7 charm decays during tevatron run ii ( ca .
20002005 ) .
while this `` btev / c0 '' effort aims at both charm and beauty physics , i focus here on charm .
sensitivity at the proposed level will substantially advance such programmatic " charm physics as spectroscopy , lifetimes , and qcd tests .
it will also give substantial new - physics reach in the areas of _ cp _ violation , flavor - changing neutral - current and lepton - number - violating decays , and @xmath2 mixing . if direct _
violation occurs in cabibbo - suppressed @xmath3 decays at the level predicted by the standard model , it could be observable in the btev / c0 experiment .
violation is recognized as one of the central problems of particle physics .
the mechanism(s ) responsible for it have yet to be definitively established . a leading candidate , the kobayashi - maskawa ( km ) model,@xcite has the attractive feature of explaining the small size of @xmath8 asymmetries as a manifestation of the small mixing between the third quark generation and the first two.@xcite thus in the km model , large asymmetries are expected in the beauty sector .
other models attribute the effect to the exchange of massive particles such as @xmath9 s with right - handed coupings or extra higgs scalars.@xcite in these models asymmetries should be more democratic " and may be too small to observe in beauty ( @xmath10@xmath11 ) .
many of these models predict large mixing in charm
. we do not know whether violation arises exclusively from any one of these mechanisms , whether many contribute , or whether some other mechanism not yet thought of is the answer .
thus a balanced program of investigation in all available quark ( and lepton@xcite ) sectors is desirable .
as is well known , -violation , mixing , and rare - decay studies in beauty are the goal of several projects in progress around the world
. such studies in charm are important precisely because the small standard model predictions can allow new physics to appear in a striking manner .
direct _ cp _ violation in charm decay is expected in the standard model ( sm ) at the @xmath12 level@xcite ( see table [ tab : sens ] ) . in the sm it is significant only for singly cabibbo - suppressed decays ( scsd ) , for which tree - level graphs can interfere with penguin diagrams , leading to partial - decay - rate asymmetries : @xmath13 where @xmath14 is the decay width for a @xmath3 meson to final state @xmath15 and @xmath16 that for the _ cp_-conjugate process . [ cols= " < , < , < , < , < " , ] @xmath17 to save space , sources for the measurements and predictions in this table are not cited here ; most may be found in refs.@xcite and @xcite.@xmath18 at 90% confidence level .
these asymmetries reflect interference due to the ckm phase in combination with phase differences from final - state interactions .
experimental evidence suggests substantial final - state effects in charm decay .
for example , the mode @xmath19 occurs with a branching ratio@xcite @xmath20 even though no spectator diagram can produce this final state , and the two possible @xmath9-exchange diagrams cancel each other ( by the gim mechanism ) to good approximation .
this mode could be fed by rescattering of @xmath21 into @xmath22 .
large final - state effects are also evident in the case of multibody charm decays , where dalitz - plot analyses reveal appreciable phase differences.@xcite these and similar observations underlie the expectation of @xmath10@xmath11 direct asymmetries in charm .
additional sm mechanisms for charm violation include @xmath8 mixing and possible mixing with glueballs or gluonic hybrids . as emphasized by xing,@xcite @xmath8
mixing leads to _ cp _ asymmetries of @xmath232re@xmath24 in such decays as @xmath25 and @xmath26 . while perhaps not as interesting as direct charm violation
, this effect might provide a calibration for systematic effects in the measurement of small asymmetries . as discussed below
, it could also represent a unique window into new physics .
close and lipkin@xcite make the intriguing suggestion that @xmath3 s could be mixed with gluonic - hybrid states , with consequent large -violating effects . at present
the best limits on direct _ cp _ violation in cabibbo - suppressed charm decay come from fermilab e687@xcite and e791@xcite and cleo@xcite ( table [ tab : sens ] ) . in fixed - target experiments , to correct for the production asymmetry of @xmath3 _ vs. _ @xmath27 , the asymmetry in a cabibbo - suppressed mode is normalized to that observed in the corresponding cabibbo - favored ( cfd ) mode ; this also has the effect of reducing sensitivity to such systematic effects as trigger , reconstruction , and particle - identification efficiency differences for particles _ vs. _ antiparticles .
in e687 @xmath2310% sensitivity is achieved . by extrapolation from e687 , the definitive establishment of a @xmath12 asymmetry
requires @xmath2810@xmath29 reconstructed @xmath3 s , to give @xmath3010@xmath31 reconstructed charged and ( tagged ) neutral @xmath3 s in scsd modes .
although the ratiometric nature of these measurements makes them intrinsically insensitive to systematic effects , at the sub-@xmath12 level careful attention will be required to keep systematic uncertainties from dominating . for several reasons ,
the charm sector is an excellent place to look for _ cp _ violation arising from physics beyond the standard model : * the top - quark loops that in the standard model dominate violation in the strange and beauty sectors@xcite are absent , creating a low - background window for new physics .
* new physics may couple differently to up - type and down - type quarks@xcite or couple to quark mass.@xcite * compared to beauty , the large production cross sections@xcite allow much larger event samples to be acquired , and the branching ratios to final states of interest are also larger.@xcite * many extensions of the standard model predict observable effects in charm . direct violation in cabibbo - favored or doubly cabibbo - suppressed ( dcsd ) modes would be a clear signature for new physics.@xcite asymmetries in these as well as in scsd modes could reach @xmath3010@xmath32 in such scenarios as non - minimal supersymmetry@xcite and in left - right - symmetric models.@xcite bigi has pointed out that a small new - physics contribution to the dcsd rate could amplify the sm @xmath8-induced asymmetries to @xmath10@xmath33 as well.@xcite many authors have recently emphasized the possibility of observable indirect violation in charm.@xcite@xmath34@xcite this of course depends on charm mixing , which has not been established experimentally.@xcite however , the observation of a wrong - sign signal ( which may be mixing , dcsd , or some mixture of the two ) at cleo@xcite has stimulated theorists to consider the large variety of extensions to the sm in which @xmath35 and @xmath36 can display appreciable -violating mixing .
these include flavor - changing higgs exchange , a fourth generation , @xmath37-mediated fcnc s , left - right symmetry , supersymmetry with quark - squark alignment , leptoquarks , etc.@xcite at the level discussed in the literature , such effects are likely to be observable in a @xmath38-to-@xmath7-charm experiment . as a specific example , i consider a possible indirect asymmetry in @xmath39 .
we would expect @xmath40 tagged @xmath41 decays per few@xmath42 reconstructed charm , giving @xmath43 sensitivity . since here the final state is a eigenstate , the asymmetry is independent of final - state phases and thus directly measures the new - physics phase.@xcite
the btev / c0 experiment will have unprecedented reach in all areas of charm physics , including tests of qcd and hqet in meson and baryon spectroscopy and lifetimes , charm production , dalitz - plot analyses , semileptonic form factors , extraction of ckm elements , etc .
space restrictions preclude further discussion here .
fig . 1 is a sketch of the btev / c0 spectrometer as currently conceived .
the spectrometer is designed for the tevatron c0 collider interaction hall , to be upgraded and expanded during the 1998 main injector construction period .
it differs from existing collider experiments in that it focuses on heavy - quark states produced in the
forward " direction ( @xmath440.3 ) .
this approach allows optimal decay - time resolution and is also advantageous for hadron identification .
the apparatus must have high interaction - rate capability , large acceptance , an efficient charm trigger , high - speed and high - capacity data acquisition , good mass and vertex resolution , and good particle identification .
of these requirements , the most challenging are the trigger and the particle identification .
we intend to trigger primarily on the presence of a decay vertex separated from the primary vertex.@xcite to reduce occupancy and facilitate vertex reconstruction at trigger level 1 , pixel detectors will be used for vertex reconstruction . for efficient , reliable , and compact particle identification
, we will build a ring - imaging cherenkov counter .
in other respects the spectrometer will resemble existing large - aperture fixed - target heavy - quark experiments .
our charm sensitivity goal might be achieved in either collider or fixed - target mode . during tevatron run ii , much early running at c0 may be in fixed - target mode , _
e.g. _ using a wire target in the beam halo .
we have estimated@xcite the fixed - target reconstructed - event yield at @xmath45 per @xmath6 seconds of running for an experiment operating at a 1-mhz interaction rate . given the higher production cross section , comparable or greater sensitivity could be available in collider mode even with reduced running time . in addition , we anticipate increasing the interaction rate beyond 1mhz as the tevatron bunch separation is reduced from 396 to 132ns .
ultimately @xmath235 mhz could be feasible , leading to @xmath46 charm decays reconstructed .
whether fixed - target or collider mode is better for charm physics is a detailed question which probably can not be answered definitively until data are taken .
for one thing , the forward charm - production cross section has not yet been measured in 2tev @xmath47 collisions .
there are also subtleties , for example biases in mixing studies may arise from @xmath48 cascade decays .
these would be suppressed by two orders of magnitude in fixed - target relative to collider mode , due to the reduced beauty production cross section .
a hadroproduction experiment capable of reconstructing @xmath46 charm events is feasible using detector , trigger , and data acquisition technologies that exist or are under development .
such an experiment could observe direct _ cp _ violation in charm decay at the level expected in the standard model and substanially extend the discovery reach for new physics .
m. golden and b. grinstein , phys .
b * 222 * , 501 ( 1989 ) ; f. buccella _ et al .
_ , phys .
b * 302 * , 319 ( 1993 ) ; a. pugliese and p. santorelli , in _ proc .
third workshop on the tau / charm factory _ , marbella , spain , 16 june 1993 , edition frontieres ( 1994 ) , p. 387 ; f. buccella _ et al .
_ , phys .
d * 51 * , 3478 ( 1995 ) ; f. buccella , m. lusignoli , and a. pugliese , phys .
b * 379 * , 249 ( 1996 ) .
g. burdman , in * the future of high - sensitivity charm experiments * , d. kaplan and s. kwan , _ eds .
_ , fermilab - conf-94/190 ( 1994 ) p. 75 , and in * workshop on the tau / charm factory * , j. repond , _ ed .
_ , argonne national laboratory , june 2123 , 1995 , aip conf.proc .
349 ( 1996 ) , p. 409 .
p. frabetti _ et al .
_ , phys .
b * 331 * , 217 ( 1994 ) ; j. e. wiss , in * workshop on the tau / charm factory * , argonne national laboratory , june 2123 , 1995 , j. repond , _ ed .
_ , aip conf .
349 ( 1996 ) , p. 345 .
a. le yaouanc , l. oliver , and j .- c .
raynal , phys .
b * 292 * , 353 ( 1992 ) ; m. gronau and s. wakaizumi , phys . rev . lett . * 68 * , 1814 ( 1992 ) ; a. le yaouanc _ et al .
_ , lpthe - orsay/95 - 15 , hep - ph/9504270 ( 1995 ) .
d. husby _ et al .
_ , in _ proc .
5th int .
conf . on electronics for particle physics _
, g. j. blanar and r. l. sumner , _ eds .
_ , lecroy corp .
( 1995 ) ; w. selove , in _ proc .
workshop on b physics at hadron accelerators _ , c. s. mishra and p. mcbride , _ eds .
_ , fermilab - conf-93/267 ( 1993 ) , p. 617 ; r. isik _ et al . _ , u. of pa
. report upr-234e ( 1996 ) . | the btev / c0 " experiment at fermilab could reconstruct @xmath010@xmath1 charm decays , four orders of magnitude beyond the largest extant sample .
the experiment is likely to run during tevatron run ii ( ca .
20002005 ) .
in addition to
programmatic " charm physics such as spectroscopy , lifetimes , and qcd tests , it will have significant new - physics reach in the areas of _ cp _ violation , flavor - changing neutral - current and lepton - number - violating decays , and @xmath2 mixing , and could observe direct _ cp _ violation in cabibbo - suppressed @xmath3 decays if it occurs at the level predicted by the standard model .
15000 epsf |
the discovery of extended flat rotation curves in the outer parts of disk galaxies three decades ago @xcite was decisive in ushering the paradigm shift that led to the now standard cosmological model dominated by cold dark matter ( cdm ) and dark energy .
on cosmological ( i.e. linear ) scales there is excellent agreement between the predictions of the standard @xmath3cdm model and observations of the cmb , type ia supernovae , weak - lensing , and galaxy clustering ( e.g. * ? ? ?
* ) . on galactic and sub - galactic ( i.e. non - linear ) scales
, however , there are a number of apparent discrepancies between the predictions and observations ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) of the structure and mass function of dark matter haloes .
whether these problems reflect an incomplete understanding of galaxy formation , or signal an unresolvable problem for the standard paradigm , remains to be determined .
measuring the density profiles of dark matter halos on galaxy scales is thus a crucial test for the standard paradigm of galaxy formation , offering great potential for discovery . in most luminous galaxies the baryons ( i.e. stars and cold gas )
always make a non - negligible contribution to the gravitational potential , and due to the so - called _ disk - halo degeneracy _
@xcite the distribution of baryons and dark matter in galaxies is poorly constrained , even when high spatial resolution rotation curves are available ( e.g. * ? ? ?
* ; * ? ? ?
[ cols="^,^,^,^,^,^,^,^,^,^ " , ] all the swells deflectors show a clear disky component ( figure [ fig : cmontage ] ) , confirming our original selection based on sdss images .
furthermore , the targets have high projected ellipticity , again consistent with our original selection based on sdss images ( figure [ fig : hist ] ) .
many of the targets show clear spiral arms , which in several cases are blue .
other targets are so close to edge - on that it is hard to identify spiral arms conclusively ( notes on the morphology of individual targets are provided in the previous section ) .
the overall morphology of the targets is also consistent with the distribution of velocity dispersions ( figure [ fig : hist ] ) ; this distribution has a lower central value than the slacs sample , which is dominated by massive elliptical galaxies .
we quantify the structural properties of the deflectors by performing a bulge - disk decomposition . for the purpose of this paper
we adopt a simple parametrisation , and describe the bulge with an elliptical @xcite profile and the disk with an elliptical exponential profile .
this parametrisation is the same as that adopted by simard et al .
( 2011 , in preparation ) to analyse the entire sdss - dr7 database , and therefore is ideal for a comparison between lenses and non - lens galaxies . beyond this first analysis ,
in future papers we will derive more accurate descriptions of the surface brightness distribution by considering more flexible models , like , e.g. , the sersic profile for the bulge ( e.g. paper ii ) , and by correcting for extinction due to prominent dust lanes ( brewer et al . 2011 , in preparation ) . for each deflector
we compute the posterior probability distribution function of the parameters by comparing the models with the multi - colour images using the dedicated code developed by one of us ( mwa ; see also * ? ? ?
replaces the functionality of standard codes like , by enabling a full exploration of the likelihood surface as well as joint analysis of multiband data .
furthermore , spasmoid models fit the multi - band imaging simultaneously with the same effective radius , and the bulge effective radius is constrained to be smaller than the disk effective radius .
this ensure self - consistent colors even with data of varying depth as a function of wavelength .
the most probable values of the parameters are given in table [ tab : structure ] . ) , the middle panel shows the distribution of deflector redshifts , and the bottom panel shows the distribution of axis ratios .
systems discovered prior to cycle 16s are plotted in red , our new systems are shown in black .
solid histograms represent secure lenses ( a ) , hatched histograms represent probable and possible lenses ( b and c ) , while open histograms represent non - lenses ( x ) .
note that our cycle 16s strategy targeted preferentially lower velocity dispersion and axis ratio systems , in order to select disk edge - on galaxies .
note also how the confirmation rate increases with @xmath4 , as expected .
[ fig : hist ] ] .
the black lines represent the total mass density profile , the red lines represent the contribution of the bulge , and the blue lines represent that of the disk .
profiles for both the salpeter and chabrier imf are shown .
the top panel shows the profiles of secure ( a ) lenses , the middle panel shows the profile of probable ( b ) and possible ( c ) lenses , and the bottom panel shows the non - lenses ( x ) .
note that the bulge always dominates the enclosed average mass at small radii , while the disks start to provide a non - negligible contribution on scales of a kpc or larger .
[ fig : kappaprof],title="fig : " ] .
the black lines represent the total mass density profile , the red lines represent the contribution of the bulge , and the blue lines represent that of the disk .
profiles for both the salpeter and chabrier imf are shown .
the top panel shows the profiles of secure ( a ) lenses , the middle panel shows the profile of probable ( b ) and possible ( c ) lenses , and the bottom panel shows the non - lenses ( x ) .
note that the bulge always dominates the enclosed average mass at small radii , while the disks start to provide a non - negligible contribution on scales of a kpc or larger .
[ fig : kappaprof],title="fig : " ] .
the black lines represent the total mass density profile , the red lines represent the contribution of the bulge , and the blue lines represent that of the disk .
profiles for both the salpeter and chabrier imf are shown .
the top panel shows the profiles of secure ( a ) lenses , the middle panel shows the profile of probable ( b ) and possible ( c ) lenses , and the bottom panel shows the non - lenses ( x ) .
note that the bulge always dominates the enclosed average mass at small radii , while the disks start to provide a non - negligible contribution on scales of a kpc or larger .
[ fig : kappaprof],title="fig : " ] , the red histogram represents lenses discovered prior to cycle 16s ; the black solid , hatched and empty histograms represent lenses targeted in cycle 16 , identified respectively as secure , probable and possible , and not - lenses . note how no lens was confirmed with stellar einstein radius below @xmath5 .
[ fig : histo2 ] ] stellar mass estimates for the bulge and disk components of each candidate were derived by comparing multiband photometry with stellar population synthesis models @xcite .
the method we employ provides not only stellar masses but also uncertainties that take into account all the relevant degeneracies between , e.g. , age and metallicity of the stellar populations .
the dominant systematic uncertainty is the unknown normalisation of the stellar initial mass function ( imf ) .
determining the absolute normalisation of the imf is one of the goals of the swells survey ( see paper ii ) . for the moment ,
we adopt two popular choices for the imf , which are believed to bracket the appropriate range for spiral galaxies @xcite .
stellar masses for these @xcite and @xcite imfs are given in table [ tab : structure ] .
the overall success rate for confirmation of swells candidates in cycle 16s is @xmath6% ( 8/19 ) for a - grade and @xmath7% ( 9/19 ) for a+b - grade lenses .
as expected , this is somewhat lower than the confirmation rate for massive early - type galaxies in the slacs survey , due to the fact that our galaxies are on average less massive : the einstein radii are smaller and occupy a smaller fraction of the sdss fiber . furthermore , identifying multiple images is inherently more difficult behind spiral deflectors , due to the complexity of their surface brightness distribution and the presence of dust . we can investigate the selection function more quantitatively by studying the stellar mass structure of the deflectors and correlations of its properties with the confirmation rate . mass and surface mass density are expected to be the key parameters determining the lensing strength of a galaxy .
we will therefore investigate the dependency of confirmation rate on several proxies for mass and density .
first , in figure [ fig : kappaprof ] we show the average stellar mass within circular apertures , expressed in units of the critical density @xmath8 , where @xmath9 @xmath10 and @xmath11 are the angular diameter distance to the source , to the deflector and between the deflector and the sources , respectively ( see , e.g. , * ? ? ?
as shown in figure [ fig : kappaprof ] , the enclosed average stellar mass density is always dominated by the bulge in the inner parts , although the disk starts to contribute significantly beyond 1 kpc .
all of the secure systems have deflectors of uniformly high density , above critical density well beyond one kpc , even for a chabrier imf .
the not - lenses span a broader range in surface mass densities including several systems that are significantly underdense compared to the grade - a systems .
three out of four of the x - grade lenses do not have a detectable bulge component ( to our sensitivity limit of m@xmath12m@xmath2 ) .
in a circularly symmetric system the einstein radius corresponds to the radius within which the average mass density equals the critical density @xcite . in order to obtain a proxy for the einstein radius
even when no lensing is detected we define the `` stellar einstein radius '' as the radius within which the average stellar mass density equals the critical density .
note that high resolution multi - band imaging is necessary to estimate this quantity robustly .
clearly this is only an approximation for non - circular systems , and will be a lower limit to the true einstein radius if dark matter is present .
we are now in a position to study the distribution of integrated stellar mass @xmath13 , bulge - to - total stellar mass ratio @xmath14 and stellar einstein radii for the swells targets .
as shown in figure [ fig : histo2 ] all three quantities correlate at some level with confirmation rate .
most of the not - lenses are found at small @xmath13 , small @xmath14 , and small stellar einstein radius .
however , the correlation is not perfect : b and c grade lenses are found even at m@xmath15 and b / t@xmath16 .
all three objects with no detected bulge turned out to be not - lenses .
perhaps the cleanest predictor of confirmation is stellar einstein radius : there are no confirmed lenses below @xmath5 even though there are probable and possible lenses well above @xmath17 . in conclusion , the selection function of the new swells targets is complex and comprised of multiple steps . in the first one , we preselect based on sdss images edge - on late type galaxies
this is extremely efficient in producing a sample of lenticular and spiral galaxies .
the second step , confirmation as a lens system , depends on the properties of the deflector but also has a stochastic component .
denser systems produce larger einstein radii , which in turn make the lensing effect more likely and easier to detect at resolution .
in addition , the relative position on the sky of the source and deflector , as well as the broad band surface brightness of the source , ultimately control the presence of detectable multiple imaging .
a third layer of complication is added by the presence of dust lanes and small scale structure in the surface brightness distribution of the deflector , which may complicate identification of multiple images even for a perfectly good strong lensing system . in the next section
we will investigate the effective selection function of the swells sample by comparing the properties of the swells deflectors with those of a typical sample of spiral galaxies .
understanding the selection function of any observational sample is essential , both in order to generalise results to its parent population , as well as to compare across samples . in the case of the swells sample the selection function of the deflectors is sufficiently complicated that it is extremely hard to compute it accurately from first principles .
we therefore take an alternative approach and reconstruct the basic features of the selection function _ a posteriori _ by comparing the properties of the slacs detectors to those of a suitable comparison sample of non - lens galaxies
. the comparison sample should be as close as possible in terms of controllable galaxy parameters , so that significant differences might be interpreted unambiguously as due to the lensing selection effect .
it is particularly useful to perform this comparison for both confirmed lenses and not - confirmed lenses as this will help separate the effects of our sdss - based pre - selection from those related to the ability of the deflectors to produce detectable strong lensing events . among the quantities that are well - studied and available for large samples of galaxies , stellar mass and size ( and therefore density )
are particularly interesting and likely to influence the lensing selection function .
we therefore choose to focus on the size - stellar mass correlation as our comparison tool ( for a similar analysis of the slacs sample , based on the fundamental plane relation see * ? ? ?
* ) . a suitable comparison sample for the swells
dataset is its parent sample , the sdss - dr7 database itself .
we adopt the results of the recent structural analysis performed by l. simard and collaborators ( 2011 , in preparation ) .
these authors fit two - dimensional models to the sdss images , using the same parametrisation for bulge and disk as was adopted in this study .
the range in redshift and stellar mass of their sample is very similar to ours , and the parent sample is the same . to minimise differences with our sample ,
we construct the size mass relation for the sdss sample limited to @xmath180.6 .
following @xcite , we construct the size - mass relationship separately for the bulge , disk , and total size , adopting the semi - major axis size for all three components .
in contrast to @xcite we do not apply any selection based on stellar mass or spectral type . to minimise the impact of measurement biases
we perform the comparison by also using sizes and stellar masses estimated using sdss images for our sample of lenses dex ) , consistent with the findings of @xcite .
in contrast , the -determined sizes are on average @xmath19 ( disk ) and @xmath20 dex ( bulge ) smaller than those determined from sdss images , with larger differences for the smaller objects .
these differences are likely due to the higher spatial resolution of and/or a longer wavelength range used for our fits . ] . the size - mass relation for sdss and for our deflector galaxies
is shown in figure [ fig : sizemass ] .
the top panel shows the relation between the total half light radius and total stellar mass .
the middle panel shows the relation between the half light radius of the disk and the total stellar mass .
the bottom panel shows the relation between the half light radius of the bulge and the total stellar mass . as a population ,
the deflector galaxies are consistent with being drawn from the sdss parent population .
we quantify this statement by comparing the vertical offset of each data point with the mean and standard deviation of the correlation , and computing a @xmath21 goodness of fit statistic .
we find @xmath22 respectively for total , disk , and bulge , with 27 degrees of freedom for total and bulge and 26 for disk ( one galaxy was omitted because of an unphysically small size returned by the sdss catalogue , @xmath23 kpc ) .
this shows that our selection of targets _ at fixed stellar mass _ is unbiased with respect to the parent sdss population once the @xmath24 cut is taken into account .
clearly however , the distribution along the size - mass relation is very different than that of the parent sample , since our target list was comprised almost exclusively of systems with mass above 10@xmath25 m@xmath2 by design .
it is also interesting to consider whether the distribution of confirmed lenses is consistent with that of the parent population , or whether the confirmation rate appears to be a function of , e.g. , stellar mass density . by comparing the solid ( grade a lenses ) and open points ( grade b , c , x ) in figure [ fig : sizemass ]
, it appears that the solid points tend to lie preferentially somewhat below the open points .
however , the difference is not significant .
even if we restrict to only the a grade lenses , they still appear to be consistent with the size mass relation of the sdss parent population , at fixed stellar mass ( @xmath26 for 16 degrees of freedom , respectively for total , disk , and bulge component ) . .
the points represent the swells galaxies : solid squares are secure lenses ( a ) , open stars are probable and possible lenses ( b and c ) , while open squares are not lenses ( x ) .
red symbols identify pre - cycle 16s targets , while black symbols identify cycle 16s targets .
the top panel shows the total semi - major axis effective radius as a function of total stellar mass .
the middle panel shows the semi - major axis effective radius of the disk component as a function of total stellar mass .
the bottom panel shows the semi - major axis effective radius of the bulge component as a function of total stellar mass .
we use the symbol @xmath27 to identify the semi - major axis effective radius as measured from sdss images , to distinguish it clearly from the circularized half - light radius measured from hst images and given in table [ tab : structure ] .
[ fig : sizemass ] ]
finding spiral lens galaxies is more difficult than finding early - type lens galaxies for two reasons : 1 ) spiral galaxies tend to be less massive and less dense than early - type galaxies , and they therefore produce smaller image separations ; 2 ) their complex surface brightness distribution and the presence of dust makes it hard to identify multiple images if they are present . for these reasons
a crucial element for the success of this search was the availability of multi - colour high resolution imaging data .
however , for any given candidate it is not clear _ a priori _ which band will be most useful . in general ,
the slacs sources are by selection star - forming and therefore blue ( see , e.g. , * ? ? ?
* ) , and stand out the most in the f450w filter . a good example of this phenomenon is j1117@xmath284704 , where the quadruply imaged source near the centre gradually disappears moving to longer wavelengths . in some cases , however , the presence of dust makes the infrared data crucial for studying the multiple images .
an example of this phenomenon is j1703@xmath282451 , where the lensed source is barely visible in the optical images , while it is obvious in the infrared one .
interestingly , two of the lenses were confirmed purely on the basis of @xmath29 ao imaging .
it appears that multiband photometry is an essential ingredient for future studies . at optical wavelengths
is the only option for the time being . with current technology ao
is only feasible for @xmath301/3 of high latitude targets , and therefore a combination of ao ( when possible ) and ( when ao is not possible ) observations seems to be the most cost - effective strategy for the infrared .
the other crucial ingredient is angular resolution .
this is useful especially for the detection of compact sources or central counter - images .
limited resolution is one of the reasons why we do not confirm any lens with stellar einstein radius significantly smaller than one arcsecond .
a good example of this phenomenon is j0841@xmath283824 , where the counter - image is clearly visible in @xmath31-band , but not at the lower resolution @xmath32-band , even though this may in part be due to colour variations .
the wfpc2 images obtained in cycle 16s significantly under - sample the native psf .
better - sampled images with acs or wfc3-uvis may result in an improved success rate .
higher resolution red / infrared images from next generation advanced optics or from the james webb space telescope might combine the advantages of dust removal and angular resolution , even though colour contrast may still make blue / uv images competitive .
alternatively one may restrict the search to the systems with larger stellar einstein radii to maximize success .
however , high - resolution multiband images are needed to estimate this quantity accurately and therefore it does not appear a viable strategy for a large sample .
furthermore , one has to be mindful of the astrophysical implications of such a selection against objects where dark matter contributes significantly to the central mass density or where the imf normalization is higher than normally assumed . not to mention that finding lenses with small einstein radii is astrophysically interesting in its own right , in order to probe the structure of the lower mass galaxies and those with smaller bulges . in terms of future prospects
it seems that another way to improve the success rate would be to target emission lines in the sources at high resolution .
this would be an efficient way to disentangle light from the source and deflector , and distinguish without any doubt multiple images from peculiar features in the surface brightness of the deflector .
the rich filter complement of wfc3 aboard makes it the most promising instrument for this purpose .
targeting infrared lines might mitigate the effects of foreground extinction , although with the present targets only paschen @xmath0 is accessible with decent strehl and the line is typically too faint for present day integral field spectrographs , even on large 8 - 10 m telescopes .
a tunable filter or custom - made filters will be necessary to pursue this goal .
alternatively , one could target lines in the radio mapping co transitions or hi .
alma and ska will have the resolution and sensitivity to measure emission line maps of the source _ and _ lens , and will thus also offer the ability to measure kinematics of the lens and of the source .
molecular gas tends to trace the stars , so any source that is visibly lensed should also be lensed in co. neutral gas tends to be more extended ( typically by a factor of 2 or more ) than stars and molecular gas so this will increase the chances of strong lensing .
the goal of the swells ( sloan wfc edge - on late - type lens survey ) is to study the distribution of luminous and dark matter in disk galaxies . with this goal in mind
we have assembled an unprecedented sample of candidate gravitational lens systems , where the deflector is a high inclination disk galaxy .
the sample is comprised of systems previously discovered by the slacs survey for which we have obtained follow - up multiband high resolution imaging , and 19 new targets .
the new targets were selected from the sdss database to have multiple redshifts in the sdss fiber , as well as elongated morphology .
we present multi - colour imaging of the sample , as well as structural parameters derived from this data for the whole sample .
we then study the properties of the deflectors and compare them to a sample of non - lens galaxies selected from sdss with the same criteria as our lens targets .
our main results can be summarised as follows : 1 . of the 19 new targets ,
are classified as a - grade ( secure ) lenses , is classified as b - grade ( probable ) , are classified as c - grade ( possible ) , and are classified as x - grade ( not lenses ) .
our global success rate in cycle 16s was @xmath33 for a - grade lenses .
the total number of a - grade lenses in the swells sample is . 2 .
all the deflector galaxies have prominent disks observed at high inclination , by design . however , the sample spans a broad range of morphological and spectral types .
it includes early - type spirals and lenticulars as well as later - type spirals , galaxies with strong h@xmath0 emission and galaxies with no detectable h@xmath0 emission .
the stellar bulge to disk ratio of confirmed lenses range between 0.22 and 0.85 , with an average of 0.55 .
all confirmed lenses have stellar masses above 10@xmath25 m@xmath2 and stellar einstein radii above @xmath5 .
none of the low stellar mass density targets is confirmed to be a strong lens , consistent with the idea that stellar mass ( as opposed to dark matter ) dominates the mass density profile at the scales of a few kpc .
the swells targets and confirmed lenses follow the same size - stellar mass relation as a comparison sample of high - elongation ( @xmath24 ) non - lens galaxies selected from the sdss survey .
this implies that , _ at given stellar mass _ , swells lenses are consistent with being representative of the overall population of high - elongation galaxies as identified by sdss .
we conclude our study by summarising some of the lessons learnt during this search .
key elements for a successful disk lens search appear to be high resolution images spanning from the blue to the near infrared , to exploit colour contrast as well as low dust extinction .
the main remaining limitation is our ability to disentangle light from the deflector and source .
future studies should increase their success rate by targeting emission lines in the source at high resolution , via narrow band imaging or integral field spectroscopy .
we thank matteo barnab for useful discussions and suggestions .
tt acknowledges support from the nsf through career award nsf-0642621 , and from the packard foundation through a packard research fellowship .
aad acknowledges financial support from a cita national fellowship , from the national science foundation science and technology center cfao , managed by uc santa cruz under cooperative agreement no .
aad and dck were partially supported by nsf grant ast 08 - 08133 , and by hst grants ar-10664.01-a , hst ar-10965.02-a , and hst go-11206.02-a .
pjm was given support by the tabasgo and kavli foundations , and the royal society , in the form of research fellowships .
lvek acknowledges the support by an nwo - vidi programme subsidy ( programme number 639.042.505 ) .
this research is supported by nasa through hubble space telescope programs go-10587 , go-11202 , and go-11978 , and in part by the national science foundation under grant no .
phy99 - 07949 . and
is based on observations made with the nasa / esa hubble space telescope and obtained at the space telescope science institute , which is operated by the association of universities for research in astronomy , inc .
, under nasa contract nas 5 - 26555 , and at the w.m .
keck observatory , which is operated as a scientific partnership among the california institute of technology , the university of california and the national aeronautics and space administration .
the observatory was made possible by the generous financial support of the w.m .
keck foundation .
the authors wish to recognize and acknowledge the very significant cultural role and reverence that the summit of mauna kea has always had within the indigenous hawaiian community .
we are most fortunate to have the opportunity to conduct observations from this mountain .
funding for the sdss and sdss - ii was provided by the alfred p. sloan foundation , the participating institutions , the national science foundation , the u.s .
department of energy , the national aeronautics and space administration , the japanese monbukagakusho , the max planck society , and the higher education funding council for england .
the sdss was managed by the astrophysical research consortium for the participating institutions .
the sdss web site is http://www.sdss.org/.
the parent sample for the new swells lens search consists of all sdss dr7 spectra with either good or marginal plate quality that are classified by the sdss spectroscopic reduction pipeline as galaxies .
noise estimates for all spectra were rescaled on a plate - by - plate basis to be consistent with the statistics of sky - subtracted sky spectra on that plate . the spectroscopic selection algorithm of the original slacs sample and our new swells sample both rely upon subtracting a model for the continuum of the foreground galaxy and scanning the residual spectrum for higher redshift emission features . for slacs
, these models were taken directly from the sdss redshift pipeline , and consist of the best - fit combination of 4 pca - derived eigenspectra .
for the absorption - dominated systems typical of slacs , these models provide essentially noise - limited subtraction of the foreground galaxy continua .
however , for emission - line foreground spectra , the quality of continuum subtraction with these models is much worse .
hence , we derived new continuum models by ( 1 ) generating a new pca basis with 7 eigenspectra , and ( 2 ) masking all common emission features in the foreground galaxy spectra before fitting each spectrum with this basis .
the new models were subtracted from the target spectra , and the residual spectra were scanned automatically for either ( 1 ) multiple emission lines at a single background redshift , or ( 2 ) a single emission line consistent with a marginally resolved [ oii]3727 doublet .
potential [ oii ] identifications were all checked for the presence of veto lines that would indicate h@xmath0 , [ oiii]5007 , or h@xmath34 as a more likely identification .
neighboring fibers in the multifiber spectrograph image were also checked to make sure that candidate [ oii ] detections were not due to crosstalk from bright emission lines in adjacent spectra , or to small - scale auroral emission in the night sky . , e. , dunkley , j. , nolta , m. r. , bennett , c. l. , gold , b. , hinshaw , g. , jarosik , n. , larson , d. , limon , m. , page , l. , spergel , d. n. , halpern , m. , hill , r. s. , kogut , a. , meyer , s. s. , tucker , g. s. , weiland , j. l. , wollack , e. , & wright , e. l. 2009 , apjs , 180 , 330 , d. n. , bean , r. , dor , o. , nolta , m. r. , bennett , c. l. , dunkley , j. , hinshaw , g. , jarosik , n. , komatsu , e. , page , l. , peiris , h. v. , verde , l. , halpern , m. , hill , r. s. , kogut , a. , limon , m. , meyer , s. s. , odegard , n. , tucker , g. s. , weiland , j. l. , wollack , e. , & wright , e. l. 2007 , apjs , 170 , 377 | the relative contribution of baryons and dark matter to the inner regions of spiral galaxies provides critical clues to their formation and evolution , but it is generally difficult to determine . for spiral galaxies that are strong gravitational lenses , however , the combination of lensing and kinematic observations can be used to break the disk - halo degeneracy . in turn ,
such data constrain fundamental parameters such as i ) the mass density profile slope and axis ratio of the dark matter halo , and by comparison with dark matter - only numerical simulations the modifications imposed by baryons ; ii ) the mass in stars and therefore the overall star formation efficiency , and the amount of feedback ; iii ) by comparison with stellar population synthesis models , the normalization of the stellar initial mass function . in this first paper of a series
, we present a sample of 16 secure , 1 probable , and 6 possible strong lensing spiral galaxies , for which multi - band high - resolution images and rotation curves were obtained using the hubble space telescope and keck - ii telescope as part of the sloan wfc edge - on late - type lens survey ( swells ) .
the sample includes 8 newly discovered secure systems .
we characterize the sample of deflector galaxies in terms of their morphologies , structural parameters , and stellar masses .
we find that the swells sample of secure lenses spans a broad range of morphologies ( from lenticular to late - type spiral ) , spectral types ( quantified by h@xmath0 emission ) , and bulge to total stellar mass ratio ( 0.22 - 0.85 ) , while being limited to m@xmath1 m@xmath2 .
the swells sample is thus well - suited for exploring the relationship between dark and luminous matter in a broad range of galaxies .
we find that the deflector galaxies obey the same size - mass relation as that of a comparison sample of elongated non - lens galaxies selected from the sdss survey .
we conclude that the swells sample is consistent with being representative of the overall population of high - mass high - inclination disky galaxies .
[ firstpage ] galaxies : spiral galaxies : structure galaxies : haloes galaxies : fundamental parameters gravitational lensing |
when modeling nonlinear problems , dissipative algorithms have provided researchers with an extremely valuable tool since usually most non dissipative schemes fail to produce a stable evolution .
more precisely , when using neutrally stable algorithms , instabilities often arise which spoil the evolution .
the addition of artificial dissipation removes these instabilities by `` damping '' the growing modes of the solution , in a somewhat controlled way .
therefore , its inclusion in a discretization scheme provides a practical and economic way of achieving longer evolutions .
the most widely used algorithms with this property are the family of lax schemes @xcite ; whereby adding to the equation @xmath0 a term proportional to @xmath1 one obtains a stable discretization of the system that would otherwise be unstable .
however , one might correctly ask whether this is not tantamount to solving a completely different problem .
the beauty of these methods is that the proportionality factor depends on the discretization size , and in the continuous limit the approximation to modified pde results in a consistent approximation to the original one .
although there is much experience with these kind of schemes , most of the standard dissipative algorithms have been tailored for cauchy initial value problems , where initial data is provided at one instant of time and evolved to the next instant by means of the evolution equation . however , in radiative problems , it is sometimes more convenient to choose a sequence of hypersurfaces adapted to the propagation of the waves , and therefore one adopts a foliation adapted to the characteristics of the pde under study . in the present work , we present a new algorithm adequate for hyperbolic systems .
the underlying strategy of the proposed algorithm is quite different from the conventional cauchy - type methods .
rather , it is inspired by analytic concepts developed in the 1960s @xcite for studies of gravitational radiation in general relativity and in their subsequent numerical integrations .
the main features of this approach are the use of characteristic surfaces ( for the foliation of the spacetime ) and compactification methods ( that enable the inclusion of infinity in the numerical grid ) to describe radiation properties .
although evolution algorithms ( for systems possessing some kind of symmetry ) developed within this approach proved to be successful in the linear and mildly nonlinear regime @xcite , they produce unstable evolutions when applied to the general case ; which shows the need for devising algorithms that could be applied in this situation . in the present work we present a new algorithm having dissipative properties , making it a valuable tool to study systems where strong fields might be present . in section @xmath2 the details of the algorithm for the wave equation
are presented and its stability properties discussed .
section 3 is devoted to treat a model problem which shows how the dissipative algorithm might be a useful tool for numerical investigations in general relativity .
finally , in section @xmath3 we describe two particular applications of this algorithm in the numerical implementation of general relativity .
waves of amplitude @xmath4 traveling in one spatial direction with unit speed obey the familiar equation @xmath5 which can be solved in the region @xmath6 , assuming @xmath7 and @xmath8 are given .
if , instead , one is interested in the solving the problem restricted to the region @xmath9 , boundary data must also be provided corresponding to @xmath10 .
the analysis of this problem can be described in a simple way in terms of the characteristics of this equation , which are given by @xmath11 through each spatial point @xmath12 . in particular , when solving eq .
@xmath13 in the region @xmath14 .
the domain of dependence @xmath15 of a point @xmath16 is given by @xmath17 , with @xmath18 naturally defined by the characteristics passing through @xmath16 as @xmath19 and @xmath20 is the region to the future of * the line @xmath21 , + * the region defined by @xmath22 or @xmath23 $ ] ( where @xmath24 ) ; in these cases , boundary conditions must be imposed at @xmath25 ( and @xmath26 in the latter case ) .
suppose one introduces a coordinate system adapted to the characteristics by , say @xmath27 ; then , eq . ( [ weqcar ] ) reduces to @xmath28 further , one can then choose to foliate the spacetime by a sequence of characteristics obtained by holding the ( retarded ) time @xmath29 .
one can then define a _
characteristic initial value problem _
, where eq .
( [ weqnul ] ) is solved provided that @xmath30 is given .
( note that @xmath31 need not be provided as in the cauchy initial value problem ) .
it is straightforward to check that a solution of eq .
( [ weqnul ] ) is expressible as @xmath32 ( where @xmath33 and @xmath34 are smooth functions ) .
physically , @xmath35 describes waves propagating in the @xmath36 direction ( outgoing waves ) and @xmath37 describes waves propagating in the @xmath38 direction ( incoming waves ) .
then , if one imposes the condition of pure outgoing waves , the solution must be of the form @xmath39 ; hence , along each characteristic the value of the function is constant .
notably , boundary data at @xmath40 can be given arbitrarily since purely outgoing waves at @xmath41 will not reach @xmath40 .
more generally , boundary data consistent with the incoming waves must be prescribed at @xmath40 .
it is important to note the domain of dependence for this problem . when solving eq .
( [ weqnul ] ) in the region @xmath42 , the domain of dependence @xmath43 of a point @xmath44 is defined by @xmath45 where @xmath46 however , if the region @xmath42 is chosen to be the future of the line @xmath47 , @xmath48 extends indefinitely to the past .
therefore , the characteristic approach requires @xmath42 to have a boundary .
thus , one defines @xmath42 as the region @xmath49 ( with @xmath50 ) .
figure ( [ fig : domdep ] ) illustrates the domains of dependence corresponding to each formulation . for hyperbolic systems with two or more spatial dimensions ,
the manner in which the characteristics determine the domain of dependence and lead to evolution equations is qualitatively the same .
also , the use of coordinates adapted to them provide a tidy way for studying the system .
for instance , in @xmath51 dimensions , the wave equation is given by @xmath52 which , in term of spherical polar coordinates @xmath53 has the form @xmath54 where @xmath55 denotes the angular momentum operator @xmath56 introducing coordinates @xmath57 , which defines a natural inner boundary at @xmath40 , eq .
( [ wav3d ] ) takes the form @xmath58 thus , by defining @xmath59 and considering @xmath60 as a source term , eq .
( [ 3d1d ] ) formally looks like the @xmath61 dimensional system .
therefore , from now on we restrict our analysis to this latter case and extend our results to the @xmath51 dimensional case in section @xmath3 .
the formal integration of ( [ weqnul ] ) proceeds by an integration in the @xmath62 direction on each @xmath29 surface and then evolve to the next level .
this reformulates the integration in the characteristic formulation as an `` evolution '' in the radial direction and then another in the @xmath63 direction ( as opposed to the evolution of a `` whole '' instant of time to the next one typical of the cauchy evolution ) .
hence , standard dissipative schemes intended for cauchy - type evolutions ( like the family of lax algorithms ) are not directly applicable in the characteristic formulation of the pde and the addition of artificial viscosity to the system must be reformulated . in the numerical implementation of eq .
( [ weqnul ] ) a useful discretization was introduced in @xcite .
this scheme is basically a second order approximation based on finite difference techniques . assuming the grid discretization is given by @xmath64 and @xmath65 , the derivatives may be discretized in the following way : @xmath66 the resulting scheme ( which we will refer to as giw ) is a second order in time , second order in space accurate algorithm .
notably , the von neuman analysis shows that the giw scheme has unitary amplification factor ( i.e. a neutrally stable algorithm ) independent of the values of @xmath67 and @xmath68 .
this would imply that the algorithm is unconditionally stable which is at first sight puzzling .
this might be explained by the implicit local structure of the algorithm ( since it involves @xmath51 points at the upper time level ) and , as such , a local stability analysis need not give a condition on the discretization size .
nevertheless , the algorithm is globally explicit as the evolution proceeds by an outward march from the origin .
hence , the algorithm does require the enforcement of the cfl condition to ensure that the numerical and analytical domains of dependence are consistent
. the cfl condition for the system can be easily obtained .
the field at grid point at @xmath69 , depends critically on the field value at @xmath70 ( since all the points where @xmath71 are trivially included in the discretization ) .
the requirement for the numerical domain of dependence to include the analytical domain of dependence is @xmath72 ; therefore , the cfl condition will be satisfied if @xmath73 .
the giw algorithm has been employed successfully in the characteristic formulation of general relativity ( g.r . ) assuming either spherical symmetry @xcite ; axisymmetry @xcite or very small departures from spherical symmetry @xcite . however , when considering more general problems , as it is often the case with neutrally stable schemes , round off error or parasitic modes are enough to cause ripples in the solution which often lead to an unstable evolution . as stated earlier
, adding dissipation to the pde constitutes a way to alleviate this problem @xcite .
we now show that a rather simple modification of ( [ weqnul ] ) can be used to obtain a consistent discretization with dissipative properties .
we start by considering the following equation @xmath74 ( the @xmath75 factor is included for convenience ) .
a straightforward discretization of ( [ neweqn ] ) is obtained by the described approximation for @xmath76 ( [ gur ] ) and @xmath77 ( [ grr ] ) and by approximating the third derivative at the point @xmath78 as @xmath79 in analogy to the lax method , the inclusion of this extra term leads to a consistent difference approximation of eq .
( [ weqnul ] ) ; that is , the difference approximation converges formally to the differential equation in the limit @xmath80 .
in fact , it is straightforward to check that the resulting approximation is accurate of order @xmath81 .
an important feature of the resulting algorithm ( which we shall call da ) is its dissipative features , which make it particularly useful .
the stability properties of this algorithm can be easily obtained by introducing fourier modes such that @xmath82 . after some algebra one obtains @xmath83 where @xmath84 and @xmath85 .
therefore , the equation governing the growth of the solution s modes is @xmath86 now , since @xmath87 ( for @xmath88 ) the scheme will be stable if , @xmath89 .
moreover , @xmath90 as @xmath91 , indicating that the high frequency modes are effectively `` damped '' , while @xmath92 as @xmath93 .
the obtained discretization can also be thought of as an approximation to the original equation ( [ weqnul ] ) ( i.e. without the addition of the extra term ) where the finite differencing of @xmath76 includes @xmath3 points on the _ nth _ level as @xmath94 which can be regarded as a weighted average of the derivatives at @xmath78 obtained from field values at the points @xmath95 and @xmath96 . in the next section ,
we illustrate how this algorithm produces a stable discretization when the original strategy ( corresponding to @xmath97 ) fails .
in this section we study the stability properties of an equation bearing close resemblance to the nonlinear evolution equation encountered in the characteristic formulation of general relativity ( which will presented in the section iv ) @xmath98 in order to keep track of the nonlinearity of the equation , we introduce the parameter @xmath99 ( with @xmath100 ) , such that @xmath101 ; hence , eq . ( [ eq : biggnon ] ) becomes @xmath102 in particular , note that the principal part of eq .
( [ eq : biggnongral ] ) corresponds to the wave equation .
also , it reduces in the the linear case ( @xmath103 ) , to the wave equation . consequently ; one might expect the giw discretization to lead a stable scheme .
however this is not the case , as can be demonstrated by the following analysis .
first , in order to simplify the study of the stability of this nonlinear problem , we _ linearize the pde with respect to the previous time step_@xcite to obtain a more manageable equation . in this linearization , we approximate the values of @xmath4 and @xmath104 with respect to the _ nth _ level , but @xmath105 is centered in between the levels .
the resulting finite difference approximation is @xmath106 finally , we introduce the fourier modes @xmath107 and solve for @xmath108 , obtaining @xmath109 where @xmath110 , @xmath111 and @xmath112 it is not difficult to check that @xmath113 is a positive quantity for @xmath114 ( which will be the case if the cfl condition is satisfied ) .
thus , the value of @xmath115 is always larger than @xmath61 , indicating that this discretization is _ unconditionally unstable ! _ it is remarkable that the simple addition of some nonlinear terms , even when they do not change the equation s principal part , completely break an algorithm that would otherwise be stable .
we now modify the way @xmath76 is discretized by introducing dissipation as dictated by the d.a .
scheme , i.e. @xmath116 with this simple modification , the value of @xmath108 now becomes @xmath117 since @xmath118 and @xmath119 , the condition for stability is that @xmath120 . given @xmath121 , this a condition on @xmath122 , or vice versa .
for instance , if @xmath123 and @xmath124 , the discretization will be stable if @xmath125 . on the other hand , if we choose @xmath126 and @xmath127 , @xmath128 will be satisfied if @xmath129 .
figure [ vareps ] shows the value of @xmath115 for different choices of @xmath122 for a given @xmath99 ; the effect of the added dissipation can be clearly seen .
when solving einstein equations , one can take advantage of the coordinate invariance of the theory to simplify the modeling of a specific problem .
in particular , one is free to choose a foliation of the spacetime better suited to the problem . in the numerical implementation of g.r . , the most common approach is to choose a foliation by spacelike hypersurfaces at constant times . in this approach ,
einstein equations form a second oder pde system for the _ intrinsic geometry _ of each surface and its embedding in the spacetime , the _ extrinsic curvature_. einstein equations are split into two distinct set of equations .
one set consists of _ constraint _ equations that limit the possible configurations of the field variables on each hypersurface .
the second set constitutes the _ evolution _ equations that determine the development of the field variables onto the next hypersurface .
the main drawback of the numerical implementation of the cauchy formulation is the impossibility of having an infinite grid to completely cover the spacelike hypersurfaces .
thus , in practice , one chooses an exterior boundary in order to deal with a finite domain .
this introduces further problems since special conditions on the boundary must be imposed in order to avoid reflections which spoil long term evolutions .
although in the @xmath61 dimensional case there are sound method to achieve this ( e.g. the sommerfeld condition ) , in the general case , any local boundary condition still introduces reflections turning the task of obtaining long accurate evolutions into an almost impossible one .
a related problem arising from an outer boundary at a finite distance is that the radiation can not be rigorously calculated .
when studying gravitational radiation a more natural choice adapted to the wave propagation is to adopt a sequence of characteristic hypersurfaces to cover the spacetime .
this approach is known as the _ characteristic formulation of g.r .
_ , pioneered by bondi and sachs @xcite .
the main ingredients of this formulation are the foliation of the spacetime by a sequence of characteristic hypersurfaces and the use of compactification techniques ( which enable the inclusion of infinity in a finite grid ) to rigorously describe asymptotic properties of radiation @xcite .
the equations naturally split into _
hypersurface equations _ and _ evolution equations_. we now outline the main aspects of the numerical implementation of this formulation ( based on @xcite ; where a detailed description of the problem has been presented ) and employ the constructed algorithm to discretize the pde equations governing the evolution of the fields .
a coordinate system is introduced by labeling the outgoing lightlike hypersurfaces with a parameter @xmath63 ; ach null ray on a specific hypersurface labeled with @xmath130 @xmath131 and choose @xmath62 as a surface area coordinate ( i.e. surfaces at @xmath132 have area @xmath133 ) . in
the resulting @xmath134 coordinates , the metric takes the bondi - sachs form @xcite @xmath135 where @xmath136 and @xmath137 , with @xmath138 a unit sphere metric .
the metric components are re - expressed as @xmath139 + k),\;\nonumber \\ h_{23}&=&h_{32}=\frac{4}{p^2}\im[j],\;\nonumber \\
h_{33}&=&\frac{4}{p^2}(k-\re[j ] ) , \nonumber \\ u^2&=&\frac{p}{2}\re[u],\ ; \nonumber \\ u^3&=&\frac{p}{2}\im[u ] ; \label{eq : huq}\end{aligned}\ ] ] where @xmath140 in standard angular spherical coordinates @xmath141 . here
, the metric is expressed in terms of two real ( @xmath142 and @xmath143 ) and two complex ( @xmath144 and @xmath145 ) variables ( where @xmath146 ) .
the complex field @xmath145 measures the departure of spherical symmetry of the surfaces given by @xmath132 , and @xmath29 ; @xmath143 represents the mass distribution of the system ; @xmath142 measures the expansion of the light rays and @xmath144 measures the shift in the angular directions from one hypersurface to another ( at constant @xmath62 ) .
the hypersurface equations are expressed as : @xmath147 \label{beta } \\
u_{,r } & = & f_u[\beta , j ] \label{u } \\
( r^2 q)_{,r } & = & f_q[u,\beta , j ] \label{q } \\ v_{,r}&= & f_v[q , u,\beta , j ] , \label{w}\end{aligned}\ ] ] where @xmath148 which is introduced to deal with a first order system of hypersurface equations .
the functions @xmath149 and @xmath150 involve derivatives taken only on a particular hypersurface @xmath151 .
then , they can be easily integrated if @xmath145 is known on @xmath151 ( assuming consistent boundary conditions are provided ) in the following way .
the integration strategy proceeds by first obtaining @xmath142 from eq .
( [ beta ] ) , then @xmath144 from eq .
( [ u ] ) , followed by the calculation of @xmath152 using eq .
( [ q ] ) and finally @xmath143 using eq .
( [ w ] ) .
the evolution to the next hypersurface is prescribed by a first order ( in time ) equation for @xmath145 that takes the form @xmath153
\label{eq : wev}\end{aligned}\ ] ] where @xmath154 involve derivatives on @xmath151 only .
a code that implements einstein equations was written using ( second order ) finite difference approximations .
angular and radial derivatives are approximated along the following lines @xcite , * _ angular derivatives_. we follow the formalism given in @xcite . to expedite the numerical implementation of angular derivatives , instead of working with the standard spherical angular coordinates @xmath155 ,
we work in stereographic coordinates : @xmath156 and angular derivatives are written in terms of the ( complex differential ) eth operators @xmath157 and @xmath158 @xcite ; for instance , @xmath159 this allow us to employ a set of numerical techniques introduced in @xcite which are specially tailored to : ( _ i _ ) handle the numerical approximation of angular derivative operators and ( _ ii _ )
deal with the fact that a single coordinate patch can not be used to smoothly cover the sphere . * _ radial derivatives_. these are approximated via centered second order differences along each null ray ( i.e. holding @xmath160 ) ; for instance @xmath161 the evolution equation deserves special consideration .
its discretization ( in between levels @xmath162 and @xmath163 ) is obtained using dissipation in the following way : schematically , it can be re - expressed as @xmath164 .
\ , , \label{eq : evol}\ ] ] where @xmath165 .
the function @xmath154 can be straightforwardly approximated at each grid point @xmath166 to second order accuracy .
then , in order to introduce dissipation in the algorithm , we proceed to consider a modified version ( along the lines described in section ii ) @xmath167 we center the derivatives at the point @xmath168 , as dictated by the da scheme and obtain @xmath169 by means of an iterative procedure . in the first iteration we set @xmath170 and get a first approximation of @xmath171 via the evolution equation .
then , we use this value to obtain a guess for @xmath169 which is then used to get a better approximation for @xmath171 . this procedure is repeated a sufficient number of times to ensure convergence .
unfortunately , when solving a @xmath51 dimensional problem , the computational requirements of integrating from the origin ( @xmath40 ) are formidable .
however , it is possible to start the integration from a finite value of @xmath62 , assuming consistent values of @xmath142 , @xmath144 , @xmath152 , @xmath143 and @xmath145 are known on this boundary ( which is refered to as the worldtube boundary ) as well as the value of @xmath145 on an initial hypersurface @xcite . to illustrate the usefulness of the presented algorithm , we apply it to model ( _ i _ ) the propagation of linear waves on a minkowski background and ( _ ii _ ) the problem of scattering off a schwarzschild black hole in 3 dimensions . in the past ,
analytical solutions of linearized einstein equations ( in the characteristic formulation ) have been found which describe waves propagating on a flat background @xcite .
these solutions provide an important test bed for the algorithm , since the numerically obtained solutions must converge to the analytic values given by @xmath172 with @xmath145 and @xmath144 obtained from a solution ( @xmath173 ) of the scalar wave equation by @xmath174 in order to test the algorithm , we choose a solution of the wave equation in @xmath51 dimensions , that represents an outgoing wave with angular momentum @xmath175 of the form @xmath176 where @xmath177 is the @xmath178-translation operator . the resulting solution is well behaved above the singular light cone @xmath179 . choosing initial data of very small amplitude @xmath180 ; we used these solutions to give data at @xmath181 ( with the inner boundary set at @xmath182 ) and compared the numerical and exact solutions over time for different values of @xmath122 .
the computation was performed on grids of size @xmath183 equal to @xmath184 , @xmath185 , @xmath186 ( with the number of angular points @xmath187 and the ratio @xmath188 ) .
the @xmath189 norm of the error was calculated over the entire grid and plotted against time for different values of the dissipation parameter .
figure [ errwav ] shows the logarithm of the error in @xmath145 vs. time ( for runs with @xmath190 ) .
for @xmath97 the evolution is unstable , as can be seen by the exponential growth of the error .
for @xmath191 the instability appears at a later time , also with an exponential growth .
however , for @xmath192 the run proceeds stably and the error remains under control .
it is important to note that the magnitude of the dissipation needed to achieve a stable run is very small and therefore the `` damping '' of the solution in not severe .
the characteristic initial value problem on an outgoing null hypersurface requires inner boundary conditions on the worldtube .
here we consider an example in which the inner boundary @xmath193 consists of an ingoing nullcone ( see figure [ fig : pulse ] ) .
we adopt coordinates @xmath130 which follow the ingoing null geodesics and foliate @xmath193 ( chosen to correspond to ingoing @xmath194 surface in a schwarzschild spacetime ) by slices separated by constant parameter @xmath63 . in these coordinates , the schwarzschild line element takes form @xmath195 the initial data correspond to setting @xmath196 as data on @xmath179 , with the boundary conditions @xmath197 and @xmath198 on @xmath193 .
we pose the nonlinear problem of gravitational wave scattering off a schwarzschild black hole by retaining these boundary conditions on @xmath193 but we choose null data at @xmath179 corresponding to an incoming pulse with compact support , @xmath199 $ } \\ & \\ 0 & \mbox{otherwise , } \end{array } \right.\ ] ] where @xmath200 is the spin - two spherical harmonic @xcite .
the code was run for different values of @xmath99 under different choices of the dissipation parameter . in all cases
, unstable evolutions resulted from the choice @xmath97 , however for nonzero values of @xmath122 the code ran without any stability problem as illustrated in figure [ fig : evol1 ] ( for a run where @xmath201 ) . yet , as expected of any dissipative algorithm , the solution decreases in amplitude with time .
this highlights the need to carefully tune the value of @xmath122 .
notwithstanding this fact , it is important to stress once again that this set of runs would not have been possible without dissipation .
this problem was originally studied in the perturbative regime by price @xcite .
there is no known analytic solution to the problem in the nonlinear regime and applying numerical methods is the only way to study it .
the accuracy of the dissipative scheme can be assessed indirectly by inspection of the gravitational waves produced by the system .
gravitational waves can be described in terms of two _ polarization modes _ ( refered to as _ plus _ and _ cross _ modes ) @xcite . however , when considering spacetimes with axisymmetry , the cross mode must vanish and this fact can be used to test the algorithm . calculating
the gravitational radiation is a rather involved problem that exceeds the scope of this work .
a set of algorithms to numerically calculate the gravitational wave forms was constructed in the characteristic formulation in @xcite and tested under different situations .
we used these algorithms in the present work to calculate the polarization modes for the choice @xmath202 and with an axisymmetric pulse with @xmath203 as the initial data .
the cross polarization mode actually converges to zero in second order indicating an accurate discretization of einstein equations , as can be seen in figure [ fig : conver ] .
the algorithm described in this work represents a valuable tool for the study of nonlinear problems in the characteristic formulation .
its use enables long term evolution that would otherwise be impossible . yet
, there is still much room for improvement as the number of numerical techniques adapted to characteristic type evolutions is scarce ( as opposed to the situtation in the cauchy type evolution where one has at hand a great number of algorithms ) .
the variety of physical problems where propagating waves are to be described , highlights the need of further investigations on `` characteristic '' algorithms . of particular interest
is the application of these type of algorithms to the characteristic module constructed to model the collision of a binary black hole self - gravitating system . in this problem
, it is imperative to have robust enough schemes capable of dealing with highly nonlinear fields .
the complexity of the problem inspired the creation of the binary black hole grand challenge alliance , where a group of u.s .
universities and outside collaborators are joining efforts to tackle the problem @xcite . a strategy to study
this problem is a `` hybrid '' scheme that implements at the same time a cauchy evolution ( for the region near the black holes ) and a characteristic evolution ( for the exterior region ) .
this approach is called _ cauchy - characteristic matching ( ccm ) _ @xcite , and in principle , its implementation manages to avoid the problems and to exploit the best features of each evolution scheme .
ccm has been shown to work ( and outperform traditional outer boundary conditions ) in situations where special symmetries were assumed @xcite and its full @xmath51 dimensional application in g.r .
is currently under study .
the characteristic code is one of the pieces of this bigger algorithm and the need for robust performance prompted this investigation .
however , its use is not limited to g.r .
any hyperbolic system describing waves will have an evolution equation similar to eq .
( [ weqnul ] ) .
the algorithm presented in this work should provide a useful tool in the numerical modeling of these systems .
this work has been supported in part by the andrew mellon fellowship , by the nsf grants phy 9510895 and nsf int 9515257 to the university of pittsburgh and by the binary black hole grand challenge alliance , nsf grant phy / asc 9318152 ( arpa supplemented ) . the author would like to thank jeffrey winicour , roberto gmez and nigel bishop for valuable suggestions , also to the universities of south africa and of durban - westville for their hospitality in completing part of this work .
computer time for this project has been provided by the pittsburgh supercomputing center under grant phy860023p . | we present a dissipative algorithm for solving nonlinear wave - like equations when the initial data is specified on characteristic surfaces . the dissipative properties built in this algorithm make it particularly useful when studying the highly nonlinear regime where previous methods have failed to give a stable evolution in three dimensions .
the algorithm presented in this work is directly applicable to hyperbolic systems proper of electromagnetism , yang - mills and general relativity theories .
we carry out an analysis of the stability of the algorithm and test its properties with linear waves propagating on a minkowski background and the scattering off a scwharszchild black hole in general relativity .
key words : derivation of finite difference approximations ; stability and convergence of difference methods ; electromagnetism , other .
+ subject classifications : 65p05 , 65p10 , 77c10 , 77a99 . proposed running head : dissipative algorithm for wave - like equations .
+ * name : * luis lehner * address : * 2truemm 2truemm * tel . : * 1-(512 ) 471 - 5426 * e - mail address : * [email protected] * fax : * 1-(512 ) 471 - 0890 |
the dynamics of first order phase transitions is a fundamental ingredient in particle physics and in condensed matter .
first order phase transitions occur via the nucleation of bubbles of the true vacuum phase in the metastable or false vacuum phase . at large temperature
it is mediated by thermal or overbarrier activation and at low temperatures via quantum nucleation .
first order phase transitions are conjectured to occur in qcd and in the electroweak theory . in qcd a first order phase transition _
could _ describe the hadronization of the quark - gluon plasma , possibly produced in the early universe at about @xmath8 seconds after the big bang or in relativistic heavy ion collisions@xcite . in the electroweak theory
a first order phase transition is argued to provide the non - equilibrium setting for baryogenesis@xcite . in early universe cosmology ,
first order phase transitions had been proposed as mechanisms to generate the inflationary stage@xcite . in condensed matter physics
thermal activation results in the nucleation of bubbles of the lowest free energy phase in binary fluids@xcite and also of the decay of metastable dimerized states in quasi - one dimensional charge density wave systems and non - degenerate organic conductors@xcite .
the most comprehensive microscopic theory of nucleation via thermal activation was presented by langer@xcite and was later extended to quantum field theory to account both for thermal activation as well as for quantum nucleation@xcite . an approach to describe nucleation in a non - steady state situation in real time has been advocated in@xcite . in the limit in which nucleation is dominated by thermal activation and overbarrier transitions ,
these bubbles are produced via large thermal fluctuations .
these bubbles will grow whenever their radius is larger than a critical value and collapse if it is smaller .
supercritical bubbles will grow to convert the metastable phase into the stable phase or until they percolate achieving the total conversion of the metastable phase . for
slightly supercritical bubbles an important _ dynamical _ quantity is the growth rate of a bubble @xmath9 ; ~~ r(t ) \geq r_c\ ] ] with @xmath10 the ( time dependent ) radius of a slightly supercritical nucleating bubble and @xmath11 is the critical radius@xcite .
theory provides the nucleation rate per unit volume per unit time given by @xmath12 where @xmath13 is the free energy of a critical bubble and @xmath14 is proportional to the ratio of the determinants of the fluctuation operators around the bubble configuration to that around the homogeneous metastable state@xcite .
the regime of applicability of homogeneous nucleation theory is for @xmath15 , for @xmath16 small amplitude thermal fluctuations can trigger the phase transition and nucleation and spinodal decomposition can no longer be distinguished .
the ratio of determinants in @xmath14 has been computed analytically and numerically in several important cases@xcite .
csernai and kapusta@xcite have studied the growth rate of hadronic bubbles in a quark gas by extending langer s theory to the relativistic case .
these authors studied a coarse - grained field theory in terms of a relativistic hydrodynamic description with viscous terms in the energy momentum tensor and the baryon current .
their conclusion was that the growth rate is determined by the coefficients of the shear and bulk viscosities and that hadronic bubbles do not grow when these coefficients vanish .
another approach presented by ruggeri and friedman@xcite also based on baryon - free relativistic hydrodynamics but with a different treatment of the heat conduction and energy flow reached a different conclusion , th at the growth rate is non - zero even for vanishing bulk and shear viscosities and that viscosity effects are subleading for small viscosities .
if the hadronization phase transition is of first order there are potentially important experimental signatures associated with the homogeneous nucleation of the hadronic phase some of which had been studied in reference@xcite .
homogeneous nucleation of the quark - gluon plasma has been recently studied with a bag model of the equation of state for the quark and hadron phases and the different predictions above have been compared@xcite .
the formulation and results of csernai and kapusta had recently been used to study first order quark - hadron phase transition in the early universe for a first order hadronization transition@xcite and more recently homogeneous nucleation has been tested numerically in @xmath17 dimensional systems with qualitative agreement to the standard result@xcite . an alternative phenomenological description of the growth rate based on dissipative hydrodynamics combined with the finite temperature effective potential has been presented in @xcite . in this
work analytical and numerical studies reveal a dependence of the growth rate on the phenomenological dissipative coefficient .
khlebnikov@xcite and arnold@xcite have studied dissipative effects that slow down the growth rate of a supercritical bubble by coupling the order parameter that describes nucleation to _ other fields _ ( with a trilinear coupling ) and applie d the fluctuation dissipation relation to one loop order in the coupling to the other field .
finally we must mention a numerical approach to the description of nucleation , based on a phenomenological _ markovian _ langevin dynamics in terms of the finite temperature effective potential ( typically computed to one loop order ) with a ( local ) friction coefficient and a gaussian white noise term related by the fluctuation - dissipation theorem@xcite .
the use of the finite temperature effective potential in the description of the inhomogeneous bubble configuration that seeds the nucleation of the phase of lowest free energy is an important ingredient , in particular in coarse - grained phenomenological descriptions@xcite .
finite temperature effects are included in the coarse grained free energy which describes long wavelength physics by integrating o ut short wavelength fluctuations which are in thermodynamic equilibrium@xcite .
these finite temperature corrections modify quantitatively and sometimes substantially the equilibrium free energy functional , for example the position of the minima , masses and couplings .
these are important parameters that determine the profile of the bubble configuration , since the asymptotic behavior of the bubble is determined by the position of the minima of the equilibrium free energy .
the importance of the growth rate for describing the dynamics of nucleation in the hadronization and the electroweak phase transitions as well as the practical importance of nucleation in quasi - one dimensional organic conductors justifies a continued effort to understand from a _ microscopic _ perspective the influence of dissipation , viscosity and friction upon the growth rate of supercritical nucleating bubbles .
* focus and strategy : * our goal in this article is precisely to study dissipative effects upon the growth rate @xmath18 from a more _ microscopic _ point of view in model quantum field theories that describe the main features of nucleation .
the simplest model to study nucleation and a first order phase transition is a @xmath2 field theory with a small explicit symmetry breaking term that breaks the degeneracy between ground states and thus leads to the existence of a metastable state .
although arguably this model could hardly describe the features of the quark - hadron or electroweak phase transitions , our hope is to extract robust phenomena that will be generic to the physics of nucleation and that could enlighten the effect of viscosity on the growth rate .
the study begins by identifying the inhomogenous bubble configuration which is a solution of the static equations of motion .
we include the finite temperature effects by considering the equation of motion in terms of the _ finite temperature effective potential _ ,
this is achieved consistently by the addition of finite temperature counterterms to the lagrangian .
the quadratic fluctuations around this bubble solution feature an unstable mode that describes small perturbations from the critical bubble .
the instability is a manifestation of the growth ( or collapse ) of supercritical ( or subcritical ) bubbles and the unstable eigenvalue is directly related to the growth rate of slightly supercritical bubbles .
besides this unstable mode there are modes of zero frequency corresponding to translations of this configuration and modes of positive frequency that correspond to stable fluctuations .
the main strategy is to consider the dynamics of the coordinate that determines the unstable direction in functional space , and treating it as a _
collective coordinate_. we then focus on the _ interaction _ of this coordinate with those representing the stable fluctuations by expanding the original lagrangian in terms of these coordinates in function space and recognize the interaction terms between these coordinates .
the translational mode is anchored and ignored since we are not interested in studying the dynamics of the translational degree of freedom but that of the growth of a supercritical bubble .
we assume that the stable degrees of freedom are in thermal equilibrium and that they serve as a `` bath '' with which the unstable coordinate interacts . integrating out the degrees of freedom associated with the stable fluctuations we obtain an effective description of the dynamics for the unstable coordinate which includes the `` viscosity and friction '' effects associated with the transfer of energy to the stable modes .
this description allows us to obtain the equations of motion for the expectation value of the unstable coordinate wherefrom we can obtain the _ corrections _ to the growth rate from the interaction with the stable degrees of freedom .
furthermore , integrating out the stable degrees of freedom in the non - equilibrium functional allows us to extract the effective langevin equation for the unstable coordinate and to obtain consistently the noise terms .
we emphasize that this approach is _ fundamentally different _ from the previous studies described above .
first of all , we are _ not _ computing the ratio of determinants that enter in the rate expression .
this ratio of determinants only involves the quadratic fluctuations but _ not _ the interaction between the coordinates associated with the fluctuations .
our approach is different from that of references@xcite , in these references the coupling of the order parameter associated with the nucleating field was to _ other independent fields_. contrary to this we consider the coupling of the unstable coordinates to the stable fluctuations involving the _ same _ field .
our approach starts from the microscopic theory and does not rely on a hydrodynamic description , however as it will be seen later , long - wavelength hydrodynamic fluctuations associated with surface waves provide the largest contribution to viscosity effects in three spatial dimensions . in a very specific sense our program obtains the effective action for the degree of freedom that represents departures from a critical bubble by integrating out the degrees of freedom describing stable fluctuations , but of the _ same field_. our approach must necessarily rely on several different approximations : i ) weak coupling to allow a perturbative expansion of the effective action and the real time equations of motion , ii ) the thin wall approximation in which the critical radius is much larger than the width of the bubble wall .
this approximation is necessary to be able to provide quantitative answers .
finally as it will be justified later the important fluctuations can be described by the _
classical _ limit of the finite temperature distribution functions .
we carry out this program to lowest order in the quartic coupling and to leading order in the thin wall approximation both in @xmath0 and @xmath1 dimensions . the motivation to study the lower dimensional case
is provided by its potential application in experimentally relevant condensed matter systems@xcite .
* summary of results : * our main results can be summarized as a consistent perturbative correction to the growth rate from `` dynamical viscosity '' effects associated with the interaction of the radius of the bubble with the stable quantum and thermal fluctuations . in @xmath0 dimensions the growth rate for bubbles is given by eqn .
( [ estimate ] ) with @xmath19 the radius and width of the critical bubble respectively , @xmath4 the quartic self - coupling and @xmath20 $ ] a slowly varying dimensionless function that depends on the details of the potential and spectrum of stable fluctuations , in particular scattering states . in @xmath1 dimensions we find that in the thin wall limit the important low energy excitations are _ hydrodynamic fluctuations _ associated with surface waves .
these are identified as _ quasi - goldstone _ bosons and give the dominant contribution to viscosity effects on the growth rate .
these low energy excitations are a robust feature of _ any _ model and not particular to the one considered in this article , however , the _ coupling _ of the unstable radial coordinate to these fluctuations depends on the model .
to lowest order in the coupling the growth rate is given by eqn .
( [ finalgrowthrate ] ) .
furthermore , we obtain the effective langevin equation for the unstable coordinate that reveals the correct microscopic correlations of the noise term showing that the coupling to the hydrodynamic modes determine that the noise is correlated on time scales @xmath7 .
the article is organized as follows : section ii introduces the model , describes the strategy in detail and sets up the general form of the correction to the growth rate to lowest order ( one loop ) and in the classical limit . in section
iii the case of @xmath0 dimensions is analyzed in detail within the model considered . in section
iv we study the @xmath1 dimensional case , and discuss the low energy fluctuations associated with surface waves , providing the argument that these are quasi - goldstone hydrodynamic fluctuations .
we argue that these low energy fluctuations are present independent of the model and will dominate the viscosity corrections to the growth rate .
the conclusions are presented in section v , along with a critical discussion on the validity of our results to describe the growth rate of supercritical bubbles in coarse - grained descriptions of a first order quark - hadron phase transition .
an appendix presents the effective langevin equation for small departures of the critical radius and analyzes the generalized fluctuation - dissipation relation .
we consider a real scalar field , @xmath21 , whose dynamics is determined by the following lagrangian density @xmath22 where @xmath23 is a double well potential with a metastable minimum at @xmath24 and a stable minimum at @xmath25 .
it proves convenient to parametrize it in the following form : @xmath26 this form of the potential is depicted in fig.([potfig ] ) and is very similar to the coarse - grained free energy proposed in@xcite identifying the local energy variable _ e _ in that reference with the scalar field @xmath27 , and also similar to the effective potential description used in numerical simulations for the electroweak theory@xcite .
the stable minimum @xmath25 is given by @xmath28 where the point @xmath29 is parametrized by the mass @xmath30 of small oscillations around the _ metastable _ minimum @xmath24 as @xmath31 where @xmath32 and @xmath4 is the quartic coupling .
although this parametrization does not look familiar it will prove advantageous later .
in particular , @xmath33 gives a measure of the free energy difference between the true and the false vacua and is therefore related to the supercooling temperature .
as @xmath34 , the two minima become degenerate and the condition @xmath35 defines the thin - wall limit ( this will become clear later when we analyze the bubble profile ) , when the radius of the nucleating bubble is much larger than the width of its surface @xmath36 ( or the correlation length ) .
this corresponds to the case of small supercooling @xmath3 but @xmath37 in the description of reference@xcite . in the thin wall limit
, we find that @xmath25 has the following simple form @xmath38 we anticipate that there will be renormalization of the parameters in the potential , not only arising from ultraviolet divergences but more importantly from medium effects , i.e. finite temperature contributions to the mass , couplings and explicit symmetry breaking terms .
in particular in three spatial dimensions we expect a correction to the mass and the linear symmetry breaking term both proportional to @xmath39 .
the origin of these finite temperature corrections are the usual tadpole diagrams , the correction to the linear symmetry breaking term is a consequence of the trilinear coupling .
these finite temperature corrections are the usual ones obtained in an effective potential description and provide a finite renormalization to the potential .
our aim is to describe the dynamical aspects of the thermal fluctuations that are responsible for thermal activation and the nucleation of the true phase in the false vacuum phase .
these are the solutions of the static field equations but _ in presence of the thermal fluctuations _ ,
i.e. with the potential renormalized by the finite temperature effects .
that is to say , the potential that must enter in the field equations must be the finite temperature effective potential@xcite .
the finite temperature effective potential includes the finite temperature corrections for static and homogeneous field configurations . including these finite temperature effects in the equations
o f motion guarantees that the _ inhomogeneous _ bubble configurations will asymptotically tend to the homogeneous values of the field representing the _ correct _ extrema of the equilibrium free energy .
furthermore , these finite temperature renormalizations will determine the _ correct
_ scales for the bubble size , its width and the unstable frequency associated with the growth of a supercritical bubble .
we account for these finite temperature renormalizations by replacing the parameters in the potential by the finite temperature renormalized parameters and adding counterterms to the lagrangian .
the counterterms are then found consistently in a coupling or loop expansion by requesting that they cancel the contributions from the tadpole ( or similar ) terms _ completely _ , i.e. including the finite temperature parts .
this prescription is at the heart of using the finite temperature effective potential to study the solutions that lead to the decay of the metastable state@xcite .
thus the lagrangian density becomes @xmath40 = { 1\over 2}(\partial_\mu\phi)(\partial^\mu\phi)- v(\phi , t)+\delta { \cal l}_{ct}[\phi ] \label{counter}\ ] ] where the potential @xmath41 is of the same form as in ( [ pot ] ) but in terms of the parameters that include the finite temperature ( and ultraviolet ) renormalizations , i.e. @xmath42 etc .
and the counterterm lagrangian density is of the form @xmath43 where space - time translational invariance dictates that the counterterms are constant .
these counterterms will be consistently chosen in the loop expansion to cancel the contribution from local tadpoles .
we will argue below that this procedure renormalizes consistently in perturbation theory the _
static _ properties of the bubble , in particular the width and radius of a critical bubble .
in particular the free energy density studied in@xcite is precisely of this form with parameters that depend on temperature . at this point
we may consider adding to the counterterms a wave - function renormalization which would involve corrections for inhomogeneous configurations .
a wave - function renormalization will arise at one loop order and beyond because of the trilinear coupling .
however , computing this correction in the usual loop expansion around a homogeneous background is not very relevant for the bubble configuration . since our goal is to obtain the _ real time
_ equations of motion for particular fluctuations around the bubble configuration , these renormalizations will arise automatically in the equations of motion obtained to one loop order .
thus the counterterms that we include in the lagrangian density are only those for _ static _
renormalization thus accounting for the correct scales and asymptotic values of the classical solution and obtain the equivalent of the wave - function renormalization directly from the real time evolution of fluctuations around the bubble configuration .
this procedure will become clear when we obtain explicitly the evolution equations in the next sections .
we will focus our attention on the cases of one and three spatial dimensions .
although the case of one spatial dimension is of limited interest in particle physics , it is important in condensed matter and statistical physics of quasi - one dimensional systems@xcite .
furthermore most aspects of the treatment are general and the one - dimensional case provides a somewhat simpler setting to introduce the main strategy as well as to explore the general features . the three dimensional case will be studied in detail subsequently . before proceeding to the specific situations , we now consider general features of the bubble solutions and the quadratic fluctuations around them to focus on the precise strategy to follow .
a critical bubble is a solution of the _ static _ field equations , which in terms of the ( effective ) potential @xmath41 is given in one spatial dimension by @xmath44 with boundary conditions such that @xmath45 for @xmath46 .
such a solution describes a `` bubble '' like configuration which approaches the false vacuum at asymptotically large distances and probes the true vacuum in a localized region in space of size @xmath47 with @xmath11 the critical `` radius '' .
such a solution will be found in detail in section iii . in three spatial dimensions ,
the critical bubble is a radially symmetric static field configuration solution of the following equation @xmath48 with the boundary condition @xmath49 as @xmath50 .
it corresponds to a field configuration that starts close to the true vacuum , @xmath25 , and tends to the false vacuum , @xmath24 , at asymptotically large radial distance .
the solution of ( [ 3dbubble ] ) will be studied in detail in the thin wall approximation in section iv . in both cases
the change from the true vacuum to the false one occurs around the radius of the bubble , @xmath11 , over a distance @xmath51 that defines the wall thickness of the bubble and is related to the correlation length in the metastable phase .
we will study the non - equilibrium dynamical viscosity of nucleating bubbles in the thin wall limit in which the radius of the bubble is much larger than the wall thickness ; i.e. @xmath52 .
these field configurations are parametrized by two important coordinates : @xmath53 which describes the position of the center of the bubble and the radius @xmath54 , i.e. a bubble configuration is of the form @xmath55 and a critical bubble corresponds to @xmath56 determined by the solution to the equations of motion ( [ 1dbubble],[3dbubble ] ) .
the coordinate @xmath57 is associated with translational invariance and typically treated as a collective coordinate@xcite , while the radius @xmath54 determines the size of the bubble and will be treated also as a collective coordinate ( see below ) . since we are not concerned here with the dynamics of the translational degrees of freedom , but rather with that of the radius , we `` clamp '' the collective coordinate @xmath53 by fixing the center of the bubble at @xmath58 .
integration over this collective coordinate results in the typical volume factor@xcite and is not relevant for our discussion . in @xmath59 space dimensions
, we can use the radius of the bubble @xmath54 as a variational parameter and introduce the total energy of a bubble configuration with radius @xmath54 by @xmath60 .
\label{bubenergy}\ ] ] and in the limit @xmath61 ( thin wall ) has two main contributions : a volume contribution proportional to @xmath62 which is negative and arises from the region of the bubble that probes the true vacuum which has lower ( free ) energy , and a surface term which is positive and arises from the region of the bubble corresponding to the wall , includes gradient terms and is therefore proportional to the area of the surface of the wall , @xmath63 .
in one spatial dimension the `` surface '' of the bubble corresponds to two points and the gradient energy saturates to a constant independent of the radius for large bubble radius .
the general behavior of the total energy of a bubble configuration as a function of @xmath64 is depicted in fig .
( [ e1dfig ] ) for one spatial dimension and in fig .
( [ e3dfig ] ) for three spatial dimensions ( see sections iii and iv ) .
the maximum of the energy function determines the value of the critical radius @xmath11 as discussed in detail in the next sections .
clearly a critical bubble is an _ unstable _ static solution of the equations of motion . for @xmath65
the surface energy term dominates and the bubble shrinks into the false vacuum phase , for @xmath66 the volume energy dominates and the bubble will grow as the gain in the volume energy is greater than the cost in elastic surface energy . a more clear description is obtained in the case of three spatial dimensions for a spherically symmetric bubble . let @xmath67 be the surface tension ( energy per unit area ) of the spherical bubble configuration separating the metastable phase from the stable one
. then the total energy of the bubble is @xmath68 where @xmath69 , and the dots stand for corrections that are subleading in the thin wall approximation , see section iv ( eqn .
( [ energy3dbubb ] ) for details . these ( small ) corrections will be neglected for the arguments presented below .
the energy attains its maximum @xmath70 at the critical radius @xmath71 and it is given by @xmath72 near the maximum of the energy function , it can be written as an expansion in terms of the ( small ) departures from the critical radius as @xmath73 where the frequency @xmath74 is independent of the critical radius and @xmath75 and as it will be seen in detail below , it is related to the growth rate of slightly supercritical bubbles
. the study of the fluctuations around the classical bubble configuration begins by studying the spectrum of the fluctuation operator @xmath76 { \cal u}_n(x ) = \omega^2_n { \cal u}_n(x ) ~~\mbox{in 1d } \label{1dfluc}\\ & & \left[-\vec{\nabla}^2+v''(\phi_b(\vec x , r_c))\right ] { \cal u}_n(\vec x ) = \omega^2_n { \cal u}_n(\vec x ) ~~\mbox{in 3d } \label{3dfluc}\end{aligned}\ ] ] where the prime represents differentiation with respect to the field @xmath27 . taking a spatial derivative of the equation of motion satisfied by the bubble configuration it is straightforward to find that @xmath77 is a solution of the one dimensional eigenvalue problem ( [ 1dfluc ] ) with @xmath78 and that @xmath79 are eigenfunctions of the three dimensional fluctuation operator with zero eigenvalue .
these are the zero modes arising from translational invariance@xcite . in one dimension
the zero mode is an odd function of @xmath80 since the bubble solution is even , therefore it has a node .
hence there must be another solution of the schroedinger like operator ( [ 1dfluc ] ) that has no nodes and has a smaller and therefore _ negative _ eigenvalue .
the form of the bubble solution in the thin wall approximation is that of a widely separated kink - antikink pair , as it will be shown in detail in the next section we find that the eigenfunction corresponding to the negative eigenvalue must be @xmath81 .
this can be understood simply from the fact that the zero mode associated with the bubble solution is the _ antisymmetric _ combination of the zero modes associated with the individual kinks@xcite , hence the eigenfunction with lower eigenvalue must be the _ symmetric _ combination and the form of the bubble profile immediately leads to the conclusion that is the derivative with respect to the radius ( see section iii , eqn .
( [ 1dunstmode ] ) ) .
in the three dimensional case , the zero modes ( [ 3dzeromode ] ) correspond to the angular momentum @xmath82 spherical harmonics , therefore there must be an @xmath83 ( spherically symmetric ) solution of the schroedinger - like eigenvalue problem ( [ 3dfluc ] ) with a _ negative eigenvalue_. in a later section we study in detail the bubble solution and the spectrum of fluctuations around it and conclude that the eigenfunction corresponding to the negative eigenvalue is @xmath84 in the thin wall approximation ( see section iv , eqn .
( [ spcquasi ] ) for @xmath85 ) . therefore the respective spectra of the fluctuation operators are @xmath86 with @xmath87 normalization factors ( chosen real ) .
the classical bubble solution corresponds to a saddle point in functional space , the mode @xmath88 determines an unstable direction .
the field operator can now be expanded in the complete basis of eigenfunctions of ( [ 1dfluc],[3dfluc ] ) in either case and write in general @xmath89 whereas the bound state eigenfunctions are chosen real , the scattering states are chosen to satisfy the hermiticity condition @xmath90 .
the quanta associated with the coordinates @xmath91 for the stable modes will be referred generically as `` mesons '' , and the operators @xmath91 create and annihilate meson states . from ( [ spectrum ] )
it is clear that the modes @xmath92 correspond to shifts in the position of the bubble , @xmath53 and the radius @xmath54 respectively .
the collective coordinate treatment@xcite absorbs the zero mode @xmath93 into a definition of a new quantum mechanical degree of freedom @xmath94 and expands the field operator in terms of the directions perpendicular to this zero mode , @xmath95 we are not interested in the dynamics of the translational collective coordinate , therefore we will `` clamp '' the position of the bubble and focus solely on the dynamics of the unstable coordinate @xmath96 .
technically this is achieved by inserting a functional delta function for @xmath94 in the path integral , the corresponding jacobian from the change of field variables to the collective coordinate@xcite leads to the usual volume factor@xcite , hence in what follows we will set @xmath97 .
we note that for small amplitudes of the unstable mode we can write the expansion ( [ expre ] ) as @xmath98 that allows us to identify @xmath99 and the coordinate @xmath100 is associated with small departures from the critical radius . at this stage
we recognize that the fluctuation mode @xmath103 is the ( unstable ) functional direction along which the bubble either grows or collapses , therefore the coordinate @xmath104 describes the small fluctuations around the critical bubble .
an immediate question is : what is the relation between the frequency @xmath102 in ( [ energyofr],[wsecdv ] ) and @xmath101 , the eigenvalue associated with @xmath88 ( [ spectrum ] ) ?
the answer to this important question is obtained by taking the second derivative of the total energy in eq .
( [ bubenergy ] ) with respect to the radius of the bubble @xmath54 , which results in the following equation @xmath105 ^ 2 + \vec{\nabla}\phi_b.\vec{\nabla}\left(\frac{d^2\phi_b}{dr^2}\right ) + \left(\frac{d^2\phi_b}{dr^2}\right ) \frac{\partial v(\phi_b)}{\partial \phi } + \left(\frac{d\phi_b}{dr}\right)^2 \frac{\partial^2 v(\phi_b)}{\partial \phi^2}\right\ } \label{d2er}\ ] ] integrating the first two terms in eq .
( [ d2er ] ) by parts using the boundary conditions for the bubble solution and eqs .
( [ 1dbubble],[3dbubble],[1dfluc],[3dfluc ] ) , which are evaluated at the critical radius @xmath11 , we obtain the following relation @xmath106 in terms of the normalization factor of the unstable mode ( see ( [ spectrum ] ) ) which will be found for the one and three dimensional cases separately below .
we note that the relation ( [ expre2 ] ) determines that in the absence of interactions with other degrees of freedom the growth rate of a supercritical bubble is given by @xmath107= \omega \label{growthrate}\ ] ] for small departures from the critical radius .
equation ( [ omegarelation ] ) is very useful since it relates the second derivative of the total _ classical _ variational energy @xmath108 ( the energy of the classical bubble as a function of its radius ) to the eigenfrequency associated with the _ quantum _ unstable mode .
usually it is a difficult task to find the spectrum of the quantum fluctuation operator , and consequently the frequency of the unstable mode .
the above relation can be used to find the value of the unstable frequency , or the normalization factor @xmath109 if the frequencies are known .
our goal is to study the effects of quantum and thermal fluctuations upon the _ real time _
evolution of a bubble whose radius is slightly larger than critical .
this is achieved by treating the coordinate @xmath110 as a _ collective coordinate _ ,
i.e. the radius of the bubble is treated as a fully quantum mechanical variable interacting with the other degrees of freedom corresponding to the stable fluctuations and described by the coordinates @xmath111
. the `` zero mode '' ( translational degree of freedom ) is clamped and frozen since we are only interested in describing the dynamics of the unstable mode .
the interaction of the stable degrees of freedom with the collective coordinate describing the departure from the critical radius will introduce viscosity effects and slow down the growth of a supercritical bubble .
this will result in a smaller growth rate and our aim is to precisely compute this viscosity effect on the growth rate for small departures from the critical radius . by including the finite temperature _
static _ counterterms in the lagrangian and requesting that these be cancelled consistently in the perturbative expansion , we can isolate the static renormalization to the unstable frequency from the _ dynamical _ viscosity effects associated with the energy transfer to the stable modes .
this program begins by expanding the field in terms of the unstable and stable coordinates ( after clamping the translational mode ) as @xmath112 where the summation index @xmath113 runs over all stable , bound and scattering , states other than the translational and unstable modes . throughout the index
@xmath113 refers to both scattering and bound states . using the above expansion , eq .
( [ expansion ] ) , of the field @xmath114 , the lagrangian can be shown to have the following form : @xmath115\,=\,l_0[q_{-1 } ] + l_0[q_p ] + l_i[q_{-1},q_p]\ ] ] where @xmath116&=&\frac{1}{2}~\left\ { \dot q_{-1}^2(t)+ \omega^2 q_{-1}^2(t)+ \delta \omega^2 q_{-1}^2(t ) + hq_{-1}(t)\right\ } \label{unlag } \\ l_0[q_{p } ] & = & \frac{1}{2}~\sum_{p}~ \left\ { \dot{q}_p(t)\dot{q}_{-p}(t)-\omega_p^2 q_p(t)q_{-p}(t)\right\ } \label{scatlag } \\ l_i[q_{-1},q_p ] & = & - \sqrt{\lambda}~q_{-1}(t ) \sum_{p , p'}{\cal b}_{pp'}q_p(t)q_{p'}(t ) \ , - \ , q_{-1}^2(t ) \left\ { \sqrt{\lambda } \sum_{p}{\cal b}_pq_p(t ) + \lambda\sum_{p , p'}{\cal a}_{pp ' } q_p(t)q_{p'}(t)\right\ } \nonumber \\ & & -\ ; q_{-1}^3(t ) \,\left\{\sqrt{\lambda}\,{\cal b}_{-1 } \,+\,\lambda\ , \sum_{p } { \cal a}_pq_p(t ) \right\ } -\,\lambda \ , { \cal a}_{-1}\,q_{-1}^4(t)+\mbox{h.o.t}+ l_{ct}[q_{-1},q_p ] , \label{lagrainte}\end{aligned}\ ] ] where we have used the equations of motion satisfied by the bubble configurations , and equations ( [ 1dfluc ] ) and ( [ 3dfluc ] ) .
the terms @xmath117 are the quadratic and linear terms in @xmath100 from the counterterm lagrangian .
the terms denoted by h.o.t . in ( [ lagrainte ] ) correspond to higher order interactions that are not important to the order that we are studying .
the term @xmath118 $ ] arise from the non - linear ( cubic and higher ) terms in the coordinates from the counterterm lagrangian , they do not need to be specified since they will be requested to cancel tadpole and static terms in the equations of motion for the collective coordinate @xmath100 and do not contribute to lowest order ( @xmath119 ) .
the matrix elements are given by @xmath120 so that the @xmath4-dependence is explicit in each term of the effective action .
we thus see that the above lagrangian describes a quantum mechanical degree of freedom @xmath100 interacting with a bath of infinitely many degrees of freedom @xmath121 , as well as with self - interaction .
the self interaction of the coordinate @xmath100 will be neglected because we are only interested in the viscosity effects arising from interaction with the stable degrees of freedom .
the non - linear terms in @xmath100 are neglegted because we will extract the growth rate for small departures from the critical bubble .
the bath of mesons will introduce viscosity effects on evolution of the coordinate associated with departures from the critical radius and will result in a correction to the growth rate .
these effects can be studied in a consistent manner by integrating out the meson degrees of freedom in a consistent perturbative expansion , thus obtaining the non - equilibrium effective action for @xmath100 .
the variational derivative of this effective action leads to the equation of motion that includes the effects of the meson bath .
this non - equilibrium effective action is obtained up to one loop order in appendix b. the effective equation of motion is a langevin equation with a non - markovian viscosity kernel and a gaussian noise term ( to one loop order ) whose correlations are related to the viscosity kernel via the fluctuation - dissipation relation . the details are provided in appendix b. alternatively we obtain the equation of motion for the expectation value of the coordinate @xmath100 directly by means of linear response .
we have already accounted for the _ static _ renormalization of the unstable frequency @xmath18 by introducing the finite temperature counterterms and using the finite temperature effective potential in the equations of motion .
however , we anticipate that there will arise further corrections to the growth rate @xmath18 from _ velocity dependent _ terms in the effective equations of motion
. these are truly _ dynamical _ viscosity effects which can not be captured by a static calculation .
the viscosity terms that are a function of the time derivative of the unstable coordinate will be revealed by obtaining the effective non - equilibrium equation of motion for this coordinate by integrating out the stable modes , which act as a bath for the quantum mechanical degree of freedom @xmath100 .
we will restrict our study to small departures from the critical radius and obtain the linearized equation of motion for the expectation value of @xmath100 .
this is consistent with the definition of the growth rate as the logarithmic derivative of the radius for small departures from the critical value .
as is clear from ( [ unlag ] ) , in the absence of interactions the hamiltonian for the unstable coordinate @xmath100 is that of an _ inverted harmonic oscillator _ ( of negative frequency squared ) and small departures from the unstable equilibrium value @xmath122 will grow exponentially with the growth rate @xmath18 .
we study the time evolution of the expectation value of the unstable coordinate @xmath123 by proposing an initial state described by a density matrix which is a tensor product of the density matrix for the unstable coordinate and a density matrix that describes all the stable modes in thermal equilibrium at a given temperature .
therefore @xmath124 where @xmath125 describes a pure state in which the expectation value of the unstable coordinate is @xmath126 , and @xmath127 is a thermal density matrix for the stable modes . as mentioned above the translational mode
is clamped .
the time evolution of the initially prepared density matrix is given by liouville s equation , whose solution is @xmath128 with @xmath129 being the unitary time evolution operator . as described in detail in reference@xcite the time evolution can be cast in terms of a time dependent hamiltonian in which the interaction is switched on at the initial time @xmath130 .
alternatively the initial value problem can be cast in terms of linear response to an external source adiabatically turned - on from @xmath131 that determines the initial preparation@xcite .
both approaches are equivalent and the reader is referred to@xcite for more details on the initial value problem .
the equation of motion in real time is obtained by using the generating functional of non - equilibrium green s functions which requires a path integral along a contour in complex time and the following effective lagrangian@xcite : @xmath132-l[q^-_{-1},q^-_p]\ ] ] where the labels @xmath133 refer to the forward @xmath134 and backward @xmath135 branches along the complex time contour@xcite .
the equation of motion for the expectation value @xmath136 is obtained by performing the shift @xmath137 imposing the condition @xmath138 to all orders in perturbation theory leads to the retarded equation of motion for @xmath139 . for a detailed presentation of this method in many other situations
see@xcite .
the important ingredients in this program are the real time green s functions for the stable coordinates which are assumed to be in thermal equilibrium . these are the following @xmath140 \\ & & & & \vspace{-2ex}\\ \left <
q^-_p(t)q^-_{p'}(t^\prime ) \right > & = & -i\,\delta_{p,-{p'}}g^{--}_{p}(t , t^\prime ) & = & -i \delta_{p,-{p ' } } \left[{\cal g}^>_p(t , t')\theta(t'-t ) + { \cal g}^<_p(t , t')\theta(t - t ' ) \right ] \\ & & & & \vspace{-2ex}\\ \left <
q^+_p(t)q^-_{p'}(t^\prime ) \right > & = & i\,\delta_{p,-{p'}}g^{+-}_{p}(t , t^\prime ) & = & -i\,\delta_{p,-{p'}}\,{\cal g}^<_p(t , t ' ) \\ & & & & \vspace{-2ex}\\ \left <
q^-_p(t)q^+_{p'}(t^\prime ) \right > & = & i\,\delta_{p,-{p'}}g^{-+}_{p}(t , t^\prime ) & = & -i\ , \delta_{p,-{p ' } } \ , { \cal g}^>_{p}(t , t')= -i\ , \delta_{p,-{p ' } } \,{\cal g}^<_p(t',t ) , \end{array } \label{greens}\ ] ] where @xmath141 \nonumber
\\ { \cal g}^<_p(t , t ' ) & = & \frac{i}{2\omega_p } \bigg [ \left(1\,+\,n_p\right)\mbox{exp } \left\{i\omega_p(t - t')\right\ } \,+\,n_p\,\mbox{exp}\left\{-i\omega_p(t - t')\right\ } \bigg]\nonumber \\ n_p & = & \frac{1 } { \mbox{exp}\left\ { \beta\omega_p \right\}\,-\,1 } .
\label{ggreater}\end{aligned}\ ] ] we will carry our derivation of the equation of motion in the linear theory where we will neglect nonlinear terms , @xmath142 and higher , and furthermore we will neglect the self - interaction of the unstable coordinate ( cubic and quartic terms ) . as mentioned before , the linearization of the equation of motion is consistent with the focus on the dissipative corrections to the growth rate , which is defined for small departures from the critical radius . in this case
, the non - equilibrium effective action @xmath143 becomes @xmath144 + l_0[q^+_p(t)]\,+\ , \tilde{q}^+(t ) \left [ -\ddot{q}(t ) + ( \omega^2 + \delta\omega^2)q(t ) - \sqrt{\lambda}~\theta(t ) \sum_{p , p'}{\cal b}_{p , p'}q_p^+(t)q_{p'}^+(t ) \right .
\right.\nonumber \\ & & \left .
\,-\ , 2\lambda ~ \theta(t ) \sum_{p , p'}{\cal a}_{p_1 p_2 } q_{p_1}^+(t)q_{p_2}^+(t ) q(t ) \,+\ , h \right ] \;-\;\sqrt{\lambda}~\theta(t ) \ ,
q(t ) \sum_{p , p'}{\cal b}_{pp ' } q_p^+(t ) q_{p'}^+(t ) \;-\ ; \left [ ( + ) \leftrightarrow ( - ) \right]\ ; \right\ } , \nonumber \\ & & \end{aligned}\ ] ] where we have written only the terms that are relevant to the lowest order calculation @xmath119 which is the focus of the present discussion and included the time dependence of the hamiltonian to set up the initial value problem@xcite . imposing the condition @xmath145 , up to @xmath119
, we obtain the following equation to one - loop order @xmath146 \ ; = 0 \nonumber \end{aligned}\ ] ] with @xmath147 since the correlation @xmath148 does not vanish for all values of @xmath149 and @xmath150 , this implies that the quantity between the square brackets must vanish@xcite .
furthermore , we now choose the counterterm in such a way to ensure that @xmath151 .
we thus obtain the equation of motion for the expectation value of the unstable coordinate for @xmath152 @xmath153 where the self energy @xmath154 is given by @xmath155 \,-\
, ( n_p - n_{p ' } ) \sin[(\omega_p-\omega_{p'})(t - t ' ) ] \right\},\label{sigmat } \end{aligned}\ ] ] and we have used the fact that @xmath156 .
the two different terms in the self energy , proportional to the sum and difference of frequencies respectively have a simple but important interpretation . the first term proportional to the sum of frequencies corresponds to the process in which the coordinate @xmath100 `` decays '' into two mesons with probability @xmath157 minus the `` recombination '' process with probability @xmath158 .
the second term proportional to the difference of frequencies originates in landau damping and corresponds to the scattering of the unstable coordinate with mesons in the medium , with probability @xmath159 minus the reverse process with probability @xmath160@xcite .
the counterterm @xmath161 will be chosen to cancel the tadpole contribution to @xmath162 ( see eqn .
( [ eqnmot ] ) ) as well as the _
static _ contribution from the self - energy , since this contribution is a _
static _ renormalization of the unstable frequency associated with a stationary bubble . in order to obtain the influence of thermal and quantum fluctuations on the growth rate of the bubble
, we must solve the equation of motion ( [ eqmv2 ] ) .
this is achieved via the laplace transform . introducing the laplace transforms of @xmath163 as @xmath164 respectively with @xmath165 the laplace variable .
the laplace transform of ( [ eqmv2 ] ) with the specified initial conditions is given by @xmath166 where @xmath167 is given by @xmath168 we now isolate the static contribution by substracting @xmath169 from the laplace transform of the self - energy . using the identity @xmath170
we now fix the counterterm @xmath161 such that @xmath171 . introducing @xmath172
the laplace transform of the equation of motion becomes @xmath173 in order to avoid cluttering of notation we will not write explicitly the @xmath174 in eqn .
( [ gamas ] ) in what follows , but it must always be understood that the term @xmath175 actually refers to its principal part .
this principal part prescription arises from the subtraction in the limit of vanishing @xmath165 which is the equivalent of the frequency in the real time domain , since the laplace transform requires that @xmath176 with @xmath177 the frequency@xcite . at this stage
it becomes clear that the procedure of absorbing the local ( zero frequency limit ) contributions to the self energy in the _
static _ renormalization of the growth rate @xmath18 provides the correct description of the dynamics .
viscosity and dissipative effects only arise from the time dependence of the expectation value of the coordinate and the _
static _ medium renormalization had already been absorbed into the definition of the growth rate as the limit of zero frequency .
viscosity and dissipation arise from the transfer of energy of the bubble wall to other excitations ( mesons ) and the growth slows down because of these processes .
an important distinction arises in this case as compared to the familiar situation in field theory in which complex poles determine the renormalization to the mass from the _ real _ part of the self - energy on shell and the decay width from the _ imaginary _ part of the self - energy on shell . in this case , however , because the frequency associated with the unstable mode is _ imaginary _ , the real part of the self - energy will renormalize the _ growth rate _ whereas the imaginary part ( if any ) at the position of the ( purely imaginary pole ) will provide _ oscillatory _ contributions to the bubble dynamics .
therefore , we emphasize that _ viscosity _ effects that will diminish the growth rate are determined by the _ real part of the self - energy _ at the position of the pole .
this is a striking difference from the usual case in which damping and viscosity are associated with the imaginary part of the self - energy on shell .
the real time dynamics of the bubble growth , @xmath139 , is given by the inverse laplace transform @xmath178 where @xmath179 is given by eq.([eqs ] ) and @xmath180 refers to the bromwich contour running along the imaginary axis to the right of all the singularities of @xmath179 in the complex @xmath165-plane .
the analytic structure of @xmath179 consists of cuts along the imaginary axis in the @xmath165-plane and poles .
the two different processes of decay into meson pairs ( and recombination ) and landau damping discussed above yield two different cut structures . the first term in eq . ( [ gamas ] )
gives a two - meson cut and the contribution from landau damping determines a cut structure that includes the origin in the s - plane .
the structure of these cuts depend on the matrix elements of the interaction as well as the full spectrum of excitations and will be investigated in detail for particular cases in the next sections .
the growth rate of the bubble with the quantum fluctuation effects included is given by the pole @xmath181 of @xmath179 , eq . ( [ eqs ] ) , which satisfies the following relation @xmath182 to first order in @xmath4 , the pole of @xmath179 which corresponds to a growing bubble is given to lowest order ( @xmath119 ) by @xmath183 where the first order quantum and thermal fluctuation correction @xmath184 to the bubble growth is given by @xmath185 we note that both terms inside bracket in ( [ delomega ] ) are _ positive _ since @xmath186 for @xmath187 , we recall again that the term @xmath188 from the discussion following eqn .
( [ gamas ] ) .
therefore we conclude that the correction to the growth rate is _ negative _ i.e. the dissipative effects of the coupling to mesons results in a smaller growth rate of supercritical bubbles .
neither the frequencies of oscillations around the stationary bubble nor the matrix elements are quantum mechanical in origin . however the one loop contribution @xmath119 to the self - energy is of quantum origin . if we restore @xmath189 in the expression above this results in @xmath190
. obviously neither the eigenvalues of the fluctuation operator nor the matrix elements depend on @xmath189 but the one loop contribution to the self - energy includes the @xmath189 from the coupling ( one loop ) as well as from the temperature factors . in the next sections we will see that the correction to the growth rate is dominated by low lying excitations and for high temperatures
these will be such that @xmath191 . in this case
we can invoke the classical limit which is best understood by restoring @xmath189 as mentioned above @xmath192 in this form , @xmath193 and in the high temperature limit we further approximate @xmath194 these approximations lead to a simplified expression for the lowest order correction to the growth rate @xmath195 where the product @xmath196 is independent of @xmath189 and displays clearly the classical limit . the classical limit will be justified in each particular case in the next sections . the expression ( [ delomegac ] ) is the final form of the one loop @xmath119 correction to the growth rate arising from _ dynamical viscosity _ since all of the static contributions had been absorbed by the counterterms .
this is as far as we can pursue in a general manner without addressing the details of the spectrum of fluctuations around the bubble solution . in the next two sections we study the details of the model determined by the lagrangian ( [ model ] ) with the potential ( [ pot ] ) for the cases of @xmath0 and @xmath1 dimensions
as mentioned in the introduction , the @xmath0 dimensional case is relevant in statistical and condensed matter physics .
quantum field theory models based on the lagrangian ( [ lagrangeana ] ) with potentials with a stable and metastable state are proposed to describe the low energy phenomenology of quasi - one dimensional charge density wave systems@xcite , and therefore their relevance in these physical situations warrants the study of this case . for @xmath23 given by eq .
( [ pot ] ) , the solution to the static classical equation of motion ( [ 1dbubble ] ) for one spatial dimension can be found exactly@xcite .
the critical bubble is found to be given by@xcite @xmath197-\tanh \left[\frac{x}{\xi}-s_0\right]\right\ } ~~;~~ \xi = \frac{2}{m},\ ] ] where @xmath36 is the width of the bubble wall and @xmath198 is given in terms of the critical radius @xmath11 by @xmath199 with @xmath200 given by eq .
( [ epsln ] ) .
the above solution corresponds to a kink - antikink pair centered at @xmath201 , and separated by a distance @xmath11 , and is displayed in fig .
( [ critbub ] ) .
this is the one dimensional bubble that starts at the false vacuum @xmath24 almost reaches the true vacuum @xmath25 and returns to @xmath24 and it is similar to the polaron solution found in quasi - one - dimensional polymers @xcite . the total energy of the bubble , given by eq .
( [ bubenergy ] ) , as a function of its dimensionless radius @xmath165 can be calculated from the field @xmath202 by replacing @xmath203 in eq .
( [ bubble1d ] ) .
the gradient term @xmath204 contributes to the surface energy while @xmath205 has two contributions ; surface contribution and volume contribution .
substituting @xmath205 in eq .
( [ bubenergy ] ) and evaluating the integral , one finds that the total energy of the bubble is given by @xmath206 with @xmath207 where @xmath208 is defined as @xmath209 the volume contribution grows linearly with the radius of the bubble while the surface contribution saturates at about the critical radius of the bubble , @xmath198 , and attains the following asymptotic value @xmath210 in the thin wall limit ( @xmath211 ) , the surface energy is simply twice the kink mass@xcite , as the kink - antikink pair becames widely separated and the exponential interaction between kink and antikink becomes negligible .
the total energy of the bubble is depicted in fig.([e1dfig ] ) .
it attains its maximum , @xmath70 , at the critical radius @xmath198 where @xmath212\ ] ] with @xmath213\end{aligned}\ ] ] in the thin wall limit , the above expressions simplify and we find @xmath214 and @xmath215 to study the fluctuations around the bubble solution we need the complete set of eigenfunctions @xmath216 satisfying @xmath217 { \cal u}_n(x)=\omega_n^2{\cal u}_n(x ) , \label{eigen}\ ] ] with @xmath218 - \frac{3\,m^2}{2}\mbox{sech}^2\left[\frac{x}{\xi}-s_0\right].\ ] ] although the spectrum of eigenfunctions and eigenvalues is known exactly in the case of one kink or antikink@xcite , for the case of the kink - antikink pair there are not known results that we are aware of . solving for the eigenfunctions @xmath216 in this case
is a difficult task but in the thin wall limit , the above potential consists of two identical and widely separated wells centered at @xmath219 .
the spectrum of each potential well is known in the literature@xcite .
it consists of a zero frequency mode localized in the well , an excited bound state with frequency @xmath220 that is also localized in the well and a continuum of scattering states .
we can use approximate methods such as the linear combination of atomic orbitals approximation ( lcao ) ( available in elementary textbooks ) to provide a reliable estimate for low lying bound states of the above potential from the spectrum of the single potential well . in this method ,
the ground state of the potential @xmath221 is the _ symmetric _ linear combination of the ground states of the single well located at @xmath219 while the first excited state of @xmath221 is the _ anti - symmetric _ linear combination . in the thin wall limit ,
these two states are given by @xmath222 and @xmath223 these are the unstable , @xmath88 , and the zero , @xmath224 , modes which we discussed earlier , see eqs .
( [ 1dzeromode ] ) and ( [ spectrum ] ) .
obviously @xmath225 is associated with translations since it is the spatial derivative of the bubble solution and must correspond to a vanishing eigenvalue by translational invariance . since it is antisymmetric
there has to be a nodeless eigenfunction of smaller frequency .
the symmetric combination is @xmath226 and since the two combinations will be split off in energy by a tunneling amplitude that is exponentially small in the distance between the kink and the antikink , the unstable frequency must be negative and exponentially small in this separation as is clearly displayed in eqn .
( [ 1dunstmode ] ) . besides the unstable and the zero modes ,
there are two bound states that have energies @xmath227 with energy difference which is again exponentially small in @xmath198 and correspond to the symmetric and antisymmetric linear combinations of the bound states of the kink in @xmath2 theory@xcite localized at @xmath219 .
since @xmath221 does not exactly reach @xmath228 near the center of the bubble , the potential in the schroedinger equation could allow for shallow bound states near the scattering continuum with binding energies that are exponentially small in the variable @xmath198 .
these bound states if present are extremely difficult to obtain .
finally there is the scattering continuum region of the spectrum characterized by functions @xmath229 , whose eigenfunctions are asymptotically phase shifted plane waves with eigenvalues @xmath230 in order to compute the matrix elements that enter in the expression for the correction to the growth rate ( [ delomegac ] ) we need either the exact form of the eigenfunctions or an excellent approximation to them . whereas we are confident of our analysis regarding the low lying bound states , we lack a full understanding of shallow bound and continuum states .
such an understanding requires a detailed study of the spectrum which certainly lies beyond the scope of this article .
although we do not have a complete understanding of the spectrum of eigenfunctions and therefore we do not have even a good approximation to the matrix elements , we can , however , provide some physically reasonable assumptions complemented with dimensional arguments to provide an estimate for the corrections in this case .
we begin by noting that in the one kink case , the potential that enters in the schroedinger equation for the fluctuation is reflectionless@xcite and the scattering states for the one kink ( or antikink ) case have a transmission amplitude which is a pure phase .
thus we expect that in the thin wall limit when the kink - antikink pair separation is much larger than the width of an isolated kink the wave functions of the scattering states will acquire a phase shift that is at least twice as large as that in the one kink case and will have a rather smooth dependence on the kink - antikink separation .
because the potentials for each kink are reflectionless@xcite we _ assume _ that in the thin wall limit the reflection coefficients of each potential well are very small and therefore there are no substantial interference effects in the region between the kink and the antikink .
under these assumptions the matrix elements @xmath231 will be similar to those calculated in reference@xcite for a single kink case and fall off very fast at large momenta justifying the classical limit@xcite .
hence , under these suitable assumptions the matrix elements are smooth functions of @xmath232 .
the contribution from the two - meson cut , i.e. the first term in the bracket in ( [ delomegac ] ) is proportional to @xmath18 because the @xmath18 in the denominator can be neglected in comparison with the frequencies for the meson states @xmath233 .
since in @xmath0 dimensions the coupling @xmath4 has dimensions of @xmath234 the correction to the growth rate arising from the two meson cut is of the form @xmath235\label{2meson}\ ] ] with @xmath236 $ ] a dimensionless slowly varying function of @xmath232 which is rather difficult to calculate and can only be obtained from a detailed knowledge of the eigenfunctions . the contribution from the landau damping cut is more complicated to extract . the second term in the bracket in ( [ delomegac ] )
has the form of a lorentzian and since @xmath18 is exponentially small it is a function that is strongly peaked at @xmath237 and the sum ( integral ) over @xmath238 is dominated by a region of width @xmath18 near @xmath239 . assuming that the matrix elements are smooth functions of momentum , in this one dimensional case the integral over a small region @xmath240 can be done by taking a narrow lorentzian and integrating over the relative momentum within this region@xcite , leading to a contribution of the form @xmath241 $ ] where @xmath242 $ ] is a smooth function of its argument that can only be calculated from a detailed knowledge of the scattering wavefunctions .
hence @xmath243\label{ld}\ ] ] and the total shift in the frequency is given by @xmath244 .
thus we conclude that although we do not have a complete knowledge of the eigenfunctions and therefore can not provide a complete calculation of the correction to the growth rate , suitable assumptions based on the properties of the spectrum of the one kink case combined with dimensional arguments suggest that to lowest order in the coupling and in the thin wall approximation the growth rate of a slightly supercritical bubble is given by @xmath245\right]- \frac{\lambda t^2 \xi^2}{4}c[\frac{r_c}{\xi}]+\cdots \label{estimate}\ ] ] with @xmath246 , c[\frac{r_c}{\xi}]$ ] slowly varying dimensionless functions of their argument .
obviously a full calculation of @xmath246 , c[\frac{r_c}{\xi}]$ ] requires a detailed understanding of the spectrum , in particular the scattering states .
however it is clear from ( [ estimate ] ) that the validity of a perturbative expansion places a severe constrain on the coupling constant and the value of the temperature , in particular the second contribution in ( [ estimate ] ) arising from landau damping gives the leading correction in the thin wall approximation and signals a potential breakdown of perturbation theory in this limit .
a more detailed understanding of this possiblity requires a better knowledge of the scattering matrix elements , this is an extremely difficult problem that depends on the details of the potential and lies outside the scope of this article .
we now study the @xmath1 dimensional case which is more relevant from the point of view of particle physics , however before focusing on a particular form of the potential and bubble profile , we can study fundamental model independent properties of the @xmath1 dimensional case that will determine very robust predictions for the corrections to the bubble growth .
the static bubble configuration @xmath247 is radially symmetric and satisfies the static equation ( [ 3dbubble ] ) . to study quantum fluctuations around the critical bubble configuration , we need to find the spectrum of the fluctuation operator @xmath248 which in 3 spatial dimensions is given by @xmath249 since the critical bubble solution is radially symmetric
, we write the eigenfunctions @xmath250 of the differential operator @xmath248 as a product of spherical harmonics @xmath251 and radial functions @xmath252 that satisfy @xmath253 because of the translational invariance of the lagrangian ( [ lagrangeana ] ) , there is a 3-fold degenerate zero mode given by the eigenfunction @xmath254 which correspond to translations of the bubble in 3-dimensions with no energy cost .
these are the goldstone modes of the spontaneously broken translational invariance .
this can be easily seen by taking the derivative of eq .
( [ 3dbubble ] ) with respect to @xmath255 which results in the following equation @xmath256 in addition to the goldstone mode , there are other low - lying excitations corresponding to different values of @xmath257 .
the eigenfunctions of these excitations and their eigenvalues can be obtained by writing the @xmath82 term in the above equation as @xmath258 and we re - write equation ( [ zeromod ] ) in the following form @xmath259 where @xmath260 \label{deltav}\end{aligned}\ ] ] since the function @xmath261 is strongly localized at @xmath262 in the thin wall limit , the second term on the right hand side is a _ small localized perturbation_. therefore the unperturbed lowest lying eigenvalues are given by @xmath263 in the thin wall limit , @xmath264 we find that the lowest order correction ( in @xmath265 ) to these eigenvalues is of order @xmath266 . for details see@xcite and appendix [ correction ] .
this analysis reveals that there is a band of low lying modes with eigenfunctions @xmath267 with @xmath268 the normalization constants , corresponding to the @xmath269-fold degenerate eigenvalues given by ( [ goldex ] ) . using equations ( [ wsecdv],[omegarelation],[spectrum ] ) and ( [ goldex ] ) with @xmath83 ( corresponding to the unstable mode )
we find the normalization to be given by @xmath270 in terms of the critical radius and the surface tension .
we summarize several noteworthy features of these low - lying solutions : * these excitations become goldstone modes in the limit as @xmath271 , i.e. in the limit in which the radius of curvature of the bubble goes to infinity .
this statement will be understood in detail below in connection with the case of flat interfaces in @xmath1 dimensions . *
the eigenfunction with the lowest eigenvalue , corresponds to @xmath83 , i.e. a spherically symmetric solution with a negative eigenvalue given by@xcite @xmath272 this is the unstable mode which corresponds to a spherically symmetric expansion or contraction of the bubble and therefore corresponds to the unstable functional direction .
the coordinate associated with this mode is the displacement from the critical radius . * the three goldstone modes @xmath273 are the translation modes . *
the higher energy modes with @xmath274 are excitations on the surface of the bubble , or surface waves with energies given by @xmath275 .
these low lying modes will play a dominant role and we will refer to them collectively as @xmath276 with eigenvalues given by ( [ goldex ] ) and whose normalized eigenfunctions ( [ lolyin ] ) are simple functions of the bubble configuration which for the potential ( [ pot ] ) in the thin wall approximation are given by eqn .
( [ spcquasi ] ) below . that the modes with @xmath277 can be identified as wiggles of the bubble surface , or surface waves , can be seen from the following expansion@xcite where as discussed before the translational modes is not included because it is `` clamped '' @xmath278 where @xmath279 this is an important identification that we emphasize : _ this band of low - lying modes describes fluctuations of the surface of the bubble_. we will argue below that these excitations dominate the infrared behavior of the viscosity correction and will provide the largest contribution to the viscosity coefficient .
the low lying spectrum described above is fairly general and only depends on the existence of a thin wall bubble .
depending on the form of potential @xmath23 , there might be other bound states .
an analysis similar to that leading to the band of low - lying excitations reveals that for the potential ( [ pot ] ) there is another band of rotational bound state excitations that starts near @xmath280 . in the thin wall
limit the radial wave function for the lowest state in this band is given by the bound state of energy @xmath220 of the @xmath2 theory in @xmath0 dimensions@xcite which is also localized at the wall . the eigenfunctions and eigenvalues for this rotational band of excitations
are given by ( see appendix a ) @xmath281 where @xmath282 is given by the bound state of the theory with potential ( [ pot ] ) in one spatial dimension@xcite , the eigenfunctions in this band are given in eq .
( [ nextbs ] ) below .
finally there is a continuum of scattering eigenstates with eigenvalues @xmath283 .
as will be discussed later , the contribution to the growth rate from the rotational band ( [ nextband ] ) and of the scattering states is subleading in the thin wall limit . the maximum value of angular momentum available for the low - lying part of the spectrum ( [ goldex ] )
is limited by the edge of the continuum spectrum or the presence of higher bound states , hence @xmath284 or @xmath285 .
therefore in the thin wall limit the maximum value of the angular momentum @xmath286 .
the low lying spectrum of eigenfunctions given by ( [ lolyin ] ) with eigenvalues ( [ goldex ] ) has a simple physical origin , which can be understood by noticing that in the limit @xmath271 the discrete spectrum becomes a continuum . in the limit when the radius of the critical bubble is very large @xmath271 , the interface between the two phases @xmath25 and @xmath24 becomes planar and the two phases become degenerate .
let us consider a static planar interface configuration corresponding to a domain wall along the z - axis in three spatial dimensions .
such a configuration satisfies the following equation @xcite @xmath287 where @xmath288 satisfies the boundary condition @xmath289 as @xmath290 and @xmath291 is the coordinate perpendicular to the planar interface .
the quantum fluctuations around the classical static wall solution @xmath288 are given by the spectrum of the differential operator @xmath292 since the domain wall only depends on the coordinate z , the differential operator for the fluctuations is separable in terms of eigenfunctions @xmath293 , where @xmath294 denotes the transverse coordinates to @xmath291 , namely @xmath80 and @xmath295 , and @xmath296 the transverse momentum .
the functions @xmath297 are so lution of the following eigenvalue problem @xmath298\psi_n(z ) = \omega^2_n(\vec{q}_{\bot})\psi_n(z ) \label{flucwall}\ ] ] taking the derivative of eq .
( [ walleqn ] ) with respect to @xmath291 , i.e. @xmath299 \frac{d \phi_{w}}{dz}\,=\,0\ ] ] and comparing to eq .
( [ wfluctopr ] ) we see that @xmath300 is the zero mode which corresponds to the translational invariance@xcite .
therefore the eigenfunctions @xmath301 have eigenvalues @xmath302 .
these are goldstone modes associated with translational invariance and represent excitations of the surface of the planar interface @xmath303 since @xmath304 where @xmath305 is the position of the planar interface and @xmath306 this is clearly similar to the case of the spherical bubble eq .
( [ surfaceflucs ] ) and describes the same physics , i.e. fluctuations of the surface . in the case of a planar interface these surface waves
are also called capillary waves , and describe the _ hydrodynamic _ modes of long - wavelength fluctuations of interfaces in systems with two degenerate phases separated by an interface@xcite . in the case of degenerate phases , such as for example a liquid - gas or an ising system , these surface waves are goldstone modes associated with the breakdown of translational invariance by the presence of the interface . for a spherical bubble in the thin wall limit
these surface waves acquire a gap given by the inverse radius ( proportional to the gaussian curvature of the surface ) .
therefore in analogy with the case of interfaces for degenerate separated phases , we identify these surface fluctuations as _ quasi goldstone modes_. since , as argued before the maximum frequency of the surface waves is @xmath307 they are _ classical _ in the high temperature limit @xmath308 .
hence the surface waves are identified as _ classical hydrodynamic fluctuations _ of the bubble shape and quasi - goldstone modes in the thin wall limit .
we want to emphasize that these low energy excitations are a robust feature of the thin wall approximation and are _
model independent_. having studied in detail the _ general _ aspects of fluctuations around a thin wall bubble , we now focus on the specific details of the theory with a potential given by ( [ pot ] ) so as to be able to compute the matrix elements and provide a quantitative analysis of the viscosity effects .
it corresponds to a field configuration that starts around @xmath313 at the true vacuum @xmath25 , given by eq .
( [ phipls ] ) , and goes to the false vacuum , @xmath24 , as @xmath50 with a surface width @xmath314 and a critical radius @xmath11 .
having specified the critical bubble solution @xmath315 , we now go back and determine explicitly the general expressions which we discussed in the previous sections . the total energy of the bubble as a function of its radius @xmath54 , given by eq.([bubenergy ] ) , can be calculated from the field @xmath247 by replacing @xmath11 in eq .
( [ bubble ] ) by @xmath54 .
the gradient term @xmath316 contributes to the surface energy while @xmath205 has two contributions : a surface contribution @xmath317 and volume contribution @xmath318 given by @xmath319 \left ( 3 - \tanh\left[\frac{r - r}{\xi } \right ] \right ) \right\ } \\
v_v(\phi_b ) & = & -\frac{\eta \,m^4}{8\,\lambda } \left(1 - \tanh\left[\frac{r - r}{\xi } \right ] \right)\end{aligned}\ ] ] where @xmath208 is defined by eq . ( [ eta ] ) . substituting the above expressions in eq . ( [ bubenergy ] ) and evaluating the integral
, we find that the total energy of the bubble is given by @xmath320 \label{energy3dbubb}\ ] ] where in the thin wall limit @xmath321 the total energy attains its maximum at @xmath322.\ ] ] using the fact that @xmath208 is small in the thin wall limit , we find that the critical radius @xmath11 is given by @xmath323\;\ ; \rightarrow\;\ ; \frac{\xi}{r_c } \,\simeq \,3 \ , \eta\ ] ] and the total critical energy is given by @xmath324 which is equivalent to eqs .
( [ rc ] ) and ( [ ec ] ) with the surface tension given by @xmath325 the low - lying fluctuation modes @xmath326 eqn .
( [ lolyin ] ) are given by @xmath327 y_{lm}(\theta,\varphi ) \;\;\;\mbox{with}\;\;\;\omega^2_{0l}\,=\ , \frac{(l-1)(l+2)}{r_c^2 } \left[1+{\cal o } \left ( \frac{\xi^2}{r_c^2}\right ) \right]\label{spcquasi}\end{aligned}\ ] ] and the next rotational band of bound states is given by ( for details see appendix a ) @xmath328\mbox{tanh}\left[\frac{r - r_c}{\xi}\right ] y_{lm}(\theta,\varphi ) \nonumber \\ & & \mbox{with}\;\;\;\omega^2_{1l}\,=\ , \frac{3m^2}{4}+\frac{l(l+1)}{r^2_c}+{\cal o}\left(\frac{\xi}{r_c}\right ) \label{nextbs}\end{aligned}\ ] ] from eq .
( [ delomegac ] ) , the corrections from quantum and thermal fluctuations to the bubble growth in the present case has the following form in the classical limit @xmath329 where the index @xmath330 runs over bound and scattering states and @xmath331 with @xmath332\ ] ] since @xmath333 is spherically symmetric , the angular integral leads to @xmath334 with @xmath335 \mbox{sech}^2 \left[\frac{r - r_c}{\xi } \right]\psi_{ql}(r)\psi_{q'l}(r ) \label{bql}\ ] ] we now note that the bound state energies only depend on @xmath336 , and that the energy of the scattering states is independent of @xmath336 .
therefore there is _ no landau damping contribution _ from the bound states as consequence of the principal part prescription which subtracts the contribution from @xmath237 in the landau damping term as discussed in detail below equations ( [ gamas ] , [ delomega ] ) .
therefore the correction to the growth rate can be written as @xmath337 with the correction from the bound states given by @xmath338 and that from the scattering states given by @xmath339 the bound state correction has three contributions ; the _ quasi - goldstone _ modes contribution corresponding to surface waves with @xmath340 , the contribution from the higher energy rotational band of bound states near @xmath341 and the mixed contribution from the quasi - goldstone and the higher energy bound states modes .
the denominators in @xmath342 are of order @xmath343 for the quasi - goldstone modes contribution as compared to @xmath344 for the mixed and the higher energy bound states contributions .
since in the thin wall approximation @xmath345 , the largest contribution arises from the quasi - goldstone surface modes with the lowest energy denominators .
it can be easily seen that the matrix elements can not compensate for the difference in powers of @xmath11 and that in fact for the higher energy bound states these matrix elements are _ smaller _ than those for the surface modes because the wave functions @xmath346 actually vanish at the position of the bubble wall ( see eqn .
[ nextbs ] ) ) .
the contribution from the scattering states is also seen to be much smaller than that from the surface waves .
the frequencies for the continuum states @xmath347 and the two meson cut give a contribution of order @xmath18 ( since in the denominators the @xmath18 can be neglected as compared to @xmath30 ) .
for the landau damping cut , the argument is similar to that of the case of one space dimension .
for @xmath348 this contribution has a lorentzian shape of width @xmath349 , and the integral over the momenta can be performed in the narrow width approximation .
the subtraction of the static contribution guarantees that the integral is dominated by the lorentzian@xcite and the region @xmath350 can be integrated by changing to relative variables and now in three spatial dimensions the phase space in the region @xmath351 give extra powers of @xmath18 as compared to the one dimensional case .
furthermore , the matrix elements are smooth functions of the radius of the bubble as can be understood simply by a scattering argument from a sharply peaked potential in three spatial dimensions .
these matrix elements do not introduce any singularity in the limit @xmath352 , hence the contribution of the scattering states is at least proportional to @xmath18 and is therefore subleading in the thin wall limit .
this analysis leads to the conclusion that the largest contribution to the correction to the growth rate is given by the _ quasi - goldstone modes _ , i.e. the surface waves , since these are the lowest lying excitations and hence provide the smallest energy denominators .
the matrix element @xmath353 for the surface waves can be calculated easily and we find @xmath354 leading to the following correction @xmath355 where @xmath356 the above series converges rapidly and only the first few terms contribute to the sum , we have evaluated the sum numerically with @xmath357 .
hence we summarize one of the main results of this article : the lowest order correction to the bubble growth for three dimensional bubbles is dominated by viscosity effects arising from the excitation of long - wavelength surface waves and is given by @xmath358 therefore to lowest order in @xmath4 ( one - loop ) and to leading order in the thin wall approximation we find that the growth rate of slightly supercritical bubbles is given by @xmath359 \label{finalgrowthrate}\ ] ] with @xmath36 and @xmath11 are the width and radius of the critical bubble .
this is one of the important results of our study .
the validity of perturbation theory places a very stringent constraint on the quartic coupling constant in the thin wall limit @xmath352 and in the classical limit @xmath360 .
the validity of the classical limit in this case is warranted : we are studying the dynamics of nucleation via thermal activation for temperatures below the critical temperature @xmath361 but for temperatures much larger than the energy of the low lying excitations , the relevant regime for thermal activation is @xmath362 in the weak coupling limit .
furthermore the low - lying excitations with frequencies @xmath363 are obviously classical .
the focus of this article is to provide a microscopic calculation of the growth rate of slightly supercritical nucleation bubbles .
the model under consideration is a @xmath2 scalar theory with an explicitly symmetry breaking term that produces a metastable and a stable ground state , we studied the case of nucleation in @xmath0 dimensions as well as @xmath1 dimensions .
the former is relevant in the case of quasi - one dimensional charge density wave systems and organic conductors .
we begin our analysis by obtaining the critical bubble solution by including finite temperature effects in the potential that enters in the classical equation of motion , counterterms are added to the lagrangian to compensate for the finite temperature corrections consistently in a perturbative expansion .
our approach to obtaining the growth rate is very different from previous treatments in that we begin by expanding the quantum field around the critical bubble in terms of the quadratic fluctuations around the critical bubble configuration .
these fluctuations describe an unstable direction associated with small departures from the critical radius , translational zero modes and stable fluctuations .
the translational modes are anchored by fixing the center of the bubbles and we treat explicitly the interaction between the coordinate associated with the growth ( or collapse ) of the bubble with those associated with the stable fluctuations . we obtain the growth rate by obtaining the effective linearized equation of motion for the unstable coordinate by integrating out the coordinates associated with the stable fluctuations .
two different approximations are involved ; i ) a weak coupling expansion in terms of @xmath4 the quartic self - coupling , and ii ) the thin wall approximation in terms of @xmath364 with @xmath36 the width of the bubble wall and @xmath11 the critical radius .
the first approximation allows a consistent perturbative expansion of the self - energy of the unstable coordinate , the second allows a quantitative calculation of the relevant matrix elements , furthermore our analysis reveals that the important fluctuations are _ classical _ for temperatures @xmath365 , with @xmath366 the critical temperature and @xmath30 the mass of quanta in the metastable phase . in the one dimensional case we are able to provide an estimate for the growth rate given by eq .
( [ estimate ] ) where the functions @xmath20~,~ c[r_c/\xi]$ ] depend in a detailed manner upon the scattering states solutions of the eigenvalue problem for the quadratic fluctuations , and clearly will depend on the details of the potential .
this estimate points out the potential breakdown of perturbation theory in the thin wall limit . in the case of three dimensions we are able to extract some robust features that trascend the form of the potential and are solely a consequence of the thin wall limit .
in particular we identify a rotational band of low lying excitations which describe _ surface waves _
i.e. ripples on the surface of the bubble .
we establish the connection of these surface waves to the capillary waves of flat interfaces in the case of degenerate but phase separated thermodynamic states ( such as the ising or liquid - gas at coexistence ) , the surface waves are then identified as _ classical long - wavelength hydrodynamic fluctuations_. the unstable coordinate couples to these hydrodynamic fluctuations and as a result the friction term arising from the self - energy is dominated by the coupling to these hydrodynamic modes .
clearly the _ coupling _ of the unstable coordinate to these hydrodynamic modes depends on the model , and in the case under consideration we find in the thin wall limit and to lowest order in perturbation theory the following expression for the growth rate @xmath359\ ] ] we also obtain the effective _ non - markovian _ langevin equation for the coordinate describing small departures from the critical radius and establish the generalized fluctuation dissipation relation between the viscosity and the noise kernel .
the noise is correlated on time scales comparable to @xmath7 precisely as a consequence of the coupling to the hydrodynamic modes and can not be treated simply with white ( delta function ) correlations .
* discussion : * although we have studied a specific microscopic model , in the case of @xmath1 dimensions we have been able to identify some robust features that trascend the particular model .
these are the existence and dominance of hydrodynamic fluctuations associated with surface waves in the thin wall limit .
the coupling of the coordinate associated with small departures from the critical radius to these low energy fluctuations induces friction or viscosity corrections to the growth rate of slightly supercritical bubbles and in the weak coupling and thin wall limit these fluctuations give the largest contribution to the friction corrections .
one of our original motivations is to make contact with previous studies of nucleation as applied to the quark - hadron phase transition .
in particular csernai and kapusta@xcite have parametrized the coarse grained free energy that describes a quark - hadron first order phase transition in terms of a local energy variable that can be identified with our scalar field @xmath27 .
the form of the potential taken by these authors coincides with our potential @xmath23 ( [ pot ] ) , with coefficients that depend on temperature , just as we have argued in this article .
their form of the critical bubble solution and the variational energy as a function of the radius of a bubble are very similar to those studied in this article .
these authors have established that for about @xmath367 supercooling the critical radius for such a model is about @xmath368 the width of the wall is about @xmath369 and the thin wall approximation must be reliable in this regime .
we can obtain the quartic self - coupling @xmath4 in ( [ pot ] ) from the parameters used in@xcite by relating the width of the bubble @xmath370 and the surface tension @xmath67 to @xmath4 via eqn .
( [ surfacetension ] ) .
in@xcite the value of the surface tension for the particular quark - hadron model is @xmath371 for @xmath372 , yielding a value @xmath373 which leads to a very large correction to the growth rate .
certainly the large value of the coupling invalidates the perturbative scheme and we can not draw a definite conclusion as to the relevance of our lowest order estimate for the quark - hadron transition beyond the statement that the viscosity induced by the hydrodynamic fluctuations _ could _ result in a very large ( negative ) correction to the growth rate and therefore a rather small nucleation rate . for larger
supercooling the critical radius becomes smaller and the thin wall approximation breaks down , but in this case nucleation and spinodal decomposition will be indistinguishable and homogenous nucleation theory may not be the proper description .
however despite the limitations of the perturbative expansion and the thin wall approximation , we have provided a consistent approach to obtain friction or viscosity corrections to the growth rate from a _ microscopic _ perspective without invoking a phenomenological description .
the observation that in the thin wall limit the most important corrections arise from the coupling to _ classical _ hydrodynamic fluctuations could perhaps pave the way to a systematic hydrodynamic treatment that would circumvent the weak coupling expansion and allow to extract non - perturbative physics .
d. b. thanks the n.s.f for partial support through the grant awards : phy-9605186 and int-9512798 .
s. m. a. thanks king fahad university of petroleum and minerals ( saudi arabia ) for financial support .
s. e. j. thanks cnpq for finantial support .
f. i. t. thanks fapemig for support .
e. s. f. thanks cnpq and faperj for support and d. g. b. thanks faperj and cnpq for support through a binational collaboration .
in this appendix we calculate the corrections to the eigenvalues of the quasi - goldstone modes that obey the eq .
( [ lowener ] ) using first order perturbation theory .
first we write @xmath374 with @xmath375 first order energy correction @xmath376 are given by @xmath377 where we have integrated out the angular degrees of freedom . for the @xmath2 potential given by eq .
( [ pot ] ) , the normalized wavefunctions are given by eq .
( [ spcquasi ] ) and when it is substituted in eq .
( [ fenecor ] ) one finds that @xmath378 thus to first order in perturbation theory , the eigenvalues are given by @xmath379.\ ] ] a similar treatment can be used for the next rotational band based on the bound state of the @xmath2 theory in one spatial dimension . for this
we treat the term @xmath380\ ] ] as a perturbation .
the unperturbed wave function is the positive energy bound state of the one dimensional @xmath2 theory@xcite and the first order correction is obtained as before and is @xmath381 .
the semiclassical langevin equation is obtained by performing the path integrals over the bath degrees of freedom , i.e. the stable modes , thus obtaining a non - equilibrium effective functional for the unstable mode .
this is achieved consistently in perturbation theory@xcite , and to lowest order we find where the lagrangians @xmath383 $ ] are given by eq .
( [ unlag ] ) and @xmath384 $ ] is the influence functional@xcite which to one loop order and neglecting the tadpole contributions is given by @xmath385 & = & \mbox{exp}\bigg\{-i^2 \lambda\ , \sum_{p , p'}|{\cal b}_{pp'}|^2 \int dt\ , dt^\prime \biggl[q_{-1}^+(t ) \ ,
g_p^{++}(t , t^\prime ) g_{p'}^{++}(t , t^\prime)\ , q_{-1}^+(t^\prime ) \nonumber \\ & & \,+\ ; q_{-1}^-(t ) \ , g_p^{--}(t , t^\prime)g_{p'}^{--}(t , t^\prime ) \ , q_{-1}^-(t^\prime ) - \ ; q_{-1}^+(t ) \,g_p^{+-}(t , t^\prime)g^{+-}_{p'}(t , t^\prime ) \,q_{-1}^-(t^\prime ) \nonumber \\ & & \,-\ ; q_{-1}^-(t ) \ , g_p^{-+}(t , t^\prime ) g_{p'}^{-+}(t , t^\prime)\ , q_{-1}^+(t^\prime ) \biggr ] \ , \bigg\}\end{aligned}\ ] ] where the green s functions are given by eq .
( [ greens ] ) . introducing the wigner coordinates or center of mass and relative coordinates , @xmath386 and @xmath387 respectively given by @xmath388 the generating functional becomes@xcite-@xcite @xmath389 \,=\ , \int { \cal d}x\,{\cal d}r\ , \;e^{is[x , r ] } \label{slanvin}\ ] ] with the non - equilibrium effective action given by @xmath390 \,=\ , \int dt \ , r(t ) \ , \left [ - \ddot{x}(t ) \,+\ , \omega^2\ , x(t)\,+ \lambda \int dt^\prime\,\biggl ( \sigma_{\mbox{ret } } ( t - t^\prime)\,x(t^\prime ) \,+ \frac{i}{2}\,k(t - t^\prime)\,r(t^\prime ) \biggr ) \right ] .
\label{lanaction}\ ] ] where for clarity we have neglected the counterterms as well as the tadpoles arising in the computation of the influence functional . the kernels @xmath391 and @xmath392 are given by @xmath393 \right.\nonumber \\ & & \;\;\;+\,\left .
( n_p+n_{p'}+2n_pn_{p ' } ) \cos[(\omega_p-\omega_{p'})(t - t ' ) ] \right\},\nonumber\end{aligned}\ ] ] with @xmath394 given by ( [ sigmat ] ) . at this stage
it proves convenient to introduce the identity @xmath395 with c(t ) being an inessential normalization factor , to cast the non - equilibrium effective action of the unstable mode in terms of a stochastic noise variable with a definite probability distribution@xcite-@xcite . using the above relation , the generating functional becomes @xmath396 & = & \int { \cal d}x\,{\cal d}r\,{\cal d } \xi \,p[\xi]\ ; \mbox{exp } \biggl\{i \int dt\ ; r(t ) \biggl [ - \ddot{x}(t ) \,+\ , \omega^2 \ , x(t)\,+\,\lambda \int_{-\infty}^t dt^\prime\ , \sigma ( t - t^\prime)\,x(t^\prime)\,+\ , \xi(t ) \biggr ] \biggr\ } , \nonumber \\ \label{zlanng}\end{aligned}\ ] ] where the probability distribution of the stochastic noise , @xmath397 $ ] , is given by @xmath398\,=\,\int { \cal d } \xi \;\mbox{exp}\left\ { -\frac{1}{2 \lambda } \int dt \ , dt^\prime \xi(t)k^{-1}(t - t^\prime)\xi(t^\prime ) \right\}.\ ] ] in this approximation we find that the noise - noise correlation function is given by @xmath399 which is in general colored , i.e. it is not a delta function @xmath400 .
the semiclassical langevin equation is obtained by extremizing the effective action in eq .
( [ zlanng ] ) with respect to @xmath387 @xmath401 taking the average of the above equation with the noise probability distribution @xmath397 $ ] and identifying @xmath402 yields the equation of motion for the expectation value of the unstable mode , eq.([eqmv2 ] ) .
the relationship between the kernels @xmath391 and @xmath392 constitutes a generalized quantum fluctuation dissipation relation .
this relation is established by considering the time fourier transforms of the functions @xmath403 and @xmath404 which are denoted by @xmath405 and @xmath406 respectively .
these fourier transforms obey the kms condition@xcite @xmath407 using the above relation we find that @xmath408 , the fourier transform in time of the retarded self energy @xmath391 is given by @xmath409}{\omega-\omega'+i\epsilon},\ ] ] leading to the imaginary part @xmath410 = - \left(1-e^{-\beta \omega}\right ) \sum_{p , p ' } \vert { \cal b}_{p p ' } \vert^2 { \cal g}_{pp'}^>(\omega).\ ] ] on the other hand the kernel that determines the noise - noise correlation function @xmath411 has a fourier transform given by @xmath412 with @xmath413 im[\sigma_{\mbox{ret}}(\omega ) ]
. \label{flucdiss}\end{aligned}\ ] ] the above relation between the fourier transform of the noise - noise correlation function and the imaginary part of the self - energy is the generalized fluctuation - dissipation relation .
this function oscillates with constant amplitude on a time scale @xmath415 which is the same time scale for growth of a supercritical bubble .
therefore the noise term _ can not _ be taken to be uncorrelated over the time scales associated with the growth of a bubble , i.e. the noise is _ colored _ and a langevin description based on a white noise would miss the long - time correlations . the contribution of the scattering states _ could _
lead to a short range part of the kernel , but the long time behavior of the kernel will be dominated by the low energy surface waves .
the reason that a markovian langevin equation with white noise fails to describe the dynamics of the unstable coordinate is that there are slow hydrodynamic fluctuations with time scales comparable to the growth rate that couple to the unstable coordinate , precisely the same type of fluctuations that dominate the viscosity or friction .
99 e. v. shuryak , phys . rep . * 61 * , 71 , ( 1980 ) ; h. stcker and w. greiner , phys
137 * , 277 ( 1986 ) ; b. muller , _ the physics of the quark gluon plasma _ lecture notes in physics , vol . 225 ( springer - verlag , berlin , heidelberg , 1985 ) ; l. p. csernai , `` introduction to relativistic heavy ion collisions '' , ( john wiley and sons , england , 1994 ) ; c. y. wong , `` introduction to high - energy heavy ion collisions '' , ( world scientific , singapore , 1994 ) .
v. a. rubakov and m. e. shaposhnikov , phys .
usp . * 39 * , 461 ( 1996 ) ; g. r. farrar and m. e. shaposhnikov , phys .
d * 50 * , 774 ( 1994 ) ; v. a. kuzmin , v. a. rubakov and m. e. shaposnikov , phys .
b*155 * , 36 ( 1985 ) ; g. farrar , nucl.phys.proc.suppl . * 43*,312 ( 1995 ) . m. e. shaposhnikov , phys . lett .
b*277*,324 ( 1992 ) . for a review of possible mechanisms for baryogenesis
see : a. dolgov , physics reports , bf 222 , 309 ( 1992 ) ; n. turok , _
`` electroweak baryogenesis '' _ , in _ perspectives on higgs physics _
g. l. kane , ( page 300 ) ( 1993 ) ( world scientific , singapore ) ; m. trodden , _
`` electroweak baryogenesis '' _ , hep- ph/9803479 ( to appear in rev .
mod . phys . ) ; a. riotto , _ `` theories of baryogenesis '' _ hep - ph/9807454 . for a review of nucleation in soft condensed matter physics see : j. d. gunton , m. san miguel and p. s. sahni , in _ phase transitions and critical phenomena_,ed .
by c. domb and j. l. lebobowitz
( academic press , london , 1983 ) , vol . 8 .
yu - lu , _ solitons and polarons in conducting polymers _
( world scientific , singapore , 1988 ) ; a. j. heeger , s. kivelson , j. r. schrieffer and w. p. su , rev .
phys . * 60 * , 781 ( 1988 ) ; g. grner , `` density waves in solids '' ( addison - wesley , 1994 ) .
d. boyanovsky , c. a. a. de carvalho and e. s. fraga , phys .
b * 50 * , 2889 ( 1994 ) ; c. a. a. de carvalho , acta physica polonica b * 19 * , 875 ( 1988 ) ; mod .
phys . lett . *
b 3 * , 125 ( 1989 ) ; e. s. fraga and c. a. a. de carvalho , phys .
b * 52 * , 7448 ( 1995 ) ; d. g. barci , e. s. fraga and c. a. a. de carvalho , phys . rev .
d * 55*,4947 ( 1997 ) .
j. s. langer , ann . of phys .
( ny ) * 41 * , 108 ( 1967 ) , ann .
( ny ) * 54 * , 258 ( 1969 ) ; ann .
( ny ) * 65 * , 53 ( 1971 ) ; j. s. langer and l. a. turski , phys
. rev . a*8 * , 3230 ( 1973 ) ; l. a. turski and j. langer , phys . rev . a * 22 * , 2189 ( 1980 ) .
j. and m. gleiser , phys.rev .
d51 ( 1995 ) , 4111 ; m. gleiser , phys . rev .
73 ( 1994 ) 3495 ; m. alford and m. gleiser , phys.rev .
d48 ( 1993 ) , 2838 ; m. gleiser , g. c. marques , and r. o. ramos , phys.rev .
d48 ( 1993 ) 1571 ; m. gleiser , a. f. heckler , e. w. kolb , phys.lett .
b405 ( 1997 ) 121 ; m. gleiser and a. f. heckler , phys.rev.lett .
76 ( 1996 ) 180 .
d. boyanovsky , h. j. de vega , phys.rev .
* d47*,2343 ( 1993 ) ; d. boyanovsky , d.s . lee and anupam singh , phys.rev.*d48 * 800 ( 1993 ) ; d. boyanovsky , m. dattanasio , h. j. de vega , r. holman and d. s. lee , phys . rev . * d52 * , 6805 ( 1995 ) .
j. schwinger , j. math .
* 2 * , 407 ( 1961 ) .
l. v. keldysh , sov .
jetp * 20 * , 1018 ( 1965 ) .
k. t. mahanthappa , phys . rev . * 126 * , 329 ( 1962 ) ; p. m. bakshi and k. t. mahanthappa , j. math . phys . * 4 * , 1,12 ( 1963 ) . v. korenman , ann .
phys . * 39 * , 72 ( 1966 ) .
g. z. zhou , z. b. su , b. l. hao , and l. yu , phys .
rep . * 118 * , 1 ( 1985 ) .
j. rammer and h. smith , revs . of mod
58 * , 323 ( 1986 ) .
e. m. lifshitz and l. p. pitaevskii , _ physical kinetics _
( pergamon , new york , 1981 ) ; g. d. mahan , _ many particle physics _ , 2nd ed .
( plenum , new york , 1990 ) ; h. kleinert , _ path integrals in quantum mechanics , statistics and polymer physics _ , 2nd ed .
( world scientific , singapore , 1996 ) ; r. mills , _
propagators for many particle systems _
( gordon and breach , new york , 1969 ) .
d. boyanovsky , h. j. de vega , r. holman , d .- s .
lee and a. singh , phys . rev .
* d51 * , 4419 ( 1995 ) .
d. boyanovsky , h. j. de vega and r. holman , proceedings of the second paris cosmology colloquium , observatoire de paris , june 1994 , pp . 127 - 215 , h. j. de vega and n. snchez , editors ( world scientific , 1995 ) ; advances in astrofundamental physics , erice chalonge school , n. snchez and a. zichichi editors , ( world scientific , 1995 ) .
d. boyanovsky , h. j. de vega and r. holman , vth .
erice chalonge school , current topics in astrofundamental physics , n. snchez and a. zichichi editors , world scientific , 1996 , p. 183 - 270 .
n. j. gnther , d. a. nicole and d. j. wallace , j. phys .
a , * 13 * , 1755 ( 1980 ) ; d. j. wallace , in _ recent advances in field theory and statistical mechanics _ , edited by j - b .
zuber and r. stora ( north - holland publishing co. , amsterdam , 1984 ) . | we study the viscosity corrections to the growth rate of nucleating bubbles in a slightly supercooled first order phase transition in @xmath0 and @xmath1 dimensional scalar field theory .
we propose a _
microscopic _
approach that leads to the non - equilibrium equation of motion of the coordinate that describes small departures from the critical bubble and allows to extract the growth rate consistently in a weak coupling expansion and in the thin wall limit .
viscosity effects arise from the interaction of this coordinate with the stable quantum and thermal fluctuations around a critical bubble . in the case of @xmath0 dimensions we provide an estimate for the growth rate that depends on the details of the free energy functional . in @xmath1 dimensions we recognize robust features that are a direct consequence of the thin wall approximation that trascend a particular model .
these are _ long - wavelength hydrodynamic fluctuations _ that describe surface waves .
we identify these low energy modes with quasi - goldstone modes which are related to surface waves on interfaces in phase ordered ising - like systems . in the thin wall limit the coupling of this coordinate to these hydrodynamic modes results in the largest contribution to the viscosity corrections to the growth rate . for a @xmath2 scalar field theory at temperature @xmath3
the growth rate to lowest order in the quartic self - coupling @xmath4 is @xmath5 $ ] with @xmath6 the critical radius and the width of the bubble wall respectively .
we obtain the effective non - markovian langevin equation for the radial coordinate and derive the generalized fluctuation dissipation relation , the noise is correlated on time scales @xmath7 as a result of the coupling to the slow hydrodynamic modes .
we discuss the applicability of our results to describe the growth rate of hadron bubbles in a quark - hadron first order transition . |
x - ray emission from hot gas appears to be a generic feature of massive halos .
hot gaseous halos suffuse nearly every galaxy cluster and many galaxy groups ( @xcite , @xcite , @xcite , @xcite ) , and they are also very common ( possibly ubiquitous ) around massive elliptical galaxies ( @xcite , @xcite ) , including field ellipticals @xcite .
the hot halo is strongly affected by both feedback from the galaxy and by accretion from the intergalactic medium ( e.g. @xcite , @xcite ) .
these processes , which are fundamental for understanding galaxy formation , can therefore be studied through x - ray observations of the hot halos . in the aggregate
, x - ray emission can be described by power - law relations as a function of stellar mass ( e.g. @xcite , @xcite , @xcite , @xcite ) and halo mass ( @xcite , @xcite ) , although the scatter in these relations is considerable .
potentially active galactic nucleus ( agn ) feedback ( e.g. @xcite , @xcite ) might also be important .
for example , the slopes of these power - law relations differ from the self - similar prediction , which can be used as a constraint on the effect of agn feedback ( @xcite , @xcite ) . for clusters and groups the luminosity and temperature of the gas
are also related ( e.g. @xcite , @xcite , @xcite , @xcite , @xcite ) , but this relation seems to break down at the scale of galactic halos ( @xcite , @xcite , @xcite ) . some of the scatter in these relations may be correlated with the degree of rotational support in the central galaxy as well ( @xcite , @xcite ) .
another important issue is the baryon fraction of massive halos . in galaxy clusters ,
the hot halos contain the overwhelming majority of the baryons associated with the cluster ( @xcite , @xcite , @xcite , @xcite ) , although the stellar phase begins to grow in importance as the halo mass decreases .
an open question is the contribution of the gas phase in group and galaxy halos .
if it is sufficiently massive , then the baryon fraction remains constant from galaxy clusters ( which are nearly baryon - complete ; @xcite ) down to isolated galaxies .
current censuses appear to show a declining baryon ( stars + ism + hot gas ) fraction as the halo mass decreases ( @xcite , @xcite , @xcite ) , suggesting a problem of `` missing baryons '' from galaxies , but significant questions remain .
studies by @xcite and @xcite find a self - similar relation as a function of halo mass for the hot gas pressure as probed by the sunyaev - zeldovich ( sz ) effect .
this suggests that , at least within the planck beam ( typically several galactic virial radii ) , the cosmic baryon fraction is approximately recovered in hot gas ( although see @xcite ) .
there are also claims of individual elliptical galaxies with extremely massive gaseous halos that bring the baryon fractions of the systems up to the cosmological value ( @xcite , @xcite ) , although for at least one of these claims it has been shown that the spectral modeling employed for the analysis violates x - ray surface brightness constraints and therefore significantly overestimates the total hot gas mass @xcite .
many simulations predict that the density profile of the hot halo should systematically flatten as the halo mass decreases , due to the increasing importance of galactic feedback which inflates the entropy of the hot halo ( @xcite , @xcite )
. a flatter density profile would have less pronounced x - ray emission , potentially reconciling the sz results with the x - ray constraints .
around spiral galaxies , the picture is a bit different .
starbursting galaxies produce x - ray emitting winds ( @xcite , @xcite , @xcite , @xcite ) , but extended gaseous halos have generally proven to be very difficult to detect around spiral galaxies ( @xcite , @xcite , @xcite , @xcite ) .
the exceptions are the most massive spiral galaxies .
extensive hot halos have been detected around the giant spirals ngc 1961 , ugc 12591 , ngc 6753 , and 2masx j23453268 - 0449256 ( @xcite , @xcite , @xcite , @xcite ) .
there is also a weak rosat detection of hot gas around ngc 266 @xcite , and suggestion of extended hot gas around a stack of rosat images of nearby isolated spirals @xcite .
it is unclear what explains the difficulty of detecting hot gaseous halos around massive spiral galaxies .
weak lensing studies ( e.g. @xcite , @xcite , @xcite , @xcite ) generally find that massive blue centrals and massive red centrals obey different @xmath3 relations , so that at fixed stellar mass , massive blue galaxies lie in less massive halos than red galaxies .
if the gaseous halo is responding primarily to the potential of the dark matter halo , this could plausibly explain the relative x - ray faintness of hot halos around spiral galaxies .
alternatively , if the hot halo is tightly coupled to the galaxy ( through feedback , accretion , or some combination of the two ) , then the differences in feedback and accretion between spirals and ellipticals might explain the different hot halo properties . around the massive spirals
, the hot gas generally appears to be smoothly distributed with approximate azimuthal symmetry , and has been detected to radii of 60 kpc or so from the galaxy ( although there is no indication that the halo truncates here ) .
the temperatures of the hot halos are generally close to the expected virial temperatures of these systems , suggesting the hot gas is roughly in hydrostatic equilibrium with the halo potential .
comparisons with cosmological simulations have been able to reproduce the x - ray emission around spirals ( @xcite , @xcite ) , although these comparisons have been unsuccessful in reproducing the x - ray properties of spirals and ellipticals simultaneously . a unified picture of the x - ray circumgalactic medium is still lacking .
it is particularly important to push the x - ray observations of spiral galaxies to larger radii , to probe the regime where most of the mass in the halo is predicted to lie , and to be able to constrain the properties of the dark halo and the hot halo simultaneously .
the real quantities of interest are the radial density and temperature profiles of the gas ( or equivalently the entropy and pressure profiles ) , since these encode the feedback history of the galaxy and determine the behavior of the hot halo .
the temperature profile can easily be measured from the x - ray spectrum , assuming enough photons are available to divide the field into several annuli .
so far no observations have been deep enough to permit this measurement for the hot halo of a spiral galaxy .
the density profile can be inferred from the x - ray surface brightness profile , using an assumed gas temperature and metallicity profile and a few assumptions to deproject the observations .
typically spherical symmetry and a flat abundance profile are assumed , as well as either a flat temperature profile ( leading to a `` beta profile '' for the density ; @xcite ) or a temperature profile appropriate for a cool - core galaxy cluster ( leading to a modified `` beta profile '' for the density ; @xcite ) .
the assumed metallicity has a significant effect on the final result , since the plausible range of gas metallicities spans more than an order of magnitude ( roughly 1/10 solar to slightly super - solar ) . for metallicities in this range ,
the conversion from soft - band surface brightness to density depends approximately on the square root of the gas metallicity . in terms of the total gas mass ,
the extrapolation to large radii is even more significant since the hot halo is typically only detected out to about a tenth of the virial radius ; most of the inferred gas mass lies at larger radii where we have few constraints on the density profile .
the farther out the surface brightness profile can be measured , the more reliable the extrapolation becomes , leading to better estimates of the total hot halo mass .
in order to improve the constraints on the density profile , and to make a first measurement of a temperature profile , we re - observed the giant spiral galaxy ngc 1961 with xmm - newton for an additional 215 ks ( adding to the 74 ks of observations of this galaxy already taken with xmm - newton and presented in @xcite ) . in this paper
we report our analysis of these observations of ngc 1961 . in section 2
we discuss the data reduction . in section 3
we present a spatial analysis of this galaxy , and in section 4 we present a spectral analysis . in section 5
we combine these analyses and measure the pressure and entropy profiles of the hot halo . in section 6 we discuss our results in the context of the `` missing baryons '' problem and in comparison to isolated elliptical galaxies .
ngc 1961 is an extremely massive late - type spiral galaxy .
based on its recessional velocity of 3934 km / s listed in the nasa extragalactic database , and an assumed planck cosmology @xcite with @xmath4 = 67.8 km / s / mpc , we estimate the distance of ngc 1961 to be 58.0 mpc , so that one arcminute corresponds to 16.6 kpc .
the k - band luminosity of this galaxy is then @xmath5 , corresponding to a stellar mass of @xmath6 ( assuming a m / l ratio of 0.6 , which is the rough expectation based on its k - band luminosity ; @xcite ) . for the b - v@xmath7 color of this galaxy , a @xcite initial mass function ( imf )
gives the same m / l ratio , but a @xcite mass function gives a m / l ratio of 1.2 , yielding an even larger stellar mass for this galaxy .
however , this latter m / l ratio is disfavored by @xcite , who use both population synthesis and imf - independent constraints to determine a nearly universal m / l ratio of 0.57 in the k - band , close to our value of 0.6 .
the inclination - corrected hi rotation velocity is 340 km / s ( haan et al . 2008 ) at a projected distance of 43 kpc ( although it reaches higher values
up to about 450 km / s at smaller radii ) .
the virial radius of this galaxy is approximately 490 kpc , as we discuss in section 5.1 .
we use the coordinates from the nasa extragalactic database for the center of the galaxy .
we obtained nine observations of ngc 1961 with xmm - newton , ranging from 22 - 27 ks in length each .
the aimpoints of each observation were varied such that they formed a @xmath8 grid , with a typical separation between aimpoints of about an arcminute .
this layout makes the data reduction and spectral fitting more cumbersome , but it offers the advantages of flattening out the exposure map around the galaxy and reducing the effects of vignetting .
it also allows us to identify and separate various non - astrophysical background components ( soft protons , instrumental background , solar wind charge exchange ) from the astrophysical backgrounds like the galactic halo and the cosmic x - ray background .
we also re - analyzed the two previous xmm - newton observations of this galaxy which were originally discussed in @xcite ( hereafter b13 ) .
the list of all 11 observations is presented in table 1 .
.xmm - newton observations of ngc 1961 [ cols="^,^,^,^,^ " , ] + the inner three regions ( which overlap with the disk of the galaxy ) show significant improvement with a 2-t model , while the outer two regions ( which only cover the hot halo ) do not show much improvement with a 2-t model .
this seems intuitively reasonable , and one can imagine that one component could describe the hot halo and the other could describe the hot ism of the disk . in general , the first component has a low metalllicity and contains most of the mass ; its temperature , metallicity , and density behave like the 1-t fits shown in the previous section .
this component can be associated with the hot halo of the galaxy .
note that , while the emission measures in table 3 are larger than the fiducial emission measures , this difference stems from the lower metallicites in the 2-t fit as compared to the fiducial model ( which has @xmath9 ) .
the second component is more poorly constrained , but generally has a higher metallicity and a lower emission measure .
this component can be associated with the hot interstellar medium of the galaxy .
the 2-t model therefore seems to broadly support our picture of the hot gas in and around ngc 1961 .
unfortunately there are not sufficient photons to do a more detailed analysis of the ism of this galaxy at present .
in this section we examine the surface brightness profile of the x - ray emission around ngc 1961 . we analyze the soft - band and hard - band images generated in section 2 , as well as model images of the various background components .
the esas ` mos_back ` and ` pn_back ` routines generate model particle background images automatically , and we use the fit parameters from our fit to the background at large radii as inputs to the the esas ` soft_proton ` and ` swcx ` routines in order to generate models of the soft proton background and the swcx backgrounds for each observation and each detector ( these two backgrounds are only generated for the soft band , however ) .
for each type of image , we use the ` merge_comp_xmm ` routine to combine all 30 images together into a single merged image .
we also use this routine to combine the 30 exposure maps together to generate a merged exposure map .
we present the merged , exposure - corrected , soft - band image in figure 5 , which was created with the ` merge_comp_xmm ` routine and has the qpb , soft proton , and swcx backgrounds all subtracted .
note that * this routine applies no weighting between the mos1 , mos2 , and pn detectors , but instead adds the images ( in counts ) directly * ; this has the effect of weighting the pn image more than the mos images due to the larger effective area of the pn detector .
we divide each image by the merged exposure map and construct surface brightness profiles in figure 6 .
the net soft - band signal flattens to a constant value at large radii ; we use this mean value as the value for the x - ray background in this field over the soft band . the x - ray background ( the combination of the cxb , galactic halo , and local bubble ) is the largest background component , followed by the particle background , the soft proton background , and finally the swcx background . the sky background estimated from figure 6 is consistent with the 3/4 kev background for the same region measured from rosat as well . note that the particle background and the proton background appear to turn upwards at large radii ; this is because these backgrounds are not focused by the telescope s mirrors and therefore do not suffer the same vignetting as the x - ray backgrounds .
dividing by the exposure map therefore overestimates these backgrounds at large radii .
this should not be a major issue for our analysis , since it does not begin to matter until radii of 100 kpc or more , which is outside the range within which we can measure the hot halo . in these profiles
we also show , but do not subtract , the estimated contribution of x - ray binaries to the soft band .
we estimate this contribution in two different ways .
one method ( the cyan line ) uses the hard - band x - ray image and model backgrounds .
we determine the surface brightness profile of the hard - band emission and multiply this profile by a factor of 1.67 ( the scaling into the soft - band summed mos1+mos2+pn counts for a @xmath10 powerlaw with galactic absorption ) to generate the estimated xrb profile in the soft band .
for the other method ( magenta line ) , we use the k - band image of this galaxy from 2mass @xcite to estimate the k - band surface brightness profile , then convert to the soft band using the scaling relation for lmxbs from @xcite ( this conversion is also described in more detail in @xcite ) .
these two methods agree reasonably well , both showing a subdominant but somewhat uncertain xrb contribution in the central region , which falls off quickly beyond about 1.5 arcmin ( 23 kpc ) and becomes insignificant at larger radii .
we also verify that our results are consistent between the spectral and spatial analyses .
this is important to check ( anderson and bregman 2014 ) , especially with the complexity of our background model . to check this
, we plot the surface brightness profile of the hot gas as determined from each technique , in figure 7 .
the black line shows the result of the spatial analysis , which is the remaining soft - band emission after subtracting the estimated cxb , quiescent particle background , soft proton background , solar wind charge exchange background , and xrb emission as estimated from the hard - band image ( the cyan line in figure 6 ) .
we restrict this plot to the mos instruments , since the pn instrument has a different effective area .
the red points are the results of the spectral analysis ( using the fiducial model ) , where we have converted the apec component describing the hot halo ( including galactic and intrinsic absorption ) into a soft - band count rate for the mos detectors .
the agreement is excellent within the @xmath1 42 kpc where the spectral fits are robust . in the outer spectral regions , where the significance of the hot gas component is less than 3@xmath11 ,
the spectral model predicts far lower surface brightness than the observed spatial profile .
however , here the hot gas emission comprises less than about 10% of the total soft - band signal , and it is not possible to say with certainty what the true emission looks like at such low surface brightnesses .
we discard these outer regions from the joint analysis below .
( dotted line ) .
black points with open circles denote negative values which have been multiplied by @xmath12 to allow plotting with a logarithmic y - axis .
red points with open circles are spectral fits where the significance of the hot gas component is below 3@xmath11 .
the results from the spectral and spatial methods largely agree within @xmath13 kpc . at larger radii , where the surface brightness of the hot gas is below about 10% of the total signal ,
we can not tell which is correct : the profile from the spatial method , the profile from the spectral method , or neither .
, title="fig:",width=302 ]
now that we have performed both a spectral analysis and a spatial analysis , and shown that they give consistent results , in this section we explore the physical properties of the hot halo .
first , we deproject the spectral results in order to estimate the hot gas density profile .
conceptually , we follow a similar procedure to that of @xcite , although these data have much lower s / n than their observations of the perseus cluster so we make a few modifications to that procedure . first , we generate a simple estimate for the emission measure of the hot halo at very large radii . to do this
, we fit a power - law to the hot gas component of the observed surface brightness profile at large radii , finding a logarithmic slope of @xmath14 for the surface brightness as a function of radius .
this corresponds to a slope of @xmath15 for the density as a function of radius , which is equivalent to @xmath16 of 0.75 in the standard @xmath16-model .
this conversion between surface brightness profile and density profile assumes that the emissivity does not vary with radius . in our 0.4 - 1.25 kev band ,
the emissivity does not change with temperature by more than 10% for temperatures between 0.3 kev and 1 kev .
if the metallicity varies , the emissivity can also change , but for this galaxy the projected metallicity profile is consistent with being flat .
we use a fiducial temperature of 0.4 kev and a fiducial metallicity of 0.2 @xmath17 to convert from surface brightness into density .
next , we convert the results of the spectral fitting into estimates of the emission measure in each region .
this is trivially derived from the normalization of the apec component of the spectrum , using our assumed distance of 58.0 mpc to ngc 1961 .
we rescale the emission measure in each region assuming the fiducial metallicity of 0.2@xmath17 . while we adopted a metallicity of
0.5@xmath17 when performing our spectral fitting , the results of the spectral fits with floating metallicity suggest that a value of 0.2@xmath17 is more reasonable for this galaxy . *
we therefore adopt a fiducial metallicity of 0.2@xmath17 for the remainder of the analysis*. it is simple to scale our results into a different metallicity , however : at the level of accuracy we can measure for this galaxy , a good approximation is that the emission measure is inversely proportional to the value of the metallicity , and the electron density is proportional to the inverse square root of the metallicity .
we treat the spectral fits to the 1 bins and the 2 bins separately , and since the regions overlap we also separate each set into two groups .
we therefore have four independently determined profiles , based on regions r0-r2-r4 , r1-r3 , r012-r345 , and r123-r456 respectively .
as noted in section 4 , we discard the r5-r8 regions since the normalization is so poorly constrained from the spectral fitting in these regions . finally ,
for each annulus , we subtract the expected emission measure from exterior annuli ( em@xmath18 ) .
we divide the remaining emission measure by the volume of the annulus , scaled by the fraction of the field of the annulus covered by the fiducial spectrum ( using the area @xmath19 listed in table 2 ) .
we also include a factor of 0.83 to convert from @xmath20 into @xmath21 assuming a standard helium abundance .
putting this all together , and using a distance of 58.0 mpc and angles in units of arcminutes , the expression for the average electron density is ( see also @xcite , who derives a more general form of this equation ) : @xmath22 these values are listed in the final column of table 2 , and displayed graphically in figure 8 .
the systematic errors seem to be larger than the 1@xmath11 statistical uncertainties , but together they are still only at about the 10% level .
the uncertain metallicity is by far the largest source of uncertainty .
we also note that in the inner three regions , deviations from hydrostatic equilibrium may be expected due to the presence of the galactic disk .
there is some evidence for this in the preference for 2-temperature fits to these spectra , but the underlying hot halo component remains dominant in these regions , and there is no evidence of the disk in the x - ray image ( figure 5 ) . in figure 8
we also plot the best - fit hot halo electron density profile for this galaxy , as measured by b13 .
they used the modified @xmath16-model profile of @xcite to parameterize the surface brightness profile , and assumed a constant metallicity of @xmath23 ( relative to the abundance tables of @xcite ) .
we also compute a `` corrected '' version of their density profile , which is multiplied by a factor of 0.64 to account for the different metallicity and the different abundance table relative to our analysis .
overall the agreement is very good between our deprojected profile and their corrected best - fit profile .
the behavior of the profile at larger radii is extremely important , however , and it is not clear whether their parameterization can be extended to larger radii . improved observations are necessary in order to understand the behavior of the hot gaseous halo within a larger fraction of the virial radius .
we can estimate an electron pressure profile for the hot gas , which is the product of the electron density and the temperature .
unfortunately , we do not have a deprojected temperature profile , and the hot gas component of the spectra is subdominant beyond a projected radius of 2 arcminutes , so we can not get robust results by subtracting scaled spectra from one another and fitting the remainders , as one can do for deprojection in the high s / n regime .
however , our projected temperature profiles do not show significant gradients , and we can produce an approximate estimate for the reprojected pressure profile by multiplying the reprojected density profile by the projected temperature profile .
we propagate the uncertainties on the temperature and the density into the total uncertainty , and neglect the additional uncertainty introduced by using a projected temperature profile instead of a deprojected temperature profile ( this should be smaller than the statistical uncertainties on the temperature , however ) .
the resulting profile is shown in figure 9 . .
, title="fig:",width=302 ] now that we have deprojected electron density and electron pressure profiles , we can estimate the total mass profile of the galaxy .
this is derived from the temperature , electron density , and the gradient of the total pressure at each radius .
we assume @xmath24 and we assume the total pressure is 1.91 times the electron pressure .
this factor is appropriate for solar metallicity gas with the @xcite abundances , but does not vary by more than 1% for other reasonable abundances .
we neglect magnetic support or other forms of non - thermal pressure .
we interpolate between the points in our pressure and density profiles in order to calculate the gradient , and extrapolate to larger and smaller radii based on the distance - weighted slope of the nearest data points ( for more details , see appendix b of @xcite ) .
the effective circular velocity corresponding to this derived mass profile is shown in figure 10 . for comparison
, we also plot a number of other constraints on the circular velocity profile of this galaxy .
@xcite measured the hi circular velocity out to 43 kpc , with independent measurements for the receding and approaching sides of the disk .
they measure an inclination angle of @xmath25 ( close to the @xmath26 listed in hyperleda ) which they use to correct their measurements .
the approaching side of the disk decreases towards zero velocity near the center , but the receding side of the disk shows a roughly flat curve towards the center . at smaller radii , co
1 - 0 measurements are also available from @xcite , which are more consistent with the behavior of the receding side of the disk when corrected using @xmath27 .
@xcite make an argument for a higher inclination angle , however , using @xmath28 as a rough value which we adopt here for comparison .
this value would reduce the inclination correction and the inferred total mass for the galaxy .
applying this inclination correction seems to introduce some tension between the cold gas measurements and our x - ray measurements , however .
we consider a third mass estimate using the k - band image to derive the stellar contribution to the potential .
this yields a lower limit to the total circular velocity .
we extract the stellar profile from the k - band image using ellipses for both @xmath27 and @xmath28 , but the results are nearly identical in both cases so we only plot the former for simplicity . using the assumed m / l ratio of 0.6 from section 1 , which is roughly the expectation based on @xcite and @xcite for this galaxy , we see that the stars contribute the majority of the mass in the central region of the galaxy , and are also in tension with the gas - dynamical measurements corrected assuming @xmath28 . if we instead consider the @xcite m / l ratio , the tension increases further .
the @xcite inclination angle gives results which seem consistent among gas - dynamical measurements , x - ray measurements , and the stars .
in the regimes where each curve is reliable , the agreement between the rotation curves is good , and points to a roughly flat rotation curve within at least 42 kpc . or @xmath28 , as indicated .
hi data are form @xcite for the approaching ( circles ) and receding ( squares ) sides of the disk .
co 1 - 0 data are from @xcite .
the red line is our estimate of the effective circular velocity based on our x - ray data , assuming hydrostatic equilibrium .
the line is shaded thick over the region where the rotation curve is constrained by data , and dotted where the curve is extrapolated .
the blue lines show the approximate contribution of the stars in ngc 1961 , based on the k - band image assuming either a chabrier ( solid line ) or salpeter ( dashed line ) imf . in general the @xmath27 model seems to be preferred by the x - ray data , and the rotation curve seems to be roughly flat within at least 42 kpc.,title="fig:",width=302 ] finally , in figure 11 we plot the enclosed stellar mass , hot gas mass , and total mass for ngc 1961 .
we use the aforementioned m / l ratio of 0.6 for the stars , so that the total stellar mass is about @xmath6 .
the rotation curves only extend to about 42 kpc , so we extrapolate the total mass profile to larger radii .
we assume that the dark matter approaches an nfw profile at larger radii @xcite , such that the virial mass of this galaxy is @xmath29 , extending to a virial radius of about 490 kpc .
this agrees well with the estimate of b13 , who estimated a virial radius of 470 kpc for this galaxy based on comparison with cosmological simulations .
still , it is a very crude estimate of the total mass , and we will therefore not use the extrapolated mass for any precise calculations . within 10 kpc , the stellar component seems to be dominant , but at larger radii the system quickly becomes dominated by dark matter . within 90 kpc , the hot halo contains less than a tenth of the mass in the stars . at this radius , the baryon fraction ( mass in stars + neutral hydrogen + hot halo ) is close to the cosmic fraction .
extrapolating to larger radii , the hot halo may become comparable to the mass of the stars , but the sum of these components is less than a third of the expected baryon content for the system .
this is an illustration of the problem of missing baryons from galaxies ( see section 6.1 ) .
uncertainties as the shaded red region .
each line is styled in boldface over the regime where the mass is measured , and dotted where the profile is extrapolated . , title="fig:",width=302 ] our measured electron pressure profile is also useful for predicting the thermal sz signal from this galaxy .
the thermal sz effect is proportional to the volume integral of the electron pressure , which we express using the compton @xmath30-parameter .
the compton @xmath30 parameter for our pressure profile , integrated over a sphere with radius 42 kpc , is @xmath31 arcmin@xmath32 .
the typical radius for comparison with other studies is @xmath33 , which for our profile is @xmath1325 kpc ( @xmath34 ) for ncg 1961 ( and the corresponding @xmath35 is @xmath36 ) .
if we extrapolate our pressure profile out to this radius , the integrated @xmath30 parameter increases to @xmath37 arcmin@xmath32 .
these values are below the sensitivity limits of planck sz catalogs for an object of this angular size ( @xcite , @xcite ) and the projected location of ngc 1961 is fairly close to the galactic disk , which is a more difficult location in the sky to isolate the sz signal . we can also compare this value to the stacking measurements of @xcite . for galaxies with log @xmath38 ,
@xcite estimate @xmath39 arcmin@xmath32 , where the tilde indicates that the value has been normalized to an angular distance of 500 mpc . using the distance of 58 mpc appropriate for ngc 1961 yields @xmath40 arcmin@xmath32 .
this is about 300 times larger than our estimated @xmath41 for this galaxy .
based on the planck selection function data , for a galaxy with the angular size of ngc 1961 , the planck map should be sensitive to such a signal at 5@xmath11 or higher with around 90% completeness , so the non - detection of this galaxy suggests it does not have the mean thermal sz signal for a galaxy of its stellar mass . on the other hand , based on the halo mass of ngc 1961 , @xcite predicts @xmath41 of just @xmath42 arcmin@xmath32 , which is undetectable in the planck map .
while this is only one galaxy , so it is difficult to draw strong conclusions , the implication is that halo mass is a better predictor of the sz signal than the stellar mass , a point to which we return in section 6.2 .
however , the predicted @xmath41 for a galaxy with the halo mass of ncg 1961 from @xcite is still an order of magnitude above our predicted @xmath41 value ( which admittedly involves an extrapolation of nearly an order of magnitude in radius beyond our outermost spectral bin ) .
more work needs to be done in order to understand this discrepancy .
we can also construct the entropy profile for the hot halo of ngc 1961 .
this is given by @xmath43 , and propagates uncertainties on both @xmath44 and @xmath45 just like the pressure profile ( figure 9 ) .
the entropy profile is shown in figure 12 . within the uncertainties ,
the entropy profile is positive in agreement with the schwarzschild stability criterion .
the central value is about 30 kev @xmath46 , which is the transition value noted by @xcite for brightest cluster galaxies ( bcgs ) .
bcgs below this value typically have bright h@xmath47 emission and star formation , and generally correspond to cool - core systems @xcite , while higher entropy values indicate that the gas is able to stratify before cooling instabilities grow ( @xcite , @xcite , @xcite , @xcite ) .
the low entropy in the core of ngc 1961 therefore leads one to expect similar behavior for this galaxy .
indeed , 21 cm hi observations of this galaxy @xcite show widespread , patchy hi emission around this galaxy , extending tens of kpc to the nw .
this extended hi emission has a low velocity dispersion ( of order 20 km / s ) and appears to be co - rotating with , and accreting onto , the disk ( see figure 6 of @xcite ) .
this extended hi appears to contain about 10% of the total neutral hydrogen content of the system , or about @xmath48 of neutral gas . in order to check
if this neutral gas can be explained by condensation from the hot halo , we compute the cooling time and the dynamical time as a function of radius . the cooling time is given by : @xmath49 where @xmath50 is the total particle density , and we assume @xmath51 as in section 5.1 .
the dynamical time is given by @xmath52 at every radius where we have data , the cooling time is about two orders of magnitude higher than the dynamical time , which is an order of magnitude away from the threshold value predicted from simulations for cooling instabilities to become important . at 40 kpc , for example , the cooling time is 13 gyr and the dynamical time is 160 myr .
therefore it seems unlikely that condensation from the hot halo is the source of most of the neutral hydrogen seen around ngc 1961 .
another possibility , proposed by @xcite is that the hi is the result of ram pressure stripping .
ngc 1961 lies several hundred kpc to the nw of a poor group of galaxies , and the hi appears extended in the direction away from the center of this group , so it appeared plausible that ngc 1961 could be falling into this group and losing its ism through interactions with a hot intragroup medium .
however , the x - ray halo of ngc 1961 shows no signs of such interactions , and there is also no x - ray emission associated with a larger intragroup medium in figure 5 .
a rosat analysis by @xcite also found no sign of an intragroup medium emitting in the x - ray band .
we think the extended hi feature is more likely to be a sign of accretion from the intergalactic medium .
this would explain its filamentary appearance , and it seems logical that an unusually massive spiral galaxy would require an unusually large supply of intergalactic gas .
moreover , the velocity field of the hi stream appears identical to the velocity field of the hi disk ( see @xcite figure 6 ) , which was also pointed out by @xcite .
alignment of the angular momentum of the accreting gas and the galactic disk is the natural expectation for accretion from the igm @xcite .
such alignment is less likely for a minor merger , so we think accretion is a better explanation than a merger for this galaxy , but a merger or accretion of a satellite galaxy is also a possibility .
it is interesting that , although the entropy profile looks like a classic cool - core system , the temperature of the hot gas actually increases in the center by factor of 50% . while cool cores are typical in galaxy groups and clusters , negative gradients are common for galaxies ( @xcite , @xcite ) , although x - ray faint ellipticals can show flat or positive gradients as well @xcite .
ngc 1961 is an isolated galaxy with the mass of a galaxy group , but it seems that the halo mass is not the determinant of the shape of the temperature profile for this system .
we speculate that feedback from the galaxy itself ( either stellar feedback or a previous episode of agn feedback ) may be responsible for heating the gas within the central few tens of kpc in this system .
in this section we return to one of the questions discussed in the introduction : where are the missing baryons from ngc 1961 ? the deep observations in this study
allow us to constrain the density profile of the hot halo out to about 42 kpc , which is about 9% of the virial radius , and it is notable that we see no evidence for any `` flattening '' in the density profile over this range .
it is still difficult to rule out a putative flat component at larger radii , however .
as figure 7 shows , the spectral and spatial techniques begin to disagree at radii larger than about 50 kpc , where the hot gas surface brightness is below about 10% of the total soft - band surface brightness .
we are unlikely to improve upon this limit ( see , e.g. @xcite , @xcite ) with the current generation of instrumentation . in figure 7
we also show the predicted surface brightness from a uniform density cloud of gas with kt = 0.5 kev , @xmath53 , and @xmath54 extending out to the virial radius .
this is the flattest possible density profile that contains the entire baryonic content of the system .
unfortunately this limiting case falls below our surface brightness sensitivity , and so we can not confirm or rule out the possibility of some sort of extremely flat halo becoming important at larger radii .
it is also important to emphasize the uncertainties in this discussion .
we can detect the hot halo to nearly 20% of the virial radius from our surface brightness analysis , and measure the density to about 10% of the virial radius with good accuracy , but the vast majority of the virial volume is unconstrained and depends on power - law extrapolations .
moreover , the total mass of ngc 1961 is also not well - constrained , since it also depends on similar extrapolations to the observed rotation curves . finally , since we find the stars have a similar mass as the hot halo for this galaxy , the m / l ratio of the stars is also somewhat important .
our estimates of each of these quantities are unlikely to be wrong by a factor of 2 or more , so the problem of missing baryons remains for ngc 1961 , but the exact baryon fraction should be considered extremely uncertain
. our measurements of the pressure in the hot halo can also be useful for understanding the baryon budget of ngc 1961 .
observations with the cos instrument on hst show ultraviolet absorption around l * galaxies from many different elements , generally with a high covering fraction ( @xcite , @xcite , @xcite ) .
photoionization modeling suggests that this absorption comes from ionized clouds , with characteristic temperatures around @xmath55 k , densities of order @xmath56 @xmath57 , and a volume filling fraction around 10% @xcite .
such clouds can not exist in pressure equilibrium with the hot halo around ngc 1961 , but at larger radii ( several hundred kpc ) if the hot halo pressure continues to decline these clouds would be in equilibrium .
if they have densities at the high end of their estimates and a 10% volume filling fraction beyond 150 kpc , such clouds could contain essentially all of the missing baryons from the galaxy .
however , such clouds are unlikely to condense from the hot halo , given the high ratio of the cooling time to the dynamical time .
if they originate from the galaxy , they would therefore have to survive the passage through the inner 100 kpc or more of hot gas .
we measure the x - ray luminosity of the hot halo for this galaxy from the surface brightness profile ( figure 6 ) .
we convert from 0.4 - 1.25 kev counts into 0.5 - 2.0 kev flux assuming an apec model with a fiducial temperature of 0.5 kev and a metallicity of 0.2 @xmath17 , although for reasonable values of @xmath58 and @xmath44 the conversion factor is constant to within 5% .
the hot gas luminosity within 50 projected kpc is @xmath59 erg s@xmath60 .
integrating out further yields a total hot gas luminosity of @xmath61 erg s@xmath60 .
this is slightly higher , but given the much larger uncertainties the difference is not very significant .
uncertainties from @xcite .
the red star is ngc 1961 , using measurements from this work .
we have converted the peak circular velocity from the hi profile into an effective velocity dispersion by dividing by @xmath62 .
, title="fig:",width=302 ] in figure 13 we compare the 0.5 - 2.0 kev luminosity of ngc 1961 ( within 50 kpc ) to various scaling relations for early - type galaxies .
the left panel shows this galaxy on a plot of @xmath63 versus @xmath64 , comparing against isolated galaxies from @xcite and from @xcite .
ngc 1961 lies below their best - fit relation by about an order of magnitude .
the scatter in these relations is also nearly an order of magnitude , so we caution against drawing strong conclusions from a single object , but this is consistent with the general observation that the x - ray halos of spiral galaxies appear less luminous than elliptical galaxies .
ngc 1961 is much more consistent with relations between @xmath63 and the total matter content . in the right panel of figure 13
we show the comparison between @xmath63 and the velocity dispersion of early - type galaxies from @xcite .
since ngc 1961 is rotationally supported , we place this galaxy in the figure by dividing the peak inclination - corrected circular velocity from the hi gas ( figure 10 ) by @xmath62 . in this case
, ngc 1961 falls squarely within the relation for early - type galaxies .
we can show this in more detail with a detailed comparison between the properties of ngc 1961 and comparable elliptical galaxies ( figure 14 ) . for this comparison
we consider the galaxies ngc 57 , ngc 4455 , and ngc 7796 , studied by @xcite and @xcite .
they examine ic 1531 as well , but this galaxy has an agn with a radio jet , and they can not constrain the hot halo properties very precisely .
these galaxies were selected as massive ellipticals which are not in any known group or cluster and which have no neighbors within 2 b - magnitudes of their own magnitude @xcite .
@xcite note that the well - known elliptical ngc 720 falls near the border of their isolation criteria , but seems to have a higher density of faint galaxies around it than the three galaxies in their sample .
as far as we can tell , ngc 1961 fulfills their isolation criteria as well , although it is less surprising for spiral galaxies to be isolated as compared to ellipticals .
error regions , from osullivan et al .
( 2004 ) and osullivan et al .
values for ngc 1961 are shown in black for comparison . , title="fig:",width=321 ] within the uncertainties , the mass profiles and the entropy profiles for these four galaxies are all roughly consistent with one another .
ngc 7796 has a slightly lower inferred total mass , so osulivan et al .
( 2007 ) speculate that the hot gas around this galaxy might not be in hydrostatic equilibrium .
there is a wider range of temperatures and abundances between these galaxies ; ngc 1961 lies in the middle of the range of temperatures but has a lower abundance than the elliptical galaxies .
an interpretation of the discrepancy with the @xmath65 relation and the agreement with the @xmath66 relation is that the halo mass governs the x - ray properties of the system , instead of the stellar mass .
as discussed in the introduction , massive blue galaxies generally seem to lie in less massive halos than their red counterparts . as an example of this , from
@xcite we find that the effective halo mass of an average galaxy with the stellar mass of ngc 1961 is around @xmath67 - much higher than the halo mass we measure for ngc 1961 .
the lower halo mass offers a natural explanation for why ngc 1961 appears to scatter so low on the @xmath65 relation when compared to elliptical galaxies in much more massive halos .
in this paper we have analyzed the deepest x - ray observation to date of a hot gaseous halo around a spiral galaxy .
ngc 1961 has already been studied with both chandra and xmm - newton , but our deep observations allow us to perform a much more precise and robust analysis . using a spatial analysis similar to @xcite and @xcite , we detect extended soft x - ray emission to very large radii ( at least 80 kpc ) .
no breaks or bright features are evident in the surface brightness profile or the soft x - ray image of the galaxy .
we also perform a spectral analysis on the xmm - newton data .
this allows us to measure the physical properties of the hot gas out to 42 kpc .
we explore eight different sets of spectral models , and the scatter between results from these different models allows us to estimate the systematic uncertainty from the choice of models . in general this is comparable to the statistical uncertainties .
the spectral and spatial methods give fully consistent results as well , further underscoring the robustness of our results .
the hot halo has a negative temperature gradient ( declining with radius ) , as is typical for elliptical galaxies of similar mass .
the metallcity of the halo is consistent with being flat at a value around 0.2@xmath17 .
such a low metallicity is not typical for elliptical galaxies of similar mass , which usually have abundances around solar , although x - ray fainter ellipticals can have sub - solar abundances as well . the higher abundances for ellipticals are thought to stem from the mixing of the hot interstellar medium and the hot halo , whereas ngc 1961 can lock up the majority of its metals in the cooler phases of its ism . in support of this picture , we find a statistical preference for a two - temperature medium within the disk of this galaxy , with the dominant component having the high temperature and low abundance characteristic of the hot gaseous halo and a less massive component having a lower temperature and a higher abundance typical of the hot interstellar medium .
we also measure deprojected density profiles for the hot halo out to 42 kpc .
the results are consistent with the parametric fit obtained by b13 , when rescaled to our fiducial metallicity of 0.2@xmath17 , as well as with the much more uncertain profile obtained by @xcite .
no evidence of a break or a flattening in the profile is seen .
extrapolating the profile to the virial radius , we estimate a hot halo mass which is comparable to the stellar mass of the galaxy . from the density profile and the temperature profile ,
we compute a pressure profile for the hot halo and then we infer a total mass profile under the assumption of hydrostatic equilibrium for the hot gas . this mass profile is roughly consistent with measurements from the circular velocity profile observed with 21 cm emission .
based on the extrapolated hot gas mass and total mass , the problem of missing baryons from ngc 1961 seems to persist , with an estimated baryon fraction ( hot gas + stars + neutral hydrogen ) around 30% of the cosmic fraction .
it is conceivable that photoionized gas at intermediate temperatures could also exist around this galaxy , as observations by the cos instrument on hst suggest for many smaller galaxies .
this could supply the missing baryons , but there are a few complications .
first , such clouds would be far out of pressure equilibrium with the hot halo out to several hundred kpc .
second , the hot halo appears to be stable against global cooling instabilities over the range we can measure , making it difficult to condense such clouds out of the hot halo .
these are not insurmountable difficulties , but they do complicate the modeling of the circumgalactic medium . finally , we show that the x - ray luminosity from ngc 1961 is about an order of magnitude fainter than one would expect from the @xmath63-@xmath64 relation for elliptical galaxies . on the other hand , this galaxy seems to have exactly the right @xmath63 for its dark matter mass , suggesting the dark matter halo governs the hot halo more fundamentally than the galaxy at the center .
the non - detection of this galaxy in the planck sz map also supports this conclusion , as an elliptical galaxy with the same stellar mass and distance is very likely to have been detected .
a number of open issues remain .
it is critically important to push measurements out to larger radii , which will require deeper x - ray observations , possibly combined with an sz and/or absorption - line analysis ( the latter two are more sensitive at larger radii than x - ray emission , which depends on the square of the gas density ) .
it is also important to increase the sample size of hot gaseous halos , in order to understand better how the properties of the gaseous halo connect with the properties of the galaxy and of the dark matter halo .
the authors would like to acknowledge helpful discussions with mark burke , massimo gaspari , rishi khatri , alex kolodzig , and stefano mineo related to this work , and to thank the referee for a thoughtful report that improved the paper . based on observations obtained with xmm - newton , an esa science mission with instruments and contributions directly funded by esa member states and nasa .
this research has made use of the nasa / ipac extragalactic database ( ned ) which is operated by the jet propulsion laboratory , california institute of technology , under contract with the national aeronautics and space administration .
this research has made use of nasa s astrophysics data system .
this research has made use of data and/or software provided by the high energy astrophysics science archive research center ( heasarc ) , which is a service of the astrophysics science division at nasa / gsfc and the high energy astrophysics division of the smithsonian astrophysical observatory . | we examine 11 xmm - newton observations of the giant spiral galaxy ngc 1961 , allowing us to study the hot gaseous halo of a spiral galaxy in unprecedented detail .
we perform a spatial and a spectral analysis ; with the former , the hot halo is detected to at least 80 kpc and with the latter its properties can be measured in detail up to 42 kpc .
we find evidence for a negative gradient in the temperature profile as is common for elliptical galaxies .
we measure a rough metallicity profile , which is consistent with being flat at @xmath0 . converting to this metallicity ,
the deprojected density profile is consistent with previous parametric fits , with no evidence for a break within 42 kpc ( @xmath10.1r@xmath2 ) . extrapolating to the virial radius
, we infer a hot halo mass comparable to the stellar mass of the galaxy , and a baryon fraction from the stars and hot gas of around 30% .
the cooling time of the hot gas is orders of magnitude longer than the dynamical time , making the hot halo stable against cooling instabilities , and we argue that an extended stream of neutral hydrogen seen to the nw of this galaxy is instead likely due to accretion from the intergalactic medium .
the low metallicity of the hot halo suggests it too was likely accreted .
we compare the hot halo of ngc 1961 to hot halos around isolated elliptical galaxies , and show that the total mass determines the hot halo properties better than the stellar mass .
galaxies : haloes , galaxies : spiral , galaxies : individual : ngc 1961 , x - rays : galaxies |
the chemical elements with atomic number higher than about 30 are very rare in nature . in the solar system
these trans - iron elements represent only roughly 1 atom in every 10 million . the synthesis of about 99% of the nuclei in this mass region can be explained by neutron capture reactions in the astrophysical s- and r - processes under various astrophysical conditions @xcite .
there are , however , more than 30 isotopes on the proton rich side of the nuclear chart which can be synthesized by none of these two processes .
there are still many open questions regarding the synthesis of these so called p - nuclei .
their production mechanism is the astrophysical p - process which may involve several different sub - processes @xcite .
the most important mechanism is thought to be the @xmath4-process which proceeds by @xmath4-induced reactions starting from pre - existing s- or r - process seed nuclei .
consecutive ( @xmath4,n ) reactions drive the material towards the neutron - deficient region , though charged - particle emitting ( @xmath5 ) and ( @xmath4,p ) reactions also contribute to the process . figure [ fig : pprocflow ] schematically shows the path of the @xmath4-process in the case of the tin isotopic chain .
modelling of the p - process in the whole relevant mass region involves a reaction network of thousands of reactions .
the aim of a p - process model calculation is to reproduce the p - isotope abundances observed in the solar system , however , even the state - of - the - art models fail to give a satisfactory reproduction of the measured abundances ( see e.g. @xcite ) .
the models use various input parameters . on one hand the astrophysical conditions where the process takes place
must be known .
this involves the knowledge of the stellar composition ( seed abundances ) , temperature , time scale , etc .
the presently favoured site for the @xmath4-process is the o / ne layers of massive stars in explosive or pre - explosive phase where temperatures of typically 3gk are reached for a period of the order of one second @xcite . on the other hand ,
the models require the knowledge of different nuclear physics parameters .
the masses and decay half - lives of the ( mainly proton - rich ) isotopes involved in the reaction network are relatively well known .
the reaction rates of the dominant @xmath4-induced reactions are , however , much more uncertain . since the networks include thousands of reactions on mainly unstable nuclei , the calculations have to rely on theoretical reaction rate calculations obtained typically from hauser - feshbach statistical models .
the calculated p - isotope abundances are very sensitive to the variation of the reaction rates ( see e.g. @xcite ) , therefore the experimental study of the reactions relevant for a @xmath4-process network is important to test the theoretical reaction rates .
as discussed above , @xmath4-induced reactions play a key role in the synthesis of p - nuclei .
the measurement of @xmath4-induced reaction cross sections is technically challenging , especially in the case of charged particle emitting ( @xmath5 ) and ( @xmath4,p ) reactions .
recently , there has been a significant development in this field ( see e.g. @xcite and references therein ) .
however , due to the huge difference between the reaction rate measured in the laboratory and the actual rate in a stellar plasma ( stellar enhancement effect , for details see @xcite ) , the experimental @xmath4-induced reaction cross section data can not be used directly in a nucleosynthesis model .
the experimental study of radiative capture reactions provides an alternative way to investigate the reactions relevant for the @xmath4-process , since the rate of the @xmath4-induced reactions can be inferred from the rate of the inverse capture reaction using the detailed balance theorem .
moreover , the stellar enhancement effect is much less pronounced in the case of capture reactions .
while the neutron capture cross sections are rather well known experimentally , at least along the line of stability @xcite , there are only very few ( @xmath6 ) and compared to the size of the reaction network a limited number of ( p,@xmath4 ) cross sections measured in the relevant mass and energy range leaving the statistical model calculations for these reactions largely untested .
c|c ( p,@xmath4 ) & ( @xmath6 ) + + @xmath7se , @xmath8sr , @xmath9y , @xmath10zr , @xmath11nb , @xmath12mo,&@xmath13ge,@xmath14ru,@xmath15cd,@xmath16sn , + @xmath17ru,@xmath18pd,@xmath19cd,@xmath20sn,@xmath21te&@xmath22 in , @xmath23sm , @xmath24au + table [ tab : measuredisotopes ] lists those isotopes with mass number higher than 70 for which experimental proton or alpha capture reactions cross sections are available .
it is apparent that in the case of ( @xmath6 ) reactions the available experimental database is very limited especially in the heavier mass region where the p - process models show the highest sensitivity to the ( @xmath6 ) reaction rates .
therefore it is highly needed to extend the experimental database of ( @xmath6 ) cross sections to the heavier mass region .
the conventional method for capture cross section measurements is the in - beam @xmath4-spectroscopy . in the mass and energy range relevant for the @xmath4-process , however ,
severe problems are encountered in the application of this technique . as it is usual in nuclear astrophysics , low cross sections from millibarn down to the nanobarn range
must be measured and the beam - induced background produced on low z target impurities may hinder the cross section measurement . moreover , the compound nucleus created in a capture process is typically excited to an energy range where the nuclear energy level density is very high .
this results in a complicated decay scheme involving many transitions .
all these transitions must be measured with their complete angular distributions which requires huge experimental effort ( see e.g. @xcite ) . with the application of a 4@xmath25 summing crystal
some of these problems can be avoided @xcite , but none of the in - beam @xmath4-spectroscopy methods succeeded so far to reach the heavier ( a @xmath26 110 ) mass region of the p - nuclei .
the activation method has been proved to be a very fruitful technique for the determination of cross sections relevant for the @xmath4-process .
most of the isotopes listed in table [ tab : measuredisotopes ] have been studied with activation .
this method is applicable only if the reaction product of the capture leads to a radioactive isotope with convenient half - life and strong enough decay signature ( like a high intensity @xmath4-radiation ) .
most of the problems encountered in in - beam @xmath4-spectroscopy can be avoided in an activation experiment . since the off - line measurement of the induced activity leads to the determination of the total number of capture reactions ,
there is no need to consider individual @xmath4-transitions , angular distributions , etc .
the activation technique is also less sensitive to target impurities and in certain cases more isotopes of a given element can be studied simultaneously . in the present work the cross section of the @xmath0eu(@xmath1 tb and @xmath0eu(@xmath2,n)@xmath3 tb reactions
have been measured with the activation method .
( the study of ( @xmath2,n ) reactions although not directly relevant to the p - process provides additional information to the investigation of different optical potentials , see below . ) in the next section some details of the measurement are given emphasizing the special requirements of an activation experiment .
as can be seen in table [ tab : measuredisotopes ] , the only isotope for which @xmath2-capture cross section measurement has been carried out near the a=150 mass region is @xmath23sm @xcite . for this isotope it was found , that the statistical model calculations using different @xmath2-nucleus optical potentials give extremely different cross section results .
one can choose the best optical model parametrization for a given reaction only if experimental data are available for comparison .
therefore , it is important to collect more experimental data in this mass region to have a larger experimental database to check the performance of different global optical potentials .
our next step towards this goal is the study of @xmath0eu .
alpha capture on @xmath0eu ( q=-1.98mev ) leads to @xmath27 tb which decays to @xmath27gd by electron capture with a half - life of 5.32d .
the decay is followed by a few relatively strong @xmath4-radiations which can be used for the cross section measurement .
moreover , the ( @xmath2,n ) reaction on @xmath0eu ( q=-10.14mev ) also produces a radioactive residual nucleus , @xmath3 tb , so the @xmath0eu(@xmath2,n ) cross section can also be measured by activation .
the decay scheme is , however , more complicated . besides its ground state decaying to @xmath3gd with 21.5h half - life , it has two long - lived isomeric states with half - lives of 9.4h and 22.7h decaying by @xmath28 , electron capture and/or internal transition .
fortunately all these different decays are followed by many strong @xmath4-radiations , a few of which can be attributed exclusively to the decay of a given state .
thus , partial cross sections to the ground and the two isomeric states can be determined separately and their sum gives the total ( @xmath2,n ) cross section .
figure [ fig : reaction ] shows the relevant part of the chart of nuclides with the studied reactions and the decay of the reaction product .
it should be noted that the precision of the decay parameters ( half - life , @xmath4-intensity ) directly influences the precision of the half - life determination . in the case of the m1 isomer in @xmath3 tb ,
the half - life has a relatively large uncertainty ( [email protected] ) and ambiguous values can be found in the literature .
therefore , we have performed a high precision experiment to determine this half - life value .
we have obtained t@[email protected] @xcite . with this measurement
we reduced the uncertainty of this half - life by one order of magnitude and avoided one possible source of systematic uncertainty in the @xmath0eu(@xmath2,n)@xmath3 tb cross section determination .
targets have been prepared by evaporating eu@xmath31o@xmath32 enriched to 99.2% in @xmath0eu onto thin al foil .
although the natural abundance of @xmath0eu is relatively high ( 47.8% ) , enriched target material was necessary to avoid the high background coming from the decay of @xmath33 tb , the ( @xmath2,n ) reaction product of the heavier eu isotope .
@xmath33 tb has a half - life of 5.4d , similar to that of @xmath27 tb and the @xmath34eu(@xmath2,n ) cross section is much higher than that of @xmath0eu(@xmath6 ) .
test experiments with natural isotopic composition targets showed that in the presence of the @xmath33 tb decay , the weak @xmath4-radiation from @xmath27 tb is not measurable .
the number of target atoms enters directly to the calculation of the cross section . using a thin al foil as target backing , it was possible to obtain the number of target atoms by measuring the weight of the al foil before and after the evaporation .
it is also important that the target remains stable under @xmath2-bombardment .
the target stability can be monitored by detecting the backscattered @xmath2-particles from the target continuously during the irradiation . for the detection of the backscattered particles , low z backing material
is preferred in order to have a clear separation of the scattering events from eu and from the backing .
thus , al was an ideal choice .
the irradiations have been carried out at the cyclotron accelerator of atomki .
the typical beam current was 11.5p@xmath35a and based on the backscattering spectra , no target deterioration was observed with this beam intensity .
the energy range between 11.5 and 17.5mev was covered with about 0.5mev steps .
owing to the relatively short half - life of the reaction products , the targets could be used more than once at different energies . at some energies
the irradiations were carried out with two different targets and consistent results were obtained .
the length of the irradiations varied between 5 and 24 hours .
the induced @xmath4-activity has been measured with a calibrated 40% relative efficiency hpge detector equipped with lead shielding .
the countings were started typically one hour after the end of the irradiation and lasted for at least 12 hours . in this period the @xmath4-spectra
were dominated by the peaks from the @xmath0eu(@xmath2,n)@xmath3 tb and the @xmath4-lines from the @xmath0eu(@xmath1 tb reaction were buried under the compton background .
therefore , a repeated @xmath4-counting was carried out about one week after the irradiation . at this time
the shorter - lived @xmath3 tb isotope was essentially absent and @xmath27 tb @xmath4-peaks could be observed on a strongly reduced background . as an example , figure [ fig : spectrum ] shows the measured @xmath4-spectrum directly after the irradiation where many peaks from the decay of the three states of @xmath3 tb can be seen .
cross sections of the @xmath0eu(@xmath1 tb and @xmath0eu(@xmath2,n)@xmath3 tb reactions have been measured in the center of mass energy range between 11.2 and 17.1mev . at the lowest two energy points the yield from the ( @xmath36 reaction product was too low and therefore only the ( @xmath2,n ) cross section could be obtained .
the gamow window at t=3gk temperature for this reaction is roughly between 8 and 12 mev , so the measurements were carried out just above the astrophysically relevant energy range . in the measured energy range the cross section values are between about 5microbarns and 100millibarns .
the final analysis of the data is still in progress .
figure [ fig : results ] shows the preliminary results of both measured reaction in the form of the astrophysical s - factor .
also shown are the statistical model calculations performed with the non - smoker@xmath37 code version 5.8w @xcite using two different global @xmath2-nucleus optical potentials .
one is the standard potential of the non - smoker@xmath37 code from mcfadden and satchler @xcite and the other from frhlich and rauscher @xcite .
the calculations reproduce well the energy dependence of the measured cross sections , while it seems that the absolute magnitude is better reproduced by the frhlich and rauscher potential .
the final analysis and the interpretation of the data are in progress .
this work was supported by erc grant 203175 , gvop-3.2.1.-2004 - 04 - 0402/3.0 . , otka ( k68801 , t49245 ) , tubitak - grant-108t508 and kocaeli university bap - grant-2007/37 .
99 busso m , gallino r and wasserburg g 1999 _ j ann .
astrophys . _ * 37 * 239 arnould m , goriely s and takahashi k 2007 _ phys .
rep . _ * 450 * 97 arnould m and goriely s 2003 _ phys .
rep . _ * 384 * 1 gyrky gy _
et al _ 2006 _ phys .
c _ * 74 * 025805 rapp w , grres j , wiescher m , schatz h and kppeler f 2006 _ astrophys . j. _ * 653 * 474 rauscher t , heger a , hoffman r d and woosley s e 2002 _ astrophys j. _ * 576 * 323 rauscher t 2006 _ phys . rev .
c _ * 73 * 015804 hasper j _ et al _ 2009 _ phys .
c _ * 79 * 055807 mohr p , flp zs and utsunomiya h 2007 _ eur .
j. a _ * 32 * 357 bao z y , beer h , kppeler f , voss f and wisshak k 2000 _ at .
data tables _ * 76 * 70 galanopulos s _ et al _ 2003 _ phys .
c _ * 67 * 015801 hasper j _ et al _ this volume spyrou a , becker h - w , lagoyannis a , harissopulos s and rolfs c 2007 _ phys .
c _ * 76 * 015802 somorjai e _ et al _ 1998 _ astron .
astrophys . _ * 333 * 1112 gyrky gy _
et al _ 2009 _ nucl .
a _ * 828 * 1 non - smoker@xmath37 version 5.8w http://www.nucastro.org mcfadden l and satchler g r 1966 _ nucl .
phys . _ * 84 * 177 frhlich c diploma thesis 2002 university of basel , switzerland rauscher t 2003 _ nucl .
a _ * 719 * 73c ; erratum : rauscher t 2003 _ nucl .
a _ * 725 * 295 | the cross sections of @xmath0eu(@xmath1 tb and @xmath0eu(@xmath2,n)@xmath3 tb reactions have been measured with the activation method .
some aspects of the measurement are presented here to illustrate the requirements of experimental techniques needed to obtain nuclear data for the astrophysical p - process nucleosynthesis .
preliminary cross section results are also presented and compared with the predictions of statistical model calculations . |
gravitation is an example of attractive interaction that is unable to confine particles .
it is always possible that some particles acquire sufficiently energy ( via encounters or any other mechanism ) and scape from gravitational influence of the system .
evaporation is an unavoidable process that drives dynamical evolution of astrophysical systems . as discussed elsewhere
@xcite , the incidence of evaporation is a very important ingredient to explain the behavior of stellar structures observed in nature .
the effects of evaporation crucially depend on the relaxation mechanisms that are present in the microscopic dynamics , which depend on concrete conditions of a given particular system . among such mechanisms , one can mention ( i ) the particles collisions @xcite , which are present in stellar systems sufficiently dense such as globular clusters ; and ( ii ) general mechanisms like _ parametric resonance _ , which is the origin of chaotic behavior of microscopic dynamics in nonlinear many - body hamiltonian systems with bound motions in the configuration space @xcite . we propose in this work a parametric family of astrophysical models that accounts for the influence of evaporation under different relaxation regimes : the _ truncated @xmath4-exponential models_. these phenomenological models constitutes a suitable generalization of some models of tidal stellar systems available in the literature , such as wooley & dickens truncated isothermal model @xcite , as well as king s models of globular clusters @xcite . this proposal enable us to analyze the influence of the evaporation on the thermodynamic behavior and distribution profiles .
moreover , it provides a unification framework of the known isothermal and polytropic profiles considered in hydrodynamic models of stellar systems .
remarkably , these models are sufficiently amenable to allow an analytical derivation of most of their associated hydrodynamic quantities .
the paper is organized into sections as follows .
second section is devoted to introduce the parametric family of quasi - stationary distributions , the derivation of their associated hydrodynamic quantities , as well as some details about the integration of the associated nonlinear poisson equation . afterwards
, third section is devoted to discuss numerical results obtained from this model , both information concerning to thermodynamical description as well as distribution profiles .
some conclusions are drawn in forth section .
finally , mathematical properties concerning to the so - called @xmath4-_exponential function _ are discussed in [ exp.append ] .
both observational evidences and theoretical analysis suggest that distribution profiles of many stellar systems can be explained considering a quasi - stationary one - particle distribution with a mathematical form close to maxwell - boltzmann distribution @xcite : @xmath6,\ ] ] being @xmath7 the mechanical energy of a given particle with momentum @xmath8 located at the position @xmath9 with a mean field gravitational potential @xmath10 .
the incidence of evaporation introduces a deformation of the above distribution , specifically , a truncation for energies above a certain _ scape energy _ @xmath2 .
two examples of distributions that introduce this kind of deformation are the truncated isothermal model proposed by wooley and dickens @xcite : @xmath11 \right\},\ ] ] as well as michie - king models @xcite : @xmath12 \right\ } -1\right\},\ ] ] which vanish for energies @xmath13 , where @xmath14 represents a normalization constant .
the first one corresponds to a maxwell - boltzmann distribution ( [ mb.prof ] ) that is discontinuously truncated at the scape energy @xmath2 , while the second one considers a progressive vanishing at this point .
corresponding to truncated @xmath4-exponential model ( [ df.model ] ) for some values of deformation parameter @xmath4 .
we have employed prefixed values @xmath15 and @xmath16 for the scape energy and the inverse temperature , respectively.,title="fig:",width=432 ] + our interest is to propose a generalization of evaporation truncation scheme considered in the above astrophysical models .
firstly , let us express them into a unifying form as follows : @xmath17 where @xmath18 $ ] , while @xmath19 is a certain function obtained from the _ truncation of exponential function _
@xmath20 , with @xmath4 being a certain parameter .
in particular , distributions ( [ wd.prof ] ) and ( [ mk.prof ] ) are described by the functions : @xmath21 these two truncation schemes can be generalized for any nonnegative integer @xmath22 subtracting the first @xmath23 terms of power - expansion of exponential function : @xmath24 noteworthy that this ansazt contains the functions ( [ f01 ] ) as particular cases with @xmath25 and @xmath26 , which now guarantees the vanishing of its @xmath27 first derivatives at @xmath28 : @xmath29 for positive values of the argument @xmath30 , the function ( [ n.truncated ] ) can be rewritten into a more convenient integral representation as follows : @xmath31 this integral representation enables a direct extension of the above truncation scheme for any nonnegative real number @xmath4 replacing the factorial @xmath32 by the known gamma function @xmath33 : @xmath34 the resulting function : @xmath35 modifies the usual exponential function using @xmath4 as a _
deformation parameter_. hereinafter , this function will be referred to as _ @xmath4-exponential function _ , and the same one will be equivalently denoted as @xmath19 or @xmath36 .
definition ( [ integral.gamma ] ) is simply the _ riemann - liouville fractional integral _ of the ordinary exponential function @xcite . considering the fractional differentiation rule for the power - function series : @xmath37 conditions ( [ first.derivatives ] ) are now generalized as follows : @xmath38 where order of fractional differentiation @xmath39 belongs to the interval @xmath40 .
readers can find some useful properties of @xmath4-exponential function @xmath41 in [ exp.append ] .
the proposed family of quasi - stationary distributions ( [ df.model ] ) describe the influence of evaporation with two independent parameters : the scape energy @xmath2 and the deformation parameter @xmath4 .
this family of distributions is schematically shown in figure [ gamma.distribution.eps ] .
the same ones exhibit a maxwell - boltzmann profile ( [ mb.prof ] ) for energies @xmath42 that are far enough from below the scape energy @xmath2 . besides , they show a _ power - law truncation _ at @xmath2 with exponent @xmath4 .
the increasing of deformation parameter @xmath4 characterizes a larger deviation from maxwell - boltzmann profile ( [ mb.prof ] ) . for a phenomenological viewpoint ,
a larger value of the deformation parameter @xmath4 describes a stronger influence of evaporation , or equivalently , the incidence of a weaker relaxation mechanism . as shown below
, the ansatz ( [ df.model ] ) can describe distribution profiles with isothermal cores and polytropic haloes , where deformation parameter @xmath4 _ determines the polytropic structure of haloes_. the scape energy @xmath43 can be derived from the called _ tidal potential _
@xmath44 , which accounts for the gravitational influence of neighboring systems @xcite .
according to this energetic truncation , a particle that is trapped under the system gravitational influence will be confined inside a finite region of the space , where gravitational potential @xmath45 fulfills the inequality @xmath46 . for a system with
spherical symmetry , the boundary of this region is the sphere of radius @xmath47 : @xmath48 which is referred to as the _ tidal radius _ @xcite , with @xmath49 being the total mass of astrophysical system .
let us now obtain some hydrodynamic observables associated with family of distributions ( [ df.model ] ) .
introducing the dimensionless variable @xmath50 as well as the dimensionless potential @xmath51 : @xmath52,\ ] ] the particles density @xmath53 can be calculated as follows : @xmath54 being @xmath55 .
the integration can be performed by using the convolution formula ( [ convolution ] ) with parameter @xmath56 , which leads to express the particles density in terms of the @xmath4-exponential function as follows : @xmath57 = ce\left [ \phi(\mathbf{q});\gamma+\frac{3}{2}\right].\ ] ] here , @xmath58 and @xmath59 , in accordance with gamma function for half - integer values : @xmath60 other important observable is the kinetic energy density @xmath61 : @xmath62 which is also expressed in terms of the @xmath4-exponential function using the convolution formula ( [ convolution ] ) with parameter @xmath63 : @xmath64=\frac{3}{2\beta}ce\left [ \phi(\mathbf{q});\gamma+\frac{5}{2}\right].\ ] ] finally , we can introduce the magnitude @xmath65 : @xmath66 which represents the kinetic energy per particle . on the dimensionless potential @xmath67 for some values of deformation parameter @xmath4 .
one can notice the asymptotic convergence of all these dependencies towards an exponential growth for large values of @xmath67 , @xmath68 , which is related to a dominance of isothermal conditions .
we show in the inset panel these same dependencies in order to show the power - law behavior for small values of @xmath67 , @xmath69 , which is related to a dominance of polytropic conditions as a consequence of evaporation .
this type of dependence can describe distributions with isothermal cores and polytropic haloes.,width=432 ] according to the kinetic theory for ideal gases , this magnitude @xmath70 can be related with the fluid pressure @xmath71 as : @xmath72 the hydrostatic equilibrium takes place in a fluid when gravitational force and pressure gradient are balanced as follows : @xmath73 with @xmath74 and @xmath75 being its pressure and mass density , respectively .
the gravitation intensity vector @xmath76 can be expressed in terms of gravitational potential @xmath45 : @xmath77 considering the total differentiations for @xmath78 and @xmath71 : @xmath79 the hydrodynamic equilibrium condition ( [ hidro ] ) can also be rewritten as : @xmath80 where @xmath81 and @xmath82 are the particles density and the pressure expressed in terms of dimensionless potential @xmath67 .
considering differentiation identity ( [ diff ] ) , it is straightforward verified that hydrodynamic quantities ( [ densidad particulas ] ) and ( [ presiongamma ] ) derived from truncated @xmath4-exponential model satisfy the condition of hydrostatic equilibrium ( [ hidro ] ) . and
@xmath83 _ versus _ the central value of dimensionless potential @xmath84 derived from integration of nonlinear poisson problem ( [ p1]).,width=432 ] mass density @xmath85 and gravitational potential @xmath45 are related via poisson equation , @xmath86 . restricting to situations with spherical symmetry , one obtains the following nonlinear poisson problem in terms of dimensionless potential ( [ dimensionless.pot ] ) : @xmath87=-4 \pi
e\left [ \phi(\xi);\gamma+\frac{3}{2}\right].\ ] ] we have considered here the dimensionless radius : @xmath88 where @xmath89 is the physical radius , @xmath90 is an auxiliary constant , @xmath47 is the tidal radius . problem ( [ p1 ] ) can be solved in a numerical way by demanding the following conditions at the origin : @xmath91 which is integrated till the vanishing of dimensionless potential at the surface of the system with dimensionless radius @xmath92 : @xmath93 this constant allows to express the mass density @xmath75 in units of the characteristic density @xmath94 as follows : @xmath95 where @xmath83 is the dimensionless inverse temperature : @xmath96 .
panel a ) caloric curves @xmath83 _ versus _ @xmath97 ; panel b ) central particles density @xmath98 _ versus _ @xmath97 .
additionally , we have included the following notable points : ( squares ) critical point of gravothermal collapse @xmath99 ; ( circles ) critical point of isothermal collapse @xmath100 ; ( triangles ) critical point of evaporation disruption @xmath101.,width=432 ] the total energy @xmath102 can be obtained from the total kinetic @xmath103 and total potential energy @xmath104 : @xmath105 \phi(\xi)~4\pi\xi^{2}d\xi,\\ k & = & \frac{3}{2\eta^{2}\xi_{c}}\int_{0}^{\xi_{c}}e\left [ \phi(\xi);\gamma+\frac{5}{2}\right ] 4\pi\xi^{2}d\xi,\label{k}\end{aligned}\ ] ] which are expressed in units of the characteristic energy @xmath106 .
obtained from numerical integration using definition ( [ entropy.integral ] ) .
only the upper branch of microcanonical caloric curves shown in figure [ curvacalorica_1.eps ] correspond to stable quasi - stationary configurations because of they exhibit the higher entropy.,width=432 ] many studies about astrophysical systems and cosmological problems @xcite start from assuming a polytropic dependence between the pressure @xmath74 and the particles density @xmath75 : @xmath107 where @xmath108 is a certain constant and @xmath109 the so - called _
polytropic index_. the phenomenological state equation ( [ state.eq ] ) can be combined with condition of hydrostatic equilibrium ( [ hidro ] ) to derive a power - law relation between the density and the dimensionless potential @xmath67 : @xmath110 where @xmath111^{n}$ ] and the exponent @xmath112 .
the marginal case @xmath113 is fully licit .
condition of hydrostatic equilibrium now predicts an exponential dependence between @xmath75 and @xmath67 : @xmath114,\ ] ] where @xmath115 is a certain integration constant . this last behavior can be associated with a system under _ isothermal conditions _ imposing the constraint @xmath116 , where @xmath117 is the mass of constituting particles and @xmath1 is the inverse temperature .
the ansatz ( [ df.model ] ) proposed in this study allows to consider the above dependencies into a unifying fashion . according the asymptotic behavior of truncated @xmath4-exponential function ( [ asymptotic ] ) ,
the particles density ( [ den ] ) describes a power - law dependence for @xmath67 small with exponent @xmath118 , while an exponential dependence for @xmath67 large enough : @xmath119{cc}% \propto\phi^{n } , & x<<1,\\ \propto\exp\left ( \phi\right ) , & x>>1 .
\end{array } \right.\ ] ] accordingly , the deformation parameter @xmath4 can be related to the polytropic index @xmath109 of polytropic state equation ( [ state.eq ] ) as @xmath120 . according to nonlinear poisson problem ( [ p1 ] ) , the dimensionless potential @xmath67 decreases from the inner towards the outer region of the system .
this means that the truncated @xmath4-exponential models could describe distribution profiles with _
isothermal cores _ and _ polytropic haloes_. since the particles density ( [ den ] ) should vanish at the system surface , where dimensionless potential @xmath121 , one should expect that the external dimensionless radius @xmath122 should be finite , @xmath123 . however , polytropic profiles with exponent @xmath124 are infinitely extended in the space and exhibit an infinite mass , @xmath125 and @xmath126 .
such profiles can not describe any realistic situation @xcite . on the other hand , finite character of quasi - stationary distribution function @xmath127
is only possible if deformation parameter @xmath128 .
consequently , the admissible values of deformation parameter @xmath4 must be restricted to the following interval : @xmath129 accordingly , these models can only describe polytropic dependencies with exponent @xmath23 restricted to the interval @xmath130 . as show below , this restriction will be manifested in both thermodynamic quantities and distribution profiles .
: panel a ) the inverse temperatures @xmath131 and the modulus of energies @xmath132 ; panel b ) the central particles densities @xmath133 .
inset panel : fit of dependence @xmath134 near the critical value @xmath135 using the power - law ( [ form.power]).,width=384 ] nonlinear poisson problem ( [ p1 ] ) is integrated using @xmath4 and @xmath84 as independent integration parameters .
this task was accomplished using runge - kutta fourth - order method , which was implemented using fortran 90 programming .
results from this integration are shown in figure [ numericos.eps ] , in particular , dependencies @xmath122 and @xmath83 _ versus _ the central value of dimensionless potential @xmath84 for some values of deformation parameter @xmath4 . according to these results , all dependencies of the dimensionless radius @xmath122
diverge when dimensionless potential @xmath84 approaches to zero , which means that admissible values of parameter @xmath84 are nonnegative .
curiously , dependencies of dimensionless radius @xmath122 corresponding to deformation parameter @xmath136 and @xmath137 diverge at a certain value @xmath138 of dimensionless potential @xmath84 that depends on deformation parameter @xmath4 , while the corresponding dependencies of dimensionless inverse temperature @xmath83 simultaneously vanishes .
this means that value of dimensionless potential @xmath84 above the point @xmath138 are also nonphysical .
a better understanding above the physical meaning of behaviors observed in dependencies of figure [ numericos.eps ] is achieved analyzing dependencies of thermodynamic quantities .
-exponential profile with @xmath26 ( michie - king profile ) corresponding to the point of gravothermal collapse and its comparison with isothermal and polytropic profiles using _ log - log scales_. panel b ) the same dependencies but now using _ linear - log scales _ to appreciate between the polytropic fit of the halo .
accordingly , the proposed family of models can describe distribution profiles that exhibit isothermal cores and polytropic haloes.,width=384 ] dependencies of the inverse temperature @xmath83 and central particles densities @xmath139 _ versus _ the dimensionless energy @xmath97 are shown in figure [ curvacalorica_1.eps ] for different values of deformation parameter @xmath4 .
all quasi - stationary configurations obtained from these models have always negative values for the energy @xmath97 .
moreover , one can recognize the existence of three notable points : * _ critical point of gravothermal collapse _
@xmath99 : quasi - stationary configuration with minimum energy @xmath99 .
there not exist quasi - stationary configuration for energies below this point .
if a system is initially prepared with an energy below this point , it will experience an instability process that leads to sudden contraction of the system under its own gravitational field , a phenomenon commonly referred to in the literature as _ gravothermal collapse _ @xcite . *
_ critical point of isothermal collapse _
@xmath100 : quasi - stationary configuration with minimum temperature @xmath140 .
there not exist a stable configurations for temperatures @xmath141 .
if a system under _ isothermal conditions _ ( in presence of a thermostat at constant temperature ) is initially put in thermal contact with a heat reservoir with @xmath141 , this system will experience a instability process fully analogous to gravitational collapse , which is referred to as _ isothermal collapse _ @xcite .
this type of thermodynamical instability is less relevant than the gravothermal collapse because of the presence of a thermostat is actually an unrealistic consideration in most of astrophysical situations . *
_ critical point of evaporation disruption _
@xmath101 : quasi - stationary configuration with maximal energy , that is , there not exist quasi - stationary configurations for energies @xmath142 .
the existence of this superior bound is a direct consequence of the incidence of evaporation , which imposes a maximum value for the individual mechanical energies of the system constituents , @xmath143 .
if the system is initially prepared prepared with an energy above , it will experience a sudden evaporation in order to release its excess of energy @xcite .
note that the inverse temperature @xmath144 always vanishes at this point regardless the value of deformation parameter @xmath4 .
all quasi - stationary configurations are located inside the energy region @xmath145 .
moreover , there exist more of one admissible value for the dimensionless inverse temperature @xmath83 for a given energy @xmath97 near the point of gravothermal collapse @xmath99 . according to results shown in figure [ entropy_1.eps ] ,
stable quasi - stationary configurations belong to the superior branch @xmath146 , since these configurations exhibit the higher value of entropy @xmath147 for a given total energy .
energy dependence of this thermodynamic potential was evaluated from numerical integration of the expression : @xmath148 which employs as a reference the value @xmath149 corresponding to critical point of evaporation instability @xmath101 .
stable quasi - stationary configurations inside the energy range @xmath150 exhibit _ negative heat capacities _ @xmath151 . the existence of this thermodynamic anomaly is a remarkable consequence of the long - range character of gravitation , in particular , because of the short - range divergence of its interaction potential energy : @xmath152 when particles separation distance @xmath153 drops to zero .
while such configurations are unstable if the system is put into thermal contact with an environment at constant temperature ( canonical ensemble ) , they are stable if the system is put into energetic isolation ( microcanonical ensemble ) . for three notable values of the internal energy @xmath97 .
while the increasing of deformation parameter @xmath4 produces profiles with more dense cores and more diluted haloes , the increasing of the energy produces the opposite effect .
interestingly , the qualitative form of the haloes are the same for a given value of deformation parameter @xmath4 .
panels a)-c ) distribution profiles at the energy of gravothermal collapse @xmath99 for three values of deformation parameter with @xmath154 .
all them exhibit isothermal cores and polytropic haloes .
panel d ) a distribution profile at gravothermal collapse with deformation parameter @xmath155 .
note that this profile does not exhibit an isothermal core , but a divergence in the central density .
panels i)-l ) distribution profiles very near the energy of evaporation disruption @xmath101 are everywhere polytropic .
panels e)-h ) transitional profiles at the energy of isothermal collapse @xmath100 .
these profiles hardly differ from polytropic profiles i)-l ) because of they exhibit more dense cores.,width=652 ] for the sake of a better understanding about the influence of deformation parameter @xmath4 , we have calculated dependence of some thermodynamic quantities at the notable points .
such results are shown in fig.[calculofino_1.eps ] .
it is clearly evident that the increasing of the deformation parameter @xmath4 provokes a systematic decreasing of inverse temperatures @xmath156 , and the increasing of absolute values of energies @xmath132 and their associated central particles densities @xmath133 .
interestingly , the inverse temperature @xmath157 at the notable point of gravothermal collapse vanishes when @xmath158 .
this means that both the total energy @xmath99 and the temperature @xmath159 diverge at this point as a consequence of divergence of the central density @xmath160 .
precisely , the existence of this divergence also manifested as a divergence of dimensionless radius @xmath122 , which was shown in figure [ numericos.eps ] for the particular cases with deformation parameters @xmath136 and @xmath137 .
the fact that the energy of gravothermal collapse diverges @xmath161 when @xmath162 , significantly reduces the dramatic character of this phenomenon . for values admissible values of deformation parameter @xmath4 below the point @xmath135 , the system develops a gravothermal collapse at the finite energy @xmath99 , which should evolves in a _ discontinuous way _ towards a certain collapsed structure that is not describable with the present model . for values
admissible values of deformation parameter @xmath4 above the point @xmath135 , the system should release an infinite amount of energy to reach a collapsed structure with a divergent central density associated with the point of gravothermal collapse . however , this collapsed structure is now described within the present models and the transition is developed in a _ continuous way _ with the decreasing of the internal energy .
it is noteworthy that an analogous divergence is observed in thermodynamic parameters of other notable points when deformation parameter @xmath4 approaches its maximum admissible value @xmath163 .
as expected , this second divergence point is related to the nonphysical character of polytropic dependencies when @xmath124 .
a more precise estimation for the critical value @xmath135 can be obtained considering an adjustment of dependency @xmath164 near with a power - law form : @xmath165 as shown in the inset panel of fig.[calculofino_1.eps ] , the proposed form ( [ form.power ] ) exhibits a great agreement with numerical results for the following set of parameters : @xmath166 , @xmath167.@xmath168 and @xmath169 .
we show in figure [ comparacion_1.eps ] a distribution profile with deformation parameter @xmath26 at the point of gravothermal collapse , which correspons to a michie - king profile with lowest energy .
as a consequence of gravitation , the highest concentration of the particles is always located in the inner region of the system , while particles density gradually decays with the increasing of the radius @xmath89 till vanishes at the tidal radius @xmath47 . as clearly shown in this figure , the inner region , the core , can be well - fitted with an isothermal profile @xcite , while the outer one , the halo , is well - fitted with a polytropic profile @xcite .
dependence of distribution profiles on the deformation parameter @xmath4 and the internal energy @xmath97 is illustrated in figure [ profiles.eps ] , specifically , twelve profiles corresponding to three notable points and forth different values of deformation parameter @xmath4 .
particles concentration in the inner regions decreases with the increasing of the internal energy @xmath97 .
the increasing of the deformation parameter @xmath4 produces distribution profiles with more diluted haloes , and consequently , more dense cores .
not all admissible profiles derived from the present family of models can exhibit isothermal cores .
however , all these profiles exhibit polytropic haloes .
in fact , distribution profiles near the point of evaporation disruption can be regarded as everywhere polytropic with high accuracy .
finally , distribution profiles corresponding to the point of gravothermal collapse with @xmath155 are divergent at the origin .
-exponential models in the plane of integration parameters @xmath170 .
dark grey region corresponds to quasi - stationary configurations with negative heat capacities , while light grey region corresponds to positive heat capacities .
white regions are nonphysical or unstable configurations .
we have emphasized inside the dark grey region those profiles that exhibit isothermal cores and polytropic haloes .
configurations corresponding to isothermal collapse are always outside this region .
quasi - stationary configurations on the line of gravothermal collapse ( thick solid line ) exhibit a weakly dependence on the deformation parameter @xmath4 if @xmath154 .
this dependence exhibits a significant change for @xmath171 , which accounts for a divergence in the central density.,width=384 ] we have shown in figure [ phiconverge_1.eps ] the phase diagram of truncated @xmath4-exponential models in the plane of integration parameters @xmath170 of the nonlinear poisson problem ( [ p1 ] ) . for each value of deformation parameter @xmath4
, the admissible values of the central dimensionless potential @xmath84 are located inside the interval @xmath172 , where @xmath173 corresponds to the critical point of gravothermal collapse @xmath99 .
central values of dimensionless potential @xmath84 above dependence @xmath173 correspond to nonphysical on unstable configurations ( white region ) .
additionally , we have included dependence @xmath174 associated with the point of isothermal collapse @xmath100 .
configuration between dependencies @xmath175 exhibit negative heat capacities ( dark gray region ) , while those one with central dimensionless potential @xmath84 belonging to interval @xmath176 exhibit positive heat capacities ( light gray region ) .
it is remarkable that dependence @xmath173 is weakly modified by a change in the deformation parameter @xmath4 for values below the critical point @xmath135 .
however , this function experiences an sudden change above this critical value .
as expected , this behavior accounts for a sudden change in behavior of distribution profiles : the proposed models can exhibit profiles with isothermal cores for @xmath154 , while they only exhibit profiles without isothermal cores for @xmath162 . according to equation ( [ asymp.density ] ) , profiles with isothermal cores
are directly related to asymptotic dependence of particles distribution to follow an exponential - law with regard to the local value of dimensionless potential @xmath177 .
such an asymptotic behavior is better described in terms of _ deviation function _
@xmath178 with regard to the exponential function , which is introduced in [ exp.append ] . for the sake of convenience
, we have denoted dependence @xmath179 as follows : @xmath180=\epsilon,\ ] ] where the convergence error @xmath181 was fixed at the value @xmath182 .
this small value guarantees the matching of this dependence at the critical value of deformation parameter @xmath135 with central dimensionless potential @xmath84 associated with the point of gravothermal collapse , @xmath183 .
quasi - stationary configurations located inside the region @xmath184 and @xmath185 exhibit isothermal cores , that is , the inner regions of particles distribution can be fitted with an isothermal profile .
we have introduced the truncated @xmath4-exponential models to characterize the properties of astrophysical systems in a quasi - stationary evolution under the incidence evaporation .
our proposal generalizes models of tidal stellar systems available in the literature , such as michie - king models .
these models exhibit many features observed in other astrophysical models . due to truncation particles distribution in the configuration space @xmath127 ; the distribution of particles in the physical space is located inside a finite region limited by the tidal radio @xmath47 .
moreover , the total energy @xmath97 is also restricted to a finite region @xmath186 .
the lower bound @xmath99 represents the energy of gravitational collapse , while the upper one @xmath101 is the energy of evaporation disruption . caloric curves of these models exhibit an anomalous branch with negative heat capacities for energies @xmath187 , where @xmath100 is the value of energy corresponding to the point of isothermal collapse , where the system reaches a minimum temperature @xmath140 .
distribution profiles with lower energy could exhibit isothermal cores and polytropic haloes with exponent @xmath188 .
moreover , polytropic profiles are always obtained for energies near the point of evaporation disruption @xmath101 .
the admissible values of deformation parameter @xmath4 are restricted to the interval @xmath189 .
this means that this family of models describes polytropic profiles with exponent @xmath23 inside the interval @xmath130 .
a nontrivial result obtained from these models is that the existence of distribution profiles with isothermal cores is not longer possible if the deformation parameter @xmath190 .
the existence of a notable value @xmath135 for deformation parameter indicates a drastic change in the behavior of thermodynamic quantities and distribution profiles .
specifically , these cases exhibit a simultaneous divergence of the energy of gravitational collapse @xmath99 , its corresponding temperature @xmath159 and central density @xmath191 . for larger energies ,
these values of deformation parameter @xmath4 describe distribution profiles almost with a polytropic form . before ending this section ,
let us briefly refer to some open questions .
a realistic improvement of the present models is the consideration of mass spectrum for constituting particles , which enable us to study mass segregation influence on the thermodynamic behavior and distribution profiles @xcite .
a second improvement is the consideration of factors that lead to a breakdown of spherical symmetry , such as the system rotation or the tidal field of a neighboring astrophysical system @xcite .
finally , we shall perform the comparison of results obtained from these models with experimental data , in particular , particles distributions of tidal stellar systems such as globular clusters and elliptical galaxies .
these problems will be analyzed in forthcoming works .
-exponential family in a log - log graph
. one can notice the asymptotic dependencies : a power - law behavior for small @xmath30 and an exponential growing for @xmath30 large enough .
inset panel : the same dependencies in a linear - linear graph for @xmath30 near the origin , which allows to appreciated the power - law character of truncation .
panel b ) dependence of deviation function ( [ defect ] ) using a linear - linear scale .
according to this figure , the relative deviation from exponential decreases with the increasing of the independent variable @xmath30 , while it grows with the increasing of the deformation parameter @xmath4.,width=384 ] the @xmath4-exponential function @xmath36 is defined with the help of called _ gamma distribution _ as follows : @xmath192 which vanishes for @xmath193 .
this family of functions is directly related to _
lower incomplete gamma function _ :
@xmath194which only differs from expression ( [ def ] ) because of the factor @xmath195 .
moreover , the same one can also be obtained applying the _ fractional integral operator _
@xcite : @xmath196 over the exponential function . changing the integration variable @xmath197 in this last definition ,
one obtains : @xmath198 which drops to expression ( [ def ] ) when @xmath199 .
performing the integration by parts , one can obtain from definition ( [ def ] ) the recurrent relation : @xmath200 which leads to the power - expansion in equation ( [ integral.gamma ] ) .
this representation allows to obtain the following particular expressions : @xmath201 where @xmath23 is any positive integer number . in general , the @xmath4-exponential @xmath202 is a continuous and differentiable function for every @xmath203 .
this function vanishes at @xmath28 whenever @xmath3 , while it exhibits a discontinuity at @xmath28 when the deformation parameter @xmath204 .
it is easy to verify that the function @xmath205 with @xmath3 exhibits an _ exponential behavior _ for @xmath30 sufficiently large and a _ power - law _
dependence for small @xmath30 : @xmath206{cc}% x^{\gamma}/\gamma\left ( \gamma+1\right ) & \mbox { if } x<<1,\\ e^{x } & \mbox { if } x>>1 .
\end{array } \right.\ ] ] according to recurrence relation ( [ recurrence ] ) , the @xmath4-exponential function @xmath207 always diverges at the origin @xmath28 if the deformation parameter @xmath4 is negative but it is not integer number . the behavior of this family of functions is shown in panel a ) of figure [ family_1.eps ] for some values of deformation parameter @xmath4 .
moreover , we have also considered the _ deviation function _ : @xmath208 that characterizes the relative deviation of @xmath4-exponential @xmath207 from the ordinary exponential .
the behavior of this later function is illustrated in panel b ) of figure [ family_1.eps ] .
using the power - expansion expression ( [ integral.gamma ] ) , one can obtain the _ integration - differentiation rules _ : @xmath209 and the _ convolution formula _ : @xmath210 this last relation is obtained using the _ beta function _ : @xmath211 note that definition ( [ def ] ) is a particular case of convolution formula ( [ convolution ] ) for @xmath25 .
moreover , this identity accounts for the fractional integration of @xmath4-exponential function : @xmath212
velazquez thanks the financial support of conicyt / programa bicentenario de ciencia y tecnologa psd * 65*. he also thanks partial financial support from vridt - ucn research programme . | we introduce a parametric family of models to characterize the properties of astrophysical systems in a quasi - stationary evolution under the incidence evaporation . we start from an one - particle distribution @xmath0 that considers an appropriate deformation of maxwell - boltzmann form with inverse temperature @xmath1 , in particular , a power - law truncation at the scape energy @xmath2 with exponent @xmath3 .
this deformation is implemented using a generalized @xmath4-exponential function obtained from the _ fractional integration _ of ordinary exponential . as shown in this work , this proposal generalizes models of tidal stellar systems that predict particles distributions with _
isothermal cores and polytropic haloes _
, e.g. : michie - king models .
we perform the analysis of thermodynamic features of these models and their associated distribution profiles .
a nontrivial consequence of this study is that profiles with isothermal cores and polytropic haloes are only obtained for low energies whenever deformation parameter @xmath5 .
pacs numbers : 05.20.-y , 05.70.-a keywords : astrophysical systems , evaporation , thermo - statistics e - mail : [email protected] ; [email protected] |
the idea that fundamental constants of physics were not properly _ constant _ , but could instead vary with time ( and possibly in space ) is not a new one @xcite , but a fruitful revival has been possible only quite recently , stimulated by the progress achieved in observational cosmology ( for a review , see @xcite ) .
on one side , the standard big bang scenario suffers some theoretical shortcomings , as the horizon and flatness problems , which are at the base of the introduction of cosmological inflation @xcite . on the other
, we have attended the discovery of the accelerated expansion of our universe @xcite and the detection of a possible variation of the fine structure constant from quasars absorption lines @xcite .
we have focused our attention onto the possibility that the speed of light might change in time during the evolution of the universe ; such scenario is generally called as a varying speed of light ( vsl ) theory . a serious theoretical approach to define in the correct way a valid vsl theory is recent , and aimed exactly at solving horizon , flatness and the acceleration problems , in a `` more natural '' way , without relying on inflationary scenarios and the cosmological constant ( the main successful candidate to lead universe accelerated expansion , but also herald of many theoretical problems , see @xcite for a review ) .
the most exemplificative literature on this topic includes @xcite . in the very own words of some of its pioneers , vsl foundations are still far from fixed , and
a lot of debate is around them @xcite .
but we think it should be honestly agreed that there as many brilliant and important scientists on the pro - vsl side , and not only on the against - vsl one .
thus , studying vsl theories and implications is not a fallacy , but deserves attention . in particular , the debate concerns breaking the lorentz invariance which vsl theories intrinsically produce , and the correctness of discussing about variation of dimensional quantities , while the only non - controversial claims should be coming from dimensionless quantities .
while the former question has some possible reliable solutions @xcite , the latter is still open .
again , we want to stress here an important point , in order to make our work judged with the right perspective : we do not want to make any claim about the vsl theoretical background .
this is out of the purpose of this paper and will be postponed to future works . here , instead
, we will study whether , if there is a vsl signal , and whatever is the way it can be explained , it can be detected or not , by present or future observations . in this context , recently , we have proposed a method to _ measure _ the speed of light _ on cosmological scales _ and _ at relatively high redshift _ @xcite using observations from galaxy surveys .
this method should overcome some of the criticisms related to the fact that the speed of light is , actually , a dimensional quantity : we can measure _ here and now _ such speed in laboratory ; we have _ relocated _ this laboratory in the outer universe , where observations provide us a ( cosmological ) ruler and a ( cosmological ) clock , which both can be employed to measure the speed of light .
given that we will use these rulers and clocks in this present work , we briefly review what they are .
both of them can be measured by a galaxy survey .
the ruler is the sound horizon measured at late times as it is imprinted in the clustering of galaxies at cosmological scales or , equivalently , in the baryon acoustic oscillations ( bao ) @xcite .
the sound horizon has some very important properties : it is generally considered as a standard ruler , because its size in comoving coordinates is constant in time and thus can be used to calibrate / measure cosmological distances ( for alternatives , see @xcite ) ; its length can be exactly calculated from theory ( @xmath6 mpc in physical units ; the best - precision value , measured by _ planck _
, is @xmath7 mpc for the baseline model @xcite ) . due to the strong correlation between photons / radiation and gas in the early times ( prior to the recombination epoch ) , by analyzing the galaxy correlation function now , it is possible to infer a correlation length which , expressed in comoving units , corresponds exactly to the sound horizon .
generally , such ruler is both present in the tangential and radial distribution of galaxies , which can respectively be defined as @xmath8 where @xmath9 is the speed of light ( generally assumed constant ) ; @xmath10 is the cosmological redshift ; @xmath11 is the angular diameter distance ; @xmath12 is the hubble function ( expansion rate ) ; and @xmath13 is the sound horizon , evaluated at recombination ( or dragging epoch ) . actually , with present data we do not have enough strong signal to measure the two directions separately , or , at least , not at the level of accuracy which should be theoretically possible @xcite .
this will eventually be possible with future surveys , when larger number of galaxies is available ; see , for example , forecast analysis for the square kilometer array ( ska ) , _ _ euclid _ _ @xcite , _ _
_ _ @xcite , the baryon oscillation spectroscopic survey ( boss ) @xcite , the extended boss survey ( eboss ) @xcite , the dark energy spectroscopic instrument ( desi ) @xcite and the hobby - eberly telescope dark energy experiment ( hetdex ) galaxy surveys are beneficial for our purposes also because they can provide us the cosmological clocks we need for our method to be implemented : a sample of the observed galaxies can be targeted as _ cosmic chronometers _ @xcite .
the key idea is to find a `` cosmological clock '' , which is able to give the variation of the age of the universe with redshift .
if one has this clock , then , one simply has to measure the age difference @xmath14 between two redshifts separated by @xmath15 , and calculate the derivative @xmath16 .
then , this quantity can be directly related to the expansion rate ( hubble function ) , defined as @xmath17 with such a method we would have a measurement of the hubble function free from any assumption on the nature of the metric , which normally affects , for example , the definition of cosmological distances . actually , passively - evolving early - type galaxies ( etg ) turned out to be reliable candidates to play the role of such clocks . since the first proposal , stellar population models have been improved ; a much larger number of galaxies has been observed and collected , up to a redshift @xmath18 ; and more precise tools to calibrate the clocks have been introduced ( e.g. , the @xmath19 break in etg spectra ) . and this scenario can still be improved using future galaxy surveys in the optical , as _ euclid _ and _ wfirst-2.4 _ , which should observe at least ten times more galaxies compared to the present ( and etg , eventually ) . our works
@xcite have been recently questioned by @xcite , where the authors point out two possible drawbacks of our method : first , that the speed of light is measured only at one single redshift @xmath20 ; second , that we ignore the spatial curvature contribution , which is degenerate with vsl . for what concerns the first point , it is true , but we were interested in the intrinsic novelty of the method .
it is well known that the angular diameter distance has a maximum at some high redshift value , which we called _ maximum redshift _
, @xmath20 ; we found that at the maximum , the relation @xmath21 holds , e.g. only at the maximum redshift , the combination of the angular diameter distance and of the expansion rate is _ exactly equal _ to the value of the speed of light at that epoch , with a minimal number of theoretical assumptions on the cosmological background , and with no need of any information at all about how the speed of light should vary or not .
direct _ measurement , albeit local . about the second point , we have always been aware that , among the minimal number of theoretical assumptions we needed in order to derive eq . ( [ eq : max_da ] ) , we have to assume a null spatial curvature .
however , we have also discussed how a non - vanishing curvature may impact our results .
thus , in this present work , we move a step forward : using the same cosmological rulers and clocks from @xcite , we build a new method ( different from @xcite ) which can be employed to recover a _
redshift - extended _ vsl signal ( no more limited to @xmath20 ) .
we also report a detailed discussion about how a non - zero curvature can influence the application of such method in a vsl theory context .
finally , we show how can the method be also applied in a classical context , where @xmath9 is constant in order to measure the spatial curvature itself . in section ( [ sec : methodology ] ) we describe the theoretical apparatus at the base of our method ; in section ( [ sec : application ] ) we describe all the steps required for our method to be built in more detail ; in section ( [ sec : results ] ) we discuss the results obtained from the application of our method to some mock data from future galaxy surveys ; and , finally , in section ( [ sec : conclusions ] ) we sum up our results .
to start , we need the observational data which is available from future bao galaxy surveys : the angular diameter distance @xmath22 and the expansion rate @xmath23 .
we define as @xmath24 and @xmath25 the results of such observation , i.e. the _ numbers _ that outcome the measurement processes . then
, we need to fix the theoretical background underlying the implementation of our method .
our work is based on two main and general assumptions : we assume a friedmann - robertson - walker metric and no spatial curvature .
the former is the simplest and most general assumption which agrees with data and , up to some statistical accuracy , one of the main ingredients of the nowadays accepted consensus model .
the latter assumption can hide a possible degeneracy between a vsl signal and the curvature ; we will discuss this point in a later section and we show that , sticking to the present observational status , it is of limited concern .
now , let us remind the theoretical definition of the angular diameter distance , @xmath26 where @xmath27 is the theoretical hubble function ; and @xmath28 is the speed of light expressed as any possible function of redshift . in a standard scenario ,
the speed of light is constant and @xmath29 ; in the more extended context of a vsl approach , it can be any function , unknown to us until we do not recover it from the data . for what concerns @xmath27 instead , in principle
it can be derived from the first friedmann equation ( in combination with a continuity equation ) once a cosmological model is given and so contains any possible information on the cosmological background .
then , we can assume that @xmath30 i.e. , that the theoretical @xmath11 function ( ignoring what is on the right hand side of eq .
( [ eq : da_real ] ) ) is _ explicitly _ equal to the function that can be directly obtained by observations . on the other hand
, we can also assume that the _ unknown _ theoretical @xmath27 is _ explicitly _ equal to the function that can be obtained by observation , @xmath25 .
actually , this is much more than just an assumption : observations always bring signatures of the real underlying cosmological model , whose ignorance we parameterize in many ways .
for example , by introducing the energy - matter dimensionless parameters ( eg . @xmath31 and @xmath32 ) , or the dark energy equation of state @xmath33 , and so on .
we highlight a very important point of our approach , which can be inferred by eq .
( [ eq : da_real ] ) : we do not need any cosmological assumption ( apart from the two we have stated above ) for @xmath27 , because we will directly use the output from the observations , i.e. @xmath34 , in order to calculate what we have called the _
real _ angular diameter distance .
the main point here is that we do nt know , _ a priori _ , if the speed of light appearing in eq .
( [ eq : da_real ] ) is a constant or not . the definition of _ real _ is just given to stress that we derive it using _ real _ observations , and not any theoretical cosmological model .
in fact , we will have to compare @xmath24 with another quantity , which we will call the _ reconstructed _ angular diameter distance : this can be defined directly using @xmath25 , i.e. @xmath35 where , again , @xmath36 refers to the expansion rate function directly derived from observations , and not from a theoretical model . as clearly shown above through eqs .
( [ eq : da_real ] ) and ( [ eq : da_rec_0 ] ) , in order to convert @xmath36 , with the dimension of the inverse of time , to a distance , we need to make some assumption on the speed of light .
the most general and common assumption is that it was constant ; thus , the reconstructed angular diameter distance is : @xmath37 strictly speaking , this quantity is quite useless from a cosmological point of view .
generally , all the information we need is hidden in @xmath27 : one proposes a cosmological model , which leads to @xmath27 as a function of some parameters ; and finally one tries a fit of this model with observational data , in order to recover some information about it .
here there is no benefit in using eqs .
( [ eq : da_rec_0 ] ) or ( [ eq : da_rec ] ) : if we use directly the numbers coming out of observations ( i.e. @xmath25 ) , without any underlying theoretical background , we are loosing any possibility to recover the information on the cosmological model .
but , if we change our perspective , and we strictly look at vsl theories , then they reveal their benefit .
in fact we face a question : what if we have a _ real _ vsl to be detected ? in this case , by comparing eqs .
( [ eq : da_theo ] ) - ( [ eq : da_real ] ) with eq .
( [ eq : da_rec ] ) , we can easily check that if @xmath38 then the assumption we made to define @xmath39 , i.e. @xmath40 , was right and , indeed , @xmath29 .
but , if instead we find out that @xmath41 then we might conclude that this same assumption is wrong , and that @xmath42 .
this also means that , by comparing @xmath24 with @xmath39 , we should be able to reconstruct the real unknown function @xmath28 .
actually , this is not really straightforward in this case because the speed of light @xmath28 enters the definition of the angular diameter distance through an integral , thus , the vsl signal can not be isolated so easily and will result to be smoothed .
but things go much easier if we focus on @xmath25 instead , and a possible _ reconstructed _ @xmath43 . in this case
, on one side we will have the _ direct _ observational data , obtained from the derivative of the real observed @xmath24 as @xmath44 \equiv \frac{c(z)}{h^{real}(z)}\ ; , \ ] ] where , again , we have identified the unknown theoretical @xmath27 function in eq .
( [ eq : da_theo ] ) with the observed @xmath36 , as we did in eq .
( [ eq : da_real ] ) .
on the other , we will have a _ reconstructed _ set of @xmath45 where we need an explicit assumption of a constant speed of light in order to convert time observations @xmath23 into distances ( @xmath46 ) . again , if we find that @xmath47 then the assumption that the speed of light is constant will reveal to be well based .
on the contrary , if @xmath48 then @xmath42 .
what is important to stress is that , by working with @xmath46 , we can directly obtain ( or reconstruct ) the analytical redshift function @xmath28 , through the ratio : @xmath49 in this way we are also circumventing the `` dimensionless - dimensional measurement '' debate , because we are going to reconstruct a ( dimensionless ) relative variation of the speed of light , not an absolute ( dimensional ) quantity .
given the methodological basis of our model , we will now describe how to apply it in the best possible way and , in particular , we will focus on what are the limits of the accessible information we should expect from future surveys .
the first point to be addressed is : what kind of data we are going to use ? as stated in the section [ sec : introduction ] , our purpose is to show how to employ future galaxy surveys for a non - standard cosmological analysis . from such surveys
we will expect to obtain separate information on @xmath11 ( from bao ) and @xmath12 ( from cosmic chronometers ) . at the present stage
, we do not have yet independent measurements on @xmath11 and @xmath12 , so we will need to produce mock data for our analysis , in the style of @xcite and @xcite . as we have pointed out in the section [ sec : introduction ] ,
there is no uniformity in approaching a vsl scenario ; but we also want to stress that the exact choice of the vsl approach is meaningless in our case .
we are not going to have any fit , or any test of any particular model ; we only need to produce some mock observational data with a vsl signal included .
the only requirement we will ask for is that , at least , such mock data were compatible with the present observations and with the present consensus model , at least in the redshift range now covered .
in such a case , clearly , the vsl model would be indistinguishable from the standard scenario , but still compatible with observations .
this sounds like a quite reasonable requirement : `` observations are observations '' , what we measure and see is independent of our understanding of the underlying theory . of course , a vsl signal can imply a different physical evolution in / of some processes , but the measured outcomes can not be different from what we see now .
for example : the sound horizon can be obviously influenced by a vsl .
but it can also be measured with a very high confidence in a cosmic microwave background experiment .
theory has to adjust to this measurement , not viceversa .
thus , if we create a mock vsl data set which is compatible with present observations , we are just implying that the vsl signal has to be consistent with them .
following @xcite , if a vsl is introduced with a minimal coupling with gravity , then we have modified versions of the first friedmann equation and of the continuity equation which are , respectively : @xmath50 and : @xmath51 where : @xmath52 and @xmath53 are , respectively , the mass density and the pressure of any fluid in the universe ; @xmath54 is the scale factor ; @xmath55 is the universal gravitational constant ; and the speed of light is expressed as a general function of time ( or redshift ) , @xmath56 . as we have anticipated before ,
a degeneracy between vsl and geometry is possible : indeed , any change produced by a vsl is connected with the spatial curvature .
we will discuss this later in more detail ; for now , we will assume that universe is spatially flat , e.g. @xmath57 , which implies that no effective change is effective in the continuity equation and , consequently , in the first friedmann equation ( at least , in terms of the energy - mass equations of state ) .
on the other hand , in the calculation of @xmath24 , the vsl also operates through the @xmath28 function which enters the integral .
it is thus clear that , in order to produce our mock data , we need to assume an ansatz for @xmath28 ; we follow @xcite and consider the ansatz : @xmath58 where @xmath59 is the scale factor , and @xmath60 sets the transition epoch from some @xmath61 ( at early times ) to @xmath62 ( now ) .
the fiducial cosmological model used to produce the mock data in this work is a slightly modified version of the baseline @xmath63cdm model from _ planck
_ 2015 releaseparameters . ] , ` base`@xmath64`plikhm`@xmath64`ttteee`@xmath64`lowteb`@xmath64`lensing`@xmath64`post`@xmath64`bao ` .
this model is characterized by a dimensionless matter density today equal to @xmath65 .
we have to modify slightly this parameter when introducing a vsl because a vsl can mimic an accelerated expansion ( this was the original motivation for starting to study vsl theories ) and , thus , can be seen as a contribution to the dark energy sector .
more precisely , in a vsl context , the acceleration would not be given by a real cosmological fluid , but would be an implicit effect due to a varying speed of light . in any way
, this means that , when adding a vsl , the contribution from a dark energy fluid diminishes and , consequently , in a spatially flat universe , @xmath31 might grow . in this work
we have considered two different vsl scenarios : one , given by the paramaters @xmath66 and @xmath67 , corresponds to a redshift - increasing speed of light , with an average variation @xmath68 at redshift @xmath69 ; the other , given by the paramaters @xmath66 and @xmath70 , corresponds to a redshift - increasing speed of light , with an average variation @xmath71 at redshift @xmath69 .
this redshift range is used as a reference , following the nomenclature used in @xcite . in table
i of @xcite we also show how both the models are consistent with present observations and with the fiducial cosmological model from _ planck _ , if we assume for them , respectively , @xmath72 and @xmath73 .
once we have defined our input cosmological model , we can produce the mock @xmath74 and @xmath75 ; but in order to produce realistic mock data , i.e. , the previously defined @xmath25 and @xmath24 , we still need the observational errors on these quantities . in @xcite ,
many on - going and future surveys are analyzed ; the authors analyze what are the errors we should expect on such quantities by each one of these surveys , assuming redshift bins of @xmath76 width . among them , we will focus on : boss , desi , and _
wfirst-2.4 _ because , in their respective redshift ranges , they show the best performances . for boss
, we will consider @xmath77 ; for desi , @xmath78 $ ] ; for _ wfirst-2.4 _ , @xmath79 $ ] . for the intermediate range
@xmath80 $ ] we use the ska results from @xcite ; the performance of ska will outcome the others by at least one order of magnitude , thus , it will be quite natural to expect the best results of our approach from this redshift range . with @xmath74 and @xmath75 and the corresponding errors
, we can now simulate realistic data : we randomly generate our values of @xmath25 and @xmath24 from a multivariate gaussian centered on the fiducial values , and with a total covariance matrix built up from the errors defined for each survey .
we also additionally assume a correlation factor between @xmath74 and @xmath75 @xmath81 , as derived in @xcite .
we want to stress here two points .
first : the errors from @xcite are not given directly in terms for @xmath12 and @xmath11 but , instead , for @xmath82 and @xmath83 , where @xmath84 is the sound horizon at the decoupling / dragging epoch .
thus , if we want to work with @xmath12 and @xmath11 derived from a bao survey , we need to multiply the previous defined combination by the sound horizon . examining all the cases covered by the _
planck _ mission and collected in the _ planck _ legacy archiveparameters . ] , it is possible to check that the dependence of the sound horizon on the cosmological model is very weak ( on the other hand , if this were not the case , it could not be considered as a standard ruler ) .
and its dispersion is much smaller than the observational error , which is @xmath85 .
thus , the contribution of the sound horizon to the total error budget , whatever is the value used for it , is quite negligible .
second : we do not have this problem for @xmath12 , because we assume it is derived from cosmic chronometers ; but we lack a forecast analysis for the errors expected on @xmath12 using this probe for future surveys . in @xcite
we have some estimations from _ euclid _ , with a minimum statistical error @xmath86 .
for our analysis we will use the @xmath12 errors estimated by a bao survey ; we have to think about them as a possible precision goal for future surveys ( next to stage iv ) , but they will always give us a precise indication of how feasible and applicable is our approach . before using eq .
( [ eq : rec_c ] ) in order to reconstruct a possible vsl signal , many questions have to be addressed and problems solved .
the first one is intrinsic to our definition of @xmath87 given by eq .
( [ eq : h_real ] ) : it is the derivative with respect to redshift of a quantity @xmath88 which is represented by a discrete set of points ( observations ) which have an intrinsic dispersion around the underlying fiducial cosmological model .
the problems related to the dispersion can not be avoided : the dispersion is intrinsic to the measurement process , and we can only hope to have , in the future , better measurements which can reduce it ( but its nature is not of statistical origin only ) .
thus , we will always have an intrinsic systematic error in the derivation of @xmath87 ; moreover , the dispersion alters the derivative calculation and thus , as it is known and expected , the errors on the derivated quantity tend to explode
. having assumed that this problem can not be avoided , we can rely on another property of our approach : given that we are not interested in the explicit form of @xmath12 , because we will directly use observations to infer a function which _ interpolates _ them , we are not forced to fit our quantities following some cosmological - model - based requirements .
thus , we can try a fit based on the best analytic functions which can work in this situation and , eventually , we can apply general physical priors to such functions .
but , how do we actually apply such priors ?
first , we generate a random mock data set @xmath89 ; then , we apply the fits we will describe in what follows in this section ; if a generated set satisfies the chosen priors , we save it ; and if not , we reject it . at the end ,
we are left with @xmath90 random _ physically realistic _ mock data sets , and we apply our algorithm to each of them . in our case , we need analytic functions for fitting both @xmath25 and @xmath24 , and they have different requirements . for @xmath25
we have found that a simple sixth - order redshift polynomial gives an optimal fit to @xmath25 in the redshift range we are covering , i.e. @xmath91 $ ] ; higher - order polynomials do not improve the fit .
we also apply a relaxed prior requiring that @xmath92 obtained by such polynomial fit lies in the range @xmath93 $ ] , where @xmath94 is an independent measurement of the hubble constant , and @xmath95 its error . in our case
, we have used the value @xmath96 from @xcite . as a further prior
, we can also ask that @xmath97 all over the redshift range @xmath98 . for @xmath24
a polynomial fit is unsatisfactory to describe the peculiar property of the angular diameter distance to have a maximum at relatively low redshift values .
a better and more flexible fit is given by the pad@xmath99 approximant : @xmath100 which clearly satisfies the expected conditions : @xmath101 for @xmath102 and @xmath103 ; moreover , we require that @xmath104 for @xmath98 , and that @xmath105 lies in the range @xmath93 $ ] .
once the fits are run for both @xmath25 and @xmath24 , we have a set of parameters ( the parameters of the polynomial and of the pad@xmath99 approximant ) , respectively , with their covariance matrix and errors bars ; after the correct propagation error rules are applied , we end up with a set of polynomial - reconstructed @xmath87 and @xmath106 , with related errors , from which we can derive the @xmath107 ratio through eq .
( [ eq : rec_c ] ) .
finally , this last quantity can also be fitted ( or reconstructed ) ; the function we have been working with is the pad@xmath99 approximant given by : @xmath108 imposing the conditions : @xmath109 , and that @xmath28 is always positive for @xmath98 .
we have verified that the functions we have finally chosen to fit @xmath25 and @xmath24 are really good approximations to the fiducial model all over the entire redshift range @xmath98 , and not only in the redshift interval we have decided to work with because covered by next galaxy surveys .
on the other hand , the function chosen for @xmath28 has some degree of arbitrariness : it describes very well our input vsl in the galaxy surveys redshift range , but not at very high redshifts .
but it is very general , and with such a high level of flexibility that it can be used as a testing function to detect if a vsl signal is working or not in any case ( but any suggestion can be considered ) .
the first point to be examined is how good is the reconstruction of the hidden vsl signal ; then , we will move to a much detailed analysis of the possible degeneracy between the vsl signal and a non - zero spatial curvature . in our case
, we know what is the behavior of @xmath28 , so that we can easily check if the final reconstructed @xmath28 gives a reliable description of this known input , thus testing if our algorithm works well or not . in figure
[ fig : pend ] we show both the cases of a @xmath110 and of a @xmath111 vsl signal , joining results from all the @xmath112 simulations we have realized . in black , we show the @xmath4 confidence level span by all the simulations for the vsl pre - fitting signal in eq .
( [ eq : rec_c ] ) , i.e. the quantity @xmath107 calculated from @xmath87 and @xmath106 after fitting @xmath25 with a sixth - order polynomial and @xmath24 with the pad@xmath99 approximant given in eq .
( [ eq : pade_da ] ) . in red
, we plot the @xmath4 confidence level span by all the simulations for the vsl signal in eq .
( [ eq : rec_c ] ) after fitting the ratio @xmath113 with the pad@xmath99 approximant given in eq .
( [ eq : pade_cz ] ) . from a simple visual inspection
, it can be seen that , at least at @xmath4 level , a @xmath110 vsl signal can in principle be detected in the redshift range @xmath114 $ ] , where the angular diameter distance from bao and the hubble function from cosmic chronometers have been assumed to have the precision actually forecast from ska . on the other hand
, it is also clear that a @xmath111 vsl signal will be hardly detected with the same prescriptions .
a more detailed inspection of the possibility to detect a vsl signal is given in figure [ fig : pprobabend ] : for each simulation , and in each redshift bin , we calculate the residuals with respect to a constant speed of light , i.e. @xmath115 ; then , we plot the normalized number of simulations for which such residuals are positive , implying a clear detection of a non - constant @xmath107 . in blue ,
we show results when the residuals with respect to constant speed of light are calculated using the best fit pad@xmath99 approximant , eq .
( [ eq : pade_cz ] ) ; in red , the residuals are calculated using the lower @xmath4 limit derived from the same best fit function , thus indicating a detection of the vsl signal at @xmath4 confidence level ; in green and yellow , respectively , the residuals are calculated using the lower @xmath116 and @xmath1 limits .
we focused on the lower limits because our fiducial input vsl corresponds to a speed of light higher than @xmath9 at higher redshift ; in a more realistic case , one should check for both positive and negative residuals , and test , in the same way we are doing here , if there is any statistically clear evidence for one trend over the other . looking at figure [ fig : pprobabend ] we can now have a more clear and precise prediction of what could happen in the next future .
the probability to detect a @xmath110 vsl at a @xmath1 level is higher than the @xmath117 ( e.g. , in @xmath117 of our simulations we are able to detect a @xmath110 vsl signal at a @xmath1 level ) approximately in the redshift range @xmath118 $ ] . for higher redshifts , using the precision actually forecasted for next future galaxy surveys , we see that the signal degrades very rapidly .
this point is interesting : given our input @xmath28 , the deviation from the @xmath115 limit grows with redshift .
thus , higher redshifts imply larger deviations .
but this does not automatically converts in a clearer or easier detection : if the survey precision degrades too fast , we are going to lose any possibility to detect the signal at high redshift .
at the same time , this is also encouraging : given the possibly larger deviation from the constancy of the speed of light at high redshifts ( if the vsl signal is a monotonical function , of course ) , we have a lot of room to improve its detection in this range , because the precision update required for future surveys at higher redshifts is well inside our technological possibilities . another interesting point
to stress is that in @xcite we have shown how ska will be able to put a @xmath1 limit on a @xmath110 vsl signal at the maximum redshift in the angular diameter distance , which should locate at @xmath119 ( at least , basing this claim on our present knowledge of the cosmological background model ) . in the method exposed in this work ,
the detection at @xmath1 for a redshift @xmath120 is possible in the @xmath121 of our simulations ; still a high probability , even if quite lower than the @xmath122 limit assessed above .
it is not surprising that the two methods have different sensitivity at this redshift , because they rely on different algorithms ; in particular , the maximum detection method described in @xcite can be pushed to a better precision , while the present method is mainly limited by the not - perfect correspondence of the derivatives calculated from real data with those intrinsic to the unknown cosmological background .
but , still , the two methods are complementary , helping to extend the final redshift range of vsl detection .
unfortunately , from figure [ fig : pprobabend ] it is also clear that a @xmath111 signal will be hardly detected : at @xmath4 level , the probability detection of a vsl signal of such magnitude is @xmath123 in the redshift range @xmath124 $ ] ; a @xmath1 detection in the same range is achieved only in @xmath125 of our simulations , thus making it difficult to statistically state if it can be really reached or not .
another important question to be stated is what level of goodness has our method to reconstruct the real vsl background . in order to assess this question ,
we calculate the quantity : @xmath126 where @xmath127 is the total number of simulations we have run ; @xmath128 is the varying speed of light given by the resulting best pad@xmath99 approximant , eq .
( [ eq : pade_cz ] ) , evaluated at each redshift @xmath129 ; and @xmath130 is the fiducial speed of light given by eq .
( [ eq : ansatz_c ] ) .
thus , the quantity @xmath131 is the sum , at each redshift and all over our simulations , of the relative squared residuals between our final reconstructed vsl @xmath132 and the fiducial one @xmath133 .
smaller is its value , better is the agreement between our reconstruction and the true underlying model . in this way
, we have a criterium to establish if our reconstruction is accurate or not . in figure
[ fig : residual ] we plot the logarithm of this quantity ; conclusions are the same we have derived above : in the case of a @xmath110 vsl , it is clear that the agreement is quite good and very similar in the redshift range @xmath134 $ ] ; then it starts to decrease , with a @xmath135 that , at @xmath136 , is one order of magnitude larger then the minimum value achieved ; and things go even worse for larger redshift values .
the same is more or less valid for the @xmath111 case . as pointed out in previous sections , in the friedmann and continuity equations the vsl - based terms , @xmath56 and @xmath137 , come coupled with the spatial curvature parameter @xmath138 .
thus , it is natural to expect some degree of degeneracy between a possible vsl signal and what could instead interpreted as a geometric effect .
our main equations are derived , as said previously , assuming that the universe if spatially flat , i.e. @xmath57 . all the most updated observations confirm such assumption @xcite ; but we want to show here that even if we take into account curvature , still there is a wide range of validity for our equations and , even , our method might be generalized and used , in the standard context of constant speed of light , as an alternative way to measure the spatial curvature .
when taking into account spatial curvature , the main change is in the determination of what we have defined @xmath139 , defined in eq .
( [ eq : h_real ] ) as the derivative of observational @xmath24 with respect to redshift .
if the curvature is allowed to vary , then , the most general definition for the angular diameter distance is @xmath140 where @xmath141 is the dimensionless curvature density parameter today ; @xmath142 is the hubble distance ; and the line - of - sight comoving distance is defined as @xmath143 , where we have made use of the general ansatz @xmath144 , with @xmath145 for @xmath146 .
we are assuming here the most general case of a varying speed of light @xmath28 ; but the standard scenario can be easily recovered simply replacing @xmath28 with @xmath9 any time it appears . if we now calculate @xmath139 through the same eq .
( [ eq : h_real ] ) , we have : @xmath147 it is clear that even if we assume @xmath29 , we would still have some contribution from the @xmath148 term ; thus the case `` vsl @xmath149 spatial flatness '' would be equivalent to `` constant @xmath28 @xmath149 curvature '' .
we can easily quantify how much information we might derive , and which we might erroneously attribute to a vsl signal only , should instead be shared with a non - null curvature signal . from the _ planck _ legacy archive , the extension of the baseline model with a free curvature parameter , named ` base`@xmath64`omegak`@xmath64`plikhm`@xmath64`ttteee`@xmath64`lowteb`@xmath64`bao`@xmath64`h070p6`@xmath64`jla`@xmath64 + ` post`@xmath64`lensing ` , gives the value of @xmath150 at the @xmath151 confidence level ( and @xmath152 at the @xmath122 ) .
we can thus compare the curvature - correction terms in eq .
( [ eq : rec_c_curv ] ) with the null curvature hypothesis , using our ansatz for the vsl , eq .
( [ eq : ansatz_c ] ) .
results are shown in figure [ fig : curvature ] .
the first possible conclusion is that a realistic contribution from the spatial curvature to our method ( red line ) would be @xmath153 at the maximum in @xmath11 ( for a more direct and straightforward comparison , we use the same maximum criterium we have used to define the @xmath110 and the @xmath111 vsl models ) ; thus , it would be even smaller than the @xmath111 vsl signal ( black dashed line ) we have considered so far , and would thus result , finally , undetectable .
this result , obtained in an independent and alternative way , is also consistent with a recent attempt described in @xcite . for the sake of precision
, we have to stress again that , anyway , in general , a pure vsl and a pure curvature signal are degenerate .
we can detect a total signal , without being able to ascribe it to one or another .
what we can estimate is that , given present bounds on curvature , a @xmath110 signal ( solid black line ) could be attributed with no doubt to vsl only , rather than to any curvature contribution . even in the case of assuming both a vsl and non - null curvature ( dot - dashed redline ) , given the actual constraints on the latter one , the vsl signal might be @xmath154 , in order to have a final total @xmath110 detection
thus , at least at the scales which we have shown to be directly testable in the next future , curvature might play a negligible role .
but if the total signal should result to be less than @xmath110 , then we could have problems and would not be able to discriminate between them .
in this work we have extended the method previously described in @xcite : while the latter made it possible to measure the speed of light ( and , incidentally , detect any possible variation of the same quantity ) only in one well - located point ( the maximum redshift ) , here we show how it is possible to recover a _
redhisft - extended _ vsl signal on a much large redshift range .
we have used the cosmological observations which will be available in the next future from galaxy surveys , i.e. : estimations of the sound horizon at decoupling / dragging epoch , imprinted as an angular diameter distance @xmath11 in the clustering of the galaxies ; and the expansion rate data @xmath12 inferred from etg galaxies designated as cosmological clocks .
we have employed a quite various number of future galaxy surveys , boss , desi , _
wfirst-2.4 _ and ska , which result to have the best performances in different ( non - overlapping ) redshift windows . as we have discussed in section ( [ sec : application ] ) , and as it is shown in figure [ fig : pprobabend ] , given the sensitivities forecast for the previous surveys , there is a quite high probability @xmath155 to detect a @xmath110 vsl signal ( if any ) at @xmath1 confidence level in the redshift range @xmath156 $ ] .
smaller signals , of the order of @xmath111 , will be hardly detected by the same surveys .
we have also given a more detailed discussion about the impact that a possible non - null spatial curvature might have on the detection of the vsl signal .
in particular , we have shown that values of the curvature compatible with the present bounds given by _
planck _ are absolutely negligible with respect to a @xmath110 vsl signal .
we emphasize here that , even if we were not considering a vsl signal , but the classical constancy of the speed of light , then our method would result to be useful to detect _ curvature - only _ contributions .
in particular , if such contribution should result to be @xmath157 , then , it would be equivalent to a @xmath110 vsl signal , and all the discussion we have spent for the vsl theory might be equivalently exported to spatial curvature measurements only .
more problematic would be to disentangle smaller vsl signals , which would result to be of the same order of the geometrical contribution ; but , as we have shown here , such small signals are out of the detection possibilities of currently forecast galaxy surveys for the next @xmath158 years . in the meantime
, we may work to improve our method , and/or find alternative ones .
this work is financed by the polish national science center grant dec-2012/06/a / st2/00395 .
h. weyl , ann .
phys . , 59 , 129 ( 1919 ) a. s. eddington , the mathematical theory of relativity , ( cambridge university press , cambridge ) , 1923 a. s. eddington , new pathways in science , ( cambridge university press , cambridge ) , 1934 h.weyl , natturwissenschaften 22 , 145 ( 1934 ) p. a. m. dirac , nature 139 , 323 ( 1937 ) p. a. m. dirac , proc .
a165 , 189 ( 1938 ) a. a. , starobinsky , jtep lett .
30 , 682 ( 1979 ) a. a. , starobinsky , phys .
b 91 , 99 ( 1980 ) a. h. guth , phys .
d 23 , 347 ( 1981 ) a. albrecht , p. j. steinhardt , phys . rev .
48 , 1220 ( 1982 ) a. d. linde , phys .
b. 108 , 389 ( 1982 ) a. d. linde , phys .
b. 129 , 177 ( 1982 ) j. k. webb , v. v. flambaum , c. w. churchill , m. j. drinkwater , and j. d. barrow , phys .
82 , 884 ( 1999 ) v. a. dzuba , v. v. flambaum , and j. k. webb , phys
82 , 888 ( 1999 ) m. t. murphy , j. k. webb , v. v. flambaum , j. x. prochaska , a. m. wolfe , mon . not .
r. astron .
327 , 1237 ( 2001 ) m. t. murphy , j. k. webb , v. v. flambaum , m. j. drinkwater , f. combes , and t. wiklind , mon . not .
r. astron .
327 , 1244 ( 2001 ) j. k. webb , m. t. murphy , v. v. flambaum , et al .
87 , 9 ( 2001 ) m. t. murphy , j. k. webb , v. v. flambaum , mon . not .
r. astron .
345 , 609 ( 2003 ) j. n. bahcall , c. l. steinhardt , and d. schlegel , astrophys .
j. 600 , 520 ( 2004 ) h. chand , r. srianand , p. petitjean , and b. aracil , astronomy & astrophysics 417 , 853 ( 2004 ) r. srianand , h. chand , p. petitjean , and b. aracil , phys .
92 , 12 ( 2004 ) s. a. levshakov , m. centurin , p. molaro , m. v. kostina , astronomy & astrophysics 447 , l21 ( 2006 ) m. t. murphy , j. k. webb , v. v. flambaum , phys .
99 , 239001 ( 2007 ) r. srianand , h. chand , p. petitjean , and b. aracil , phys
99 , 239002 ( 2007 ) p. molaro , d. reimers , i. i. agafonova , and s. a. levshakov , eur .
j. special topics 163 , 173 ( 2008 ) m. t. murphy , j. k. webb , v. v. flambaum , mon . not .
r. astron .
384 , 1053 ( 2008 ) i. i. agafonova , p. molaro , s. a. levshakov , and j. l. hou , astronomy & astrophysics 529 , a28 ( 2011 ) j. c. berengut , v. v. flambaum , j. a. king , s. j. curran , and j. k. webb , phys .
d83 , 123506 ( 2011 ) j. k. webb , j. a. king , m. t. murphy , v. v. flambaum , r. f. carswell , and m. b. bainbridge , phys . rev .
107 , 191101 ( 2011 ) p. molaro , m. centurin , j. b. whitmore , et al . ,
astronomy & astrophysics 555 , a68 ( 2013 ) h. rahmani , n. maheshwari , and r. srianand , mon . not .
r. astron .
439 , l70 ( 2014 ) j. moffat , international journal of physics d 2 , 3 ( 1993 ) j. moffat , foundations of physics 23 , 411 ( 1993 ) a. albrecht and j. magueijo , phys .
d59 , 043516 ( 1999 ) j.d .
barrow , phys .
d59 , 043515 ( 1999 ) p. p. avelino , c. j. a. p. martins , phys .
b 49 , 468 ( 1999 ) j. d. barrow and j. magueijo , phys .
b 443 , 104 ( 1999 ) j. d. barrow and j. magueijo , phys . lett .
b 447 , 246 ( 1999 ) j. d. barrow and j. magueijo , class . quant .
16 , 1435 ( 1999 ) m. a. clayton , j. w. moffat , phys .
b 460 , 263 ( 1999 ) j. d. barrow and j. magueijo , astrophys
. j. lett .
532 , l87 ( 2000 ) m. a. clayton , j. w. moffat , phys .
b 477 , 269 ( 2000 ) b. a. bassett , s. liberati , c. molina - paris , m. visser , phys . rev .
d 62 , 103518 ( 2000 ) m. a. clayton , j. w. moffat , phys . lett .
b 506 , 177 ( 2001 ) m. a. clayton , j. w. moffat , int
d 11 , 187 ( 2002 ) j. magueijo , rep .
66 , 2025 ( 2003 ) j.w .
moffat , arxiv : 1404.05567 j.w .
moffat , arxiv : 1501.01872 p.
j. e. peebles & j. t. yu , apj 162 , 815 ( 1970 ) r. a. sunyaev & ya .
b. zeldovich , astrophysics and space science 7 , 3 ( 1970 ) a. g. doroshkevich , ya .
b. zeldovich and r. a. sunyaev , soviet astronomy 22 , 523 ( 1978 ) d. j. eisenstein , w. hu , apj 496 , 605 ( 1998 ) d. j. eisenstein , w. hu , m. tegmark , apj 504 , l57 ( 1998 ) a. cooray , w. hu , d. huterer , m. joffre , apj 557 , l7 ( 2001 ) c. blake & k. glazebrook , apj 594 , 665 ( 2003 ) w. hu & z. haiman , phys . rev .
d 68 , 063004 ( 2003 ) h .- j .
seo & d. j. eisenstein , apj 598 , 720 ( 2003 ) d. j. eisenstein , new astronomy reviews , v. 49 , iss . 7 - 9 , p. 360
- 365 ( 2005 ) d. j. eisenstein , i. zehavi , d. w. hogg , et al .
, apj 633 , 560 ( 2005 ) h .- j .
seo & d. j. eisenstein , apj 633 , 575 ( 2005 ) d. j. eisenstein , h .- j .
seo & m. white , apj 664 , 660 ( 2007 ) b. bassett , r. hlozek , arxiv:0910.5224 , _ `` dark energy : observational and theoretical approaches '' _ , edited by pilar ruiz - lapuente , published : cambridge , uk - new york by cambridge university press , 2010 d. h. weinberg , m. j. mortonson , d. j. eisenstein , c. hirata , a. g. riess , e. rozo , physics report 530 , 87 ( 2013 ) c. blake , e. a. kazin , f. beutler , et al . ,
mon . not .
r. astron .
418 , 1707 ( 2011 ) c. blake , s. brough , m. colless , et al . ,
mon . not .
r. astron .
soc . 425 , 405 ( 2012 ) l. anderson , e. aubourg , s. bailey , et al . , mon .
r. astron .
441 , 24 ( 2014 ) e. a. kazin , j. koda , c. blake , et al . , mon . not .
r. astron .
441 , 3524 ( 2014 ) l. samushia , b. a. reid , m. white , et al . ,
mon . not .
r. astron .
439 , 3504 ( 2014 ) t. delubac , j. e. bautista , n. g. busca , et al . , a&a 574 , a59 ( 2015 ) l. amendola , s. appleby , d. bacon , et al . , living rev .
relativity 16 , 6 ( 2013 ) r. laureijs et al .
, _ euclid _ collaboration - _ euclid _ assessment study report , arxiv:0912.0914 a. refregier et al . , _
euclid _ imaging consortium - _ euclid _ imaging consortium science book , arxiv:1001.0061 r. laureijs et al .
, _ euclid _ collaboration - _ euclid _ definition study report , arxiv:1110.3193 j. comparat , t. delubac , s. jouvel , et al . , accepted for publication in a&a , arxiv:1509.05045 a. prakash , j. newman ; sdss - iv / eboss collaboration , american astronomical society , aas meeting @xmath159 , @xmath160 ( 2015 ) j. tinker ; sdss - iv collaboration , american astronomical society , aas meeting @xmath159 , @xmath161 ( 2015 ) r. jimenez , a. loeb , astrophys .
j. 573 , 37 ( 2002 ) m. moresco , r. jimenez , a. cimatti , l. pozzetti , j. cosmol .
03 , 045 ( 2011 ) r. jimenez , p. talavera , l. verde , m. moresco , a. cimatti , l. pozzetti , j. cosmol .
astropart . phys .
03 , 014 ( 2012 ) m. moresco , l. verde , l. pozzetti , r. jimenez , a. cimatti , j. cosmol .
astropart . phys .
07 , 053 ( 2012 ) m. moresco , r. jimenez , a. cimatti , et al . , j. cosmol .
08 , 006 ( 2012 ) m. moresco , mon . not .
r. astron .
450 , l16 ( 2015 ) | we investigate a new method to recover ( if any ) a possible varying speed of light ( vsl ) signal from cosmological data .
it comes as an upgrade of @xcite , where it was argued that such signal could be detected at a single redshift location only . here
, we show how it is possible to extract information on a vsl signal on an extended redshift range .
we use mock cosmological data from future galaxy surveys ( boss , desi , _ wfirst-2.4 _ and ska ) : the sound horizon at decoupling imprinted in the clustering of galaxies ( bao ) as an angular diameter distance , and the expansion rate derived from those galaxies recognized as cosmic chronometers .
we find that , given the forecast sensitivities of such surveys , a @xmath0 vsl signal can be detected at @xmath1 confidence level in the redshift interval @xmath2 $ ] .
smaller signals @xmath3 will be hardly detected ( even if some lower possibility for a @xmath4 detection is still possible ) .
finally , we discuss the degeneration between a vsl signal and a non - null spatial curvature ; we show that , given present bounds on curvature , any signal , if detected , can be attributed to a vsl signal with a very high confidence . on the other hand , our method turns out to be useful even in the classical scenario of a constant speed of light : in this case , the signal we reconstruct can be totally ascribed to spatial curvature and , thus , we might have a method to detect a @xmath5-order curvature in the same redhift range with a very high confidence . |
numerous studies have shown a strong correlation between supermassive black hole ( smbh ) mass and host galaxy mass over broad mass ranges , implying that there is a close evolutionary relationship between them @xcite .
luminous quasars at high redshifts , which represent the most massive smbhs , are experiencing vigorous accretion of a significant portion of the final smbh mass .
this accretion activity is believed to be triggered by global processes ( galaxy mergers , interactions or perhaps secular evolution ) that also trigger major episodes of star formation in the massive host galaxies , which are also rapidly being assembled at high redshifts .
therefore , the processes of smbh formation and growth resulting in the quasar phenomenon are directly linked to the birth of massive galaxies @xcite .
however , the nature of the relationship between quasars and galaxy formation is not well understood .
it is widely believed that strong interactions between gas rich galaxies can result in ultra - luminous infrared galaxies ( ulirgs ) , defined by l@xmath1110@xmath12 , which have star formation rates ( sfrs ) of @xmath13 @xcite .
a fraction of ulirgs have been found to contain dust - enshrouded quasars @xcite .
there is evidence at both low and high redshift that these embedded quasars are precursors to optically luminous quasars .
for example , @xcite find that low redshift quasars with ir luminosities in the ulirg range fall between optically selected pg quasars and ulirgs ( starbursts ) for a variety of mid - ir spectroscopic indicators of starburst and active galactic nucleus ( agn ) contributions including : polycyclic aromatic hydrocarbon ( pah ) luminosities , fine structure emission line strengths , silicate absorption strengths , spectral slope and mid - ir colour indices .
these findings suggest that ir luminous quasars could represent an intermediate stage between a dominant starburst and a dominant agn phase .
quasars with high sfrs therefore may be at an intermediate stage between an embedded quasar in a star forming galaxy and a visible quasar in a galaxy where most star formation has ceased ( see also @xcite ) . at low redshifts ,
numerous star formation indicators in quasar host galaxies are seen , e.g. high ir luminosities , pah emission , strong ( sub)-mm emission and strong co emission @xcite . at high redshifts ( z @xmath9 2 - 6 )
, observations at sub - mm and mm wavelengths ( rest - frame far - infrared ( fir ) to sub - mm , depending on wavelength and redshift range ) of optically luminous quasars suggest that up to 30% of quasars also fall within the ulirg range , similar to the ir quasars studied by @xcite @xcite .
@xcite compare dynamical , gas and smbh masses of ten z @xmath14 sub - mm detected quasars and z @xmath14 sub - mm galaxies ( smgs ) , which are high redshift counterparts to ulirgs , though less extreme with more evenly distributed star formation instead of a localised starburst @xcite .
they find that the fainter half of their quasar sample could likely be ` transition objects ' between smgs and luminous quasars , based on their smg - like surface densities and their proximity to the local m@xmath15/m@xmath16 relation , since luminous quasars tend to lie above this relation and typical smgs tend to lie below .
the highest redshift sub - mm - selected source currently known has also been observed to posses similar stellar and gas masses to this z @xmath14 sample of transition objects , as well as a small agn contribution @xcite .
these quasars eventually may produce enough energy from accretion to effectively blow out the surrounding gas , halting star formation and clearing the view to an optically bright quasar before halting their own growth @xcite .
this regulation of star formation by the accreting smbh could naturally produce the observed black hole - galaxy mass correlation , and different stages of the process should present different sfrs , with initially high sfrs declining as the agn becomes more dominant , culminating in a phase of strong smbh accretion and little or no ongoing star formation @xcite .
fir luminosities can be used to determine the star formation rates in quasar host galaxies .
the fir emission is due to dust heated either by star formation or quasar emission ( see @xcite for discussion ) .
@xcite find evidence that fir emission in quasar host galaxies at all redshifts is caused by dust heated by star formation and not the quasar .
the strength of pah emission , which is found almost exclusively in star forming regions , tightly corresponds to the strength of fir emission in the same hosts @xcite .
@xcite measure fir emission from six high redshift quasars and derive fir to radio spectral indexes consistent with local star forming galaxies without agn .
the fir emission is seemingly uncontaminated by hotter dust potentially heated by the agn .
these results along with others ( see for example , @xcite ) further substantiate the claim that the fir luminosity in quasars is dominated by star formation and not by the agn .
fir luminosity traces ongoing star formation , but past star formation also can be observed indirectly by measuring chemical abundances .
several studies have found that high - redshift quasars typically have metallicities greater than or equal to solar metallicity in the broad emission line region ( blr ) , which requires significant previous star formation in the host @xcite .
these blr studies rely on emission line ratios such as / and + / that are sensitive to the selective enrichment of nitrogen as a secondary element and/or to the decreasing temperatures and increasing metal - line saturations that occur in metal - rich gas .
the main result for typically solar or higher metallicities near quasars has been corroborated by independent studies of the narrow emission lines @xcite and narrow absorption lines @xcite in quasar spectra .
the metal - rich blr result is true even for the highest redshifts studied , e.g. @xcite , with redshifts out to z = 6.4 .
there is no known change in metallicity with redshift ( see @xcite and references above ) .
simple chemical evolution models for quasars and elliptical galaxies find that galactic centres tend to be more metal - rich than their halos , and a centralised starburst can enrich the galactic centre to supersolar abundances in a short time ( @xmath17 ) @xcite .
these models , combined with the consistently super - solar gas abundances observed in quasar environments , imply that quasars tend to emerge after or near the end of ( potentially ) short centralised star formation epochs .
if the quasar phase emerges when star formation is on the decline , less advanced environments might have higher sfrs and lower abundances @xcite .
the absorption lines in quasar spectra provide additional information about outflows that might be related to the blowout of gas from the host galaxies , and perhaps , about the gaseous remnants of recent galaxy mergers .
these absorption lines may be more common in quasars with higher sfrs if mergers and interactions trigger the star formation as in ulirgs @xcite .
quasar ` associated ' absorption lines ( aals ) , near the emission redshifts with velocity widths less than 500 and broad absorption lines ( bals ) , with velocity widths greater than a few thousand , are examples of these potential merger , interaction and outflow signatures .
in systems with recent mergers , remnants from the interaction may manifest as a greater incidence of low - velocity ( @xmath18 ) aals .
about 25% of bright quasars contain aals with rest equivalent widths ( rews ) of @xmath190.3 , and around 10% of quasars have bals @xcite
. a higher incidence of aals or bals may occur in the quasars with higher sfrs if these quasars have more recently experienced an interaction and/or there is a progression in quasar outflow characteristics with time . in this paper , we present an exploratory observational study examining whether sfr in the host galaxies correlates with metallicity in the near - quasar environment .
we measure metallicity in the quasar blr from the rest frame ultraviolet ( uv ) spectrum and estimate galactic sfrs from the fir luminosities .
the data and analysis are described in [
sec : data ] and [ sec : anal ] .
the results and discussion are presented in [ sec : disc ] , with a brief summary in [ sec : sum ] .
we adopt cosmological parameters h@xmath20 mpc@xmath7 , @xmath21 , and @xmath22 throughout this work .
we select quasars for this study from the sample observed with mambo at iram at 1.2 mm by @xcite and @xcite and scuba at jcmt at 850@xmath23 by @xcite , @xcite and @xcite and all compiled by @xcite .
the quasars all were selected to be optically bright with absolute b - band magnitude m@xmath24 for the carilli et al .
objects , m@xmath25 for the omont et al . objects and m@xmath26 . ] for the @xcite and @xcite objects . @xcite observed a representative sample of 41 out of the more than 100 quasars with z @xmath27 3.6 found as part of the sloan digital sky survey ( sdss ) galactic cap and southern equatorial stripe survey .
@xcite observed a random selection of 97 radio quiet , optically luminous sources from the multicolour palomar digital sky survey available from g. djorgovski s web page with 3.9 @xmath28 z @xmath28 4.5 and the @xcite catalogue with 1.8 @xmath28 z @xmath28 2.8 .
@xcite selected a small sample of 6 bright @xmath29 radio quiet quasars from the apm survey @xcite .
@xcite selected a larger sample of the 76 most uv luminous @xmath30 radio quiet quasars known at the time of observation , and @xcite selected a complimentary sample of 57 @xmath31 quasars from various large surveys .
we cross - reference this sample with the optical quasar spectra in the sdss data - release 6 , finding 116 objects with available spectra .
the sdss spectra from data release 6 have resolution r @xmath322000 and wavelength coverage @xmath33 @xcite .
we consider only those objects with redshifts between 2.17 and 4.75 , compatible with sdss spectral coverage of the full 1200 1600 rest - frame wavelength range .
spectra missing regions within the specified wavelength range are further excluded from the analysis .
five spectra with very low signal to noise ratios ( s / n ) also are rejected .
the final sample consists of 34 optical sdss spectra with a range of sub - mm brightnesses and a redshift range of 2.2 @xmath34 4.6 . [ cols="<,<,^,^,^,^,^,^ " , ]
to make comparisons between different l@xmath35 , we create one composite spectrum from the sdss spectra comprising each of the three fir bins in the sample described in
[ sec : data ] . to create the composites ,
we begin by shifting each spectrum to its rest wavelength using the redshifts provided by sdss @xcite .
we manually inspect each individual spectrum and remove absorption features by interpolating across the affected regions .
the presence of narrow absorption lines and their relationship to l@xmath35 is discussed in [ subsec : aalanal ] below .
then we average together the spectra in each bin to create the final composites . to ensure that no single spectrum is dominating the composite we also compute the median spectra for each bin , compare them to the average spectra used throughout the rest of the analysis , and confirm that the two composite types are well - matched .
our analysis is limited to the spectral region between ly@xmath36 1216 and 1550 .
these limits are imposed by the incomplete spectral coverage at longer wavelengths and by suppression in the ly@xmath36 forest starting in the blue wing of the ly@xmath36 emission line and extending to shorter wavelengths . before normalising the composite spectra
, we visually compare the continuum slopes of the three composites .
we find no significant trend for reddening with fir luminosity in the continuum slopes of the composite spectra .
differences between the composite slopes are negligible compared to the dispersion of slopes among the individual quasars making up each composite .
we fit a power - law continuum to the fir - intermediate composite spectrum using wavelength regions devoid of absorption or emission , 25 wide and centred at @xmath81460 , and @xmath81770 , following @xcite .
unfortunately , the fir - faint and fir - bright composites have incomplete spectral coverage past @xmath9@xmath81700 , so we fit the continuum using the available wavelength region around @xmath81460 , plus the region from @xmath81335 - 1355 , which is also used by @xcite , and which we determine by visual inspection to be devoid of absorption or any detectable emission . the normalised average composite spectra
are shown in figure [ fig : comp ] .
we measure several quasar broad emission line fluxes to calculate line flux ratios , which we then use to estimate the gas phase metallicity in the near - quasar environment .
flux ratios are measured for the emission lines that have been shown by e.g. @xcite , to give good abundance estimates within the limited wavelength range from @xmath81216 to @xmath81600 : @xmath81240/ @xmath81549 and @xmath81397 + @xmath81402/ @xmath81549 .
these emission lines are labelled in figure [ fig : comp ] .
other ratios such as n iii ] @xmath81750/o iii ] @xmath81664 , @xmath81240/he ii @xmath81640 or n iii @xmath8991/c iii @xmath8977 are too weak given the s / n in the spectrum and/or they fall in a poorly characterised region of the spectrum . the measured line flux ratios are listed in columns 5 and 7 of table [ tab : abs ] . because the emission is strongly blended with the ly@xmath36 emission , which is itself significantly degraded by absorption in the ly@xmath36 forest ,
is the only strong , non - blended emission line in the spectra .
thus , the emission lines are fit with gaussians and used as templates for the other emission lines .
the fits to and the scaled fits to ly@xmath36 and are shown for the three composites in figure [ fig : gauss ] .
the fits are performed using gatorplot , an idl program written by c. warner , and are carried out using the smallest number of gaussians possible to provide a good match to the data : one broad and one narrow component for each of the two doublet lines at @xmath81548 and @xmath81551 .
the two gaussians comprising the @xmath81548 emission line fit are held at fixed full width at half maximum ( fwhm ) and are scaled in total flux at the fixed central wavelength of the ly@xmath36 @xmath81216 and @xmath81239 , 1243 doublet emission to match these line strengths . because is known to be blue - shifted relative to ly@xmath36 and the nominal quasar redshift by @xmath37 310 km s@xmath7 @xcite , the locations of the and ly@xmath36 centroids are shifted from their laboratory values by this amount relative to . to accurately determine the flux
, the red ly@xmath36 wing beneath the emission must be characterised . to do this
precisely , we scale the fit to align with the non - absorbed region of ly@xmath36 , which in some cases is quite small ( @xmath38 ) .
the scaled fit may even rise above the data where heavy ly@xmath36 forest absorption has eaten into the emission , as is the case for all of the three composites , or miss the peak of ly@xmath36 as is the case in the fir faint composite ; both cases are clearly shown in figure [ fig : gauss ] . in order to better match the ly@xmath36 emission peak in the fir faint composite
, we allow the ly@xmath36 gaussian to be @xmath39 narrower than the gaussian .
narrowing the gaussian for the ly@xmath36 emission line does not have a noticeable effect on the resulting strength .
regardless , a precise match to the ly@xmath36 emission line is not required for this analysis , and does not effect the final emission results ( see e.g. @xcite for further discussion ) .
the and doublets are fit at their respective fixed separations and with a 3:2 intensity ratio ( halfway between the allowed 2:1 and 1:1 intensity ratios ) for the shorter and longer wavelength respectively , as in @xcite .
possible broadening in is predicted if the emission forms at a larger distance from the quasar than the emission . @xcite and @xcite find that forms twice as close to the quasar as , so if the line widths are controlled by virial motions , could be up to @xmath40 times broader than .
ly@xmath36 , on the other hand , forms at about the same distance from the quasar as and should be no broader ( and probably narrower ) than @xcite .
therefore , the profiles we adopt , which have the same fwhm s as , may be narrower than the actual fwhm s , while the ly@xmath36 fwhm could be slightly broader .
we estimate the uncertainties in the line strengths due to uncertainty in the continuum placement by repeating the line fits with different reasonable continuum heights .
we also include estimates of line strength uncertainties due to blending with ly@xmath36 and the poorly defined fwhm by repeating the line fits with a range of different and ly@xmath36 fwhm s .
the overall uncertainties , combining these effects , are less than a factor of 1.5 , so that the measurement errors in the / ratios are also less than a factor of 1.5 . , and in the left panels and in the right panels for each normalised composite spectrum .
the vertical scale is normalised flux .
the top panels shows the fir faint spectrum , the middle panels shows the fir intermediate spectrum and the fir bright spectrum is shown in the bottom panels .
the solid curves are the data and total gaussian fits , the dashed curves show the individual gaussian fits for each emission line .
the and features each have two gaussians , one for each doublet member , which are each composed of a narrow and a wide gaussian ( not shown ) .
the error spectra for each region are shown as dot dashed lines . ]
the + emission line is weaker than the and emission lines in the composite spectra , however it also is isolated from other emission .
this isolation means that measuring the equivalent width of the emission without fitting , using iraf , produces accurate fluxes .
we confirm the accuracy of the equivalent width measurements by comparing the previously measured fluxes from gaussian fits to to the fluxes from equivalent widths , and find compatible results .
we measure + and equivalent widths to calculate the + / line flux ratios listed in table [ tab : abs ] . to estimate uncertainties in the + line strengths , we vary the equivalent width measurements by using a range of continuum heights and wavelength cutoffs for the edge of the emission lines .
the overall uncertainties in the + line strengths are less than a factor of 1.2 , so that the measurement errors in the + / ratios are also less than a factor of 1.2 .
we convert the / flux ratios into metallicities using the theoretical relationship in which nitrogen abundance increases relative to carbon as metallicity increases because of secondary enrichment ( cno nucleosynthesis ) processes , and determine the average metallicity for each l@xmath35 composite @xcite .
the correlation is characterised by @xcite with recent updates to the solar abundance ratios by @xcite .
we find supersolar metallicities for all three composite spectra , as listed in column 6 of table [ tab : abs ] .
similar characterisations for the + / flux ratio - metallicity correlation are performed by @xcite , using the older solar abundances also used by @xcite .
we apply the latest corrections for the solar abundance ratios for this characterisation and determine that the metallicities are also supersolar according to this ratio , as shown in column 8 of table [ tab : abs ] .
there is no significant trend among the three composites in the line ratios , based both on our measurements of these ratios and on a visual inspection of the stacked composites in figure [ fig : comp ] , and correspondingly there is no significant trend in metallicity .
the average metallicities across the three composites are [email protected] z@xmath5 for the / ratio and 4.2 z@xmath5 for the + / ratio .
we note that the metallicities inferred from the two line ratios can differ by as much as a factor of @xmath92 in the same spectrum .
this is typical of the dispersion found between different line ratios in other studies , and it might be a good indication of the theoretical uncertainties @xcite .
we compare our metallicities to the metallicities in the l@xmath41 range 10@xmath42 to 10@xmath43 erg s@xmath7 sampled by a large emission line study by @xcite , who find metallicities of z @xmath93 - 6 z@xmath5 , broadly consistent with the metallicities found in this sample .
we also note that the results based on /(and /he ii ) tend to be higher than other line ratios , and thus our best guess at the metallicity in the current sample overall would be near the lower end of the range 49 z@xmath5 @xcite .
the _ relative _ metallicities between the three composites , which smooth over object - to - object scatter , are robust for differences greater than a factor of 2 - 3 , based on the uncertainties in the line ratio measurements plus the theoretical uncertainties , and therefore useful for spotting strong metallicity trends among the different l@xmath35 bins .
we inspect each sdss spectrum used to make the composites and count the number of aals with rew @xmath44 per spectrum .
we find that 17/34 quasars have at least one aal and 6/34 have absorption too broad to be classified as an aal , yet too narrow to be classified as a bal , with widths more than twice the thermal width and velocity dispersion expected for unrelated gas in the line of sight , which is indicative of gas forming in an outflow @xcite .
these statistics are broadly consistent with previous work , given the small numbers involved in this study @xcite .
aals and outflow lines appear in each of the three l@xmath35 bins , with no significant differences in their occurrence fractions among the bins that would indicate a trend with l@xmath35 , particularly given the small number of objects in each bin .
the average line flux ratios / @xmath45 and + / @xmath46 across the three composites , as shown in columns 5 and 7 of table [ tab : abs ] , correspond to average metallicities of z @xmath9 9.5 and 4.2 z@xmath5 respectively , shown in columns 6 and 8 of table [ tab : abs ] .
there is no significant trend in metallicity with l@xmath35 in this sample .
we note that the fir bright bin has the most luminous average l@xmath41 , while the fir faint and intermediate bins have roughly equal , and less luminous average l@xmath41 .
this difference in l@xmath41 could appear in the relative metallicities among the bins because more luminous quasars tend to be more metal - rich than less luminous quasars @xcite .
however , the absence of a corresponding gradient in the metallicity is not surprising given the relatively high average of the total l@xmath41 , the range in average l@xmath41 among the composites of less than 0.4 dex ( table [ tab : abs ] ) , which corresponds to a range in metallicity of no more than 0.2 dex in @xcite , and the sensitivity of the data to trends in metallicity greater than @xmath90.3 dex between the three l@xmath35 bins .
the fact that the data show no significant metallicity evolution with changing sfr ( l@xmath35 ) could be due to the small sample sizes being affected by object - to - object scatter or by the small dynamic range in l@xmath35 , with values that are all within the luminosity / sfr range of powerful ulirgs .
absorption features such as aals and bals and reddened continua may be more numerous among quasar environments that have undergone a recent merger or are participating in the blowout of gas that is revealing the visible quasar @xcite .
although these are optically selected quasars chosen to be bright in the rest - frame uv and strongly reddened sources would be excluded , there could still be a more subtle trend in sub - samples by l@xmath35 .
we do not find a trend in the number of aals per quasar or in the continuum slope per l@xmath35 bin .
each bin appears to have roughly the same percentage of quasars with aals .
the absence of a trend could be due simply to the small numbers in our sample . or ,
if the consistency across bins is real , it could indicate that there is no progression in quasar outflow characteristics with changing sfrs .
a larger sample size is needed to resolve this question .
the overall conclusion from this and other studies is that quasars appear to be metal - rich at all redshifts , with only small variations due to smbh mass and luminosity dependencies @xcite .
the consistently high metallicities suggest that the star formation producing these fir - bright quasars is not the star formation that determines the metallicity of the blr . instead
, the blr gas must have been enriched prior to these fir - producing star formation episodes , possibly shortly after a merger or some other star formation trigger when the starburst first began .
there are no significant differences among the l@xmath35 bins in the outflow fraction or in the amount of debris leftover from mergers in the form of absorption , which might correlate with l@xmath35 if this is indeed an evolutionary sequence .
this scenario is consistent with models of galaxy and quasar evolution , which predict a lack of metallicity evolution .
the models also find that quasar activity , though caused by a disruption of some sort , is significantly delayed after the disruption event and most of the star formation and enrichment in the host is complete even as the quasar phase first begins to emerge @xcite .
the fir - bright quasars may be in some sense less mature than the fir - faint quasars , however , this is not manifest in significant metallicity evolution over the timescales probed by this sample , which are shorter than a quasar lifetime
. the reality of quasar - host galaxy formation could also be more complicated than any simple monotonic sequence in which high sfrs correlate cleanly with lower metallicities and an earlier stage of evolution .
@xcite suggest a ` softer ' ulirg - quasar evolution paradigm with more scatter in the path of an individual galaxy s evolution from ulirg to quasar .
they measure neon abundances of [email protected] z@xmath10 in the nuclear regions of ulirgs .
if these high metallicities are ubiquitous in ulirgs , we will never observe metal - poor quasars , regardless of their stage of evolution .
other plausible scenarios suggest that agn and star formation activity in galaxies may be episodic in nature @xcite .
thus , as a quasar emerges at the tail end of a star - forming phase , there is no guarantee that its blr metallicity is linked to this recent and/or ongoing star formation phase , because several previous episodes of star formation and agn activity could already have occurred and enriched the blr gas to supersolar values .
an alternative interpretation is provided by the works of @xcite and @xcite .
they postulate a scenario where a star - forming galaxy could quickly reach an equilibrium metallicity early in its formation history .
higher mass galaxies would have higher abundances , but for a given galaxy , the abundance would not change significantly after an initial balance of in - flowing metal - poor and out - flowing metal - rich gas was reached .
the lack of blr metallicity evolution with star formation rate found in this study is consistent with these objects having previously reached an equilibrium metallicity , unaffected by the observed ongoing star formation .
we create composite rest - frame uv spectra for a sample of 34 quasars observed in the fir and measure / and + / flux ratios .
we convert these flux ratios into a metallicity for each of three l@xmath35 bins , which we interpret as sfr bins .
we find that all three bins have supersolar metallicities of the order 4 to 9 times solar , and find no metallicity trend with l@xmath35 .
we investigate the amount of outflowing gas , as manifest by both aals and somewhat broader absorption lines , in each l@xmath35 bin and find no difference in number between the l@xmath35 bins , neither do we find a change in continuum slope with changing l@xmath35 , as would be consistent with more dusty star formation in hosts with higher sfrs .
together , these results suggest that the ongoing star formation in the host is not responsible for the metal enrichment of the blr gas . instead
, the blr gas must have been enriched before the visible quasar phase .
we note that we require a substantially larger sample size with a broader range in l@xmath35 to confirm that these results are not affected by unusual individual objects or biased by the narrow l@xmath35 range .
d. , kennicutt r. c. , engelbracht c. w. , leitherer c. , draine b. t. , kewley l. , moustakas j. , sosey m. , dale d. a. , gordon k. d. , helou g. x. , hollenbach d. j. , armus l. , bendo g. , bot c. , buckalew b. e. a. , 2007 , apj , 666 , 870 k. e. k. , smail i. , alexander d. m. , weiss a. , walter f. , swinbank a. m. , greve t. r. , kovacs a. , de breuck c. , dickinson m. , ibar e. , ivison r. j. , reddy n. , spinrad h. , stern d. , brandt w. n. , chapman s. c. e. a. , 2009 , mnras , 395 , 1905 k. , bender r. , bower g. , dressler a. , faber s. m. , filippenko a. v. , green r. , grillmair c. , ho l. c. , kormendy j. , lauer t. r. , magorrian j. , pinkney j. , richstone d. , tremaine s. , 2000 , apj , 539 , l13 f. , simon l. , 2009 , in cunha k. , spite m. , barbuy b. , eds , chemical abundances in the universe : connecting first stars to planets proceedings iau symposium no . 265 , quasar metal abundances & host galaxy evolution f. , warner c. , dietrich m. , ferland g. , 2007 , in ho l. c. , wang j .- w .
, eds , the central engine of active galactic nuclei vol .
373 of astronomical society of the pacific conference series , quasar metal abundances and host galaxy evolution .
pp 653+ t. , eracleous m. , charlton j. , ganguly r. , tytler d. , kirkman d. , omeara j. , suzuki n. , lubin d. , 2003 , in bulletin of the american astronomical society vol .
35 of bulletin of the american astronomical society , a census of intrinsic narrow absorption lines at z @xmath373.0 .
pp 1329+ m. , lutz d. , sturm e. , contursi a. , tacconi l. j. , lehnert m. d. , dasyra k. m. , genzel r. , veilleux s. , rupke d. , kim d .- c . , baker a. j. , netzer h. , sternberg a. , mazzarella j. , lord s. , 2006 , apj , 649 , 79 m. , frieman j. , bernardi m. , loveday j. , nichol b. , castander f. , meiksin a. , 2002 , in starck j .- l . , murtagh f. d. , eds , society of photo - optical instrumentation engineers ( spie ) conference series vol .
4847 of society of photo - optical instrumentation engineers ( spie ) conference series , the sloan digital sky survey 1-dimensional spectroscopic pipeline .
pp 452460 s. , gebhardt k. , bender r. , bower g. , dressler a. , faber s. m. , filippenko a. v. , green r. , grillmair c. , ho l. c. , kormendy j. , lauer t. r. , magorrian j. , pinkney j. , richstone d. , 2002 , apj , 574 , 740 s. , rupke d. s. n. , kim d. , genzel r. , sturm e. , lutz d. , contursi a. , schweitzer m. , tacconi l. j. , netzer h. , sternberg a. , mihos j. c. , baker a. j. , mazzarella j. m. , lord s. , sanders d. b. , stockton a. , joseph r. d. , barnes j. e. , 2009 , apjs , 182 , 628 | we present the results of an exploratory study of broad line region ( blr ) metallicity in 34 2.2 @xmath0 z @xmath0 4.6 quasars with far - infrared ( fir ) luminosities ( l@xmath1 ) from 10@xmath2 to @xmath310@xmath4 l@xmath5 .
quasar samples sorted by l@xmath1 might represent an evolutionary sequence if the star formation rates ( sfrs ) in quasar hosts generally diminish across quasar lifetimes .
we use rest - frame ultraviolet spectra from the sloan digital sky survey to construct three composite spectra sorted by l@xmath1 , corresponding to average sfrs of 4980 , 2130 and @xmath6 m@xmath5 yr@xmath7 after correcting for a nominal quasar fir contribution .
the measured @xmath81240/ @xmath81550 and @xmath81397 + @xmath81402/ @xmath81550 emission line ratios indicate super - solar blr metallicities in all three composites , with no evidence for a trend with the star formation rate .
the formal derived metallicities , z@xmath9 59 z@xmath10 , are similar to those derived for the blrs of other quasars at similar redshifts and luminosities .
these results suggest that the ongoing star formation in the host is not responsible for the metal enrichment of the blr gas . instead
, the blr gas must have been enriched before the visible quasar phase .
these results for high quasar metallicities , regardless of l@xmath1 , are consistent with evolution scenarios wherein visibly bright quasars appear after the main episode(s ) of star formation and metal enrichment in the host galaxies . finally
, young quasars , those more closely associated with a recent merger or a blowout of gas and dust , may exhibit tracers of these events , such as redder continuum slopes and higher incidence of narrow absorption lines . with the caveat of small sample sizes , we find no relation between l@xmath1 and the reddening or the incidence of absorption lines .
[ firstpage ] quasars : general quasars : emission lines galaxies : evolution |
measuring of the structure functions is an unique tool for the study of the nucleon internal structure - together with the quark - parton model ( qpm ) giving the relations between the structure functions and the parton distributions , which represent a final , detailed picture of the nucleon . in this sense , these relations , obtained under definite assumptions , are extraordinary important , since the distribution functions themselves are not directly measurable . at the same time , the standard , simple formulas relating the structure and distribution functions are ordinarily considered so self - evident , that in some statements , the both are identified . the experiments dedicated to the deep inelastic scattering ( dis ) , are oriented to the measuring either _ unpolarized _ or _ polarized _ structure functions . the results on the unpolarized functions are compatible with our expectations based on the qpm and qcd , but the situation for the polarized functions is much more complicated . until now ,
it is not well understood , why the integral of the proton spin structure function @xmath1 is substantially less , than expected from very natural but possibly equally naive assumption , that the nucleon spin is created by the valence quarks .
presently , there is a strong tendency to explain the missing part of the nucleon spin as a result of the considerable contributions of the sea quarks ( particularly strange quarks ) and the gluons . nevertheless , a consistent explanation of the underlying mechanism is still missing . during the last years , the hundreds of papers have been devoted to the nucleon spin structure , for the present status see e.g. @xcite , @xcite , the comprehensive overviews @xcite,@xcite and citations therein . in the present paper we continue
our discussion started in @xcite , @xcite , where we have shown , that the standard formulation of the qpm , conceptually firmly connected with the infinite momentum frame ( imf ) , oversimplifies the parton kinematics .
in @xcite we demonstrated that the effect of oversimplified kinematics in imf can have an impact particularly on the spin structure function @xmath1 , or more exactly , it can substantially modify the relation between the distribution and spin structure functions .
we have suggested this effect can be a source of the discrepancy between the experimentally measured and naturally expected magnitude of the spin function @xmath1 .
the primary aim of this paper is to precise just this point .
the paper is organized as follows . in the following section
we add some comments to our master equations connecting the structure and distribution functions and point out the distinction between them and these used in the standard approach . in the sec .
we propose the model , in which the internal motion of quarks is consistently taken into account .
further , in contradistinction to the standard treatment based on the qcd evolution of the distribution functions in dependence on @xmath2 , our model rather aims at describing the part of distribution and structure functions , which are not calculable in terms of the pertubative qcd . in the sec . [ sec4 ] .
the results of the model on the polarized and unpolarized proton structure functions are compared with the experimental data and some free parameters are fixed .
some important additional comments on the model and obtained results are done in the sec .
the last section is devoted to the overall summary and concluding remarks .
since this paper should be read together with @xcite , @xcite , for convenience we refer to the equations or figures in these papers simply with prefixes p , q . for completeness ,
in the appendix we give a more rigorous explanation , that the approximation of the term @xmath3 entering the argument of the @xmath4functions in some integrals in this and two the previous papers is for our purpose fully acceptable .
in the preceding discussion @xcite,@xcite we have shown [ see eqs .
( p3.41 ) , ( q2.1 ) ] , that if one assumes momenta distributions ( q2.2 ) , ( q2.3 ) of quasi - free quarks having mass @xmath5 are spherically symmetric in the nucleon rest frame , then the corresponding structure functions @xmath6obey the master equation@xmath7 + a(p_\alpha q_\beta + p_\beta q_\alpha ) + bq_\alpha q_\beta\ ] ] @xmath8 @xmath9 simultaneously , we have shown , how the explicit solution of this equation looks like . in this solution , the mass @xmath10 formally appears as a free parameter .
let us make a remark , in the same way a similar equation can be obtained and solved also for the set of the neutrino structure functions , nevertheless in this paper we shall deal only with the electromagnetic ones .
let us note , despite of the fact that eq .
( [ ms21 ] ) is assembled for quark momenta distributions @xmath11 in the nucleon rest frame , the equation is relativistically covariant .
its manifestly covariant form follows immediately from ( [ ms21 ] ) after the substitution @xmath12 for moving nucleon we have @xmath13 and @xmath14 which means , that the phase space of the subset of quarks with @xmath15 fixed , represented by the sphere @xmath16 in the nucleon rest frame , is in a boosted system correctly represented by the ellipsoid with the shape defined by the lorentz transform ( [ ms24 ] ) .
further , the eq .
( [ ms21 ] ) involve also the quark polarization vector @xmath17 , which as follows from @xcite , plays the crucial role for the evaluation of the spin structure functions @xmath18 so here we consider desirable to derive its form in a more rigorous way , than we suggested for eq . ( q2.12 ) .
generally , quark polarization vector should be constructed from the proton momentum @xmath19 , proton polarization vector @xmath20 and the quark momentum @xmath21 : @xmath22 where @xmath23 are invariant functions of @xmath24 then contracting of eq .
( q2.5 ) with @xmath25 and @xmath26 and using relations ( q2.4 ) give the equations @xmath27
@xmath28 @xmath29 elimination of @xmath30 gives @xmath31 and since @xmath32 are independent , @xmath33 must be zero .
the remaining invariants @xmath34 follow from eq . (
q2.4 ) , which implies @xmath35 and solution of these equations reads @xmath36 so the quark polarization vector has the form @xmath37 contributions of both possible solutions [ sign @xmath38 means that quark spin is parallel ( antiparallel ) to the proton spin in its rest frame ] are in our calculation taken into account by the difference in eq .
. apparently , for the proton rest frame and polarization @xmath39 the last equation is identical to eq . ( q2.12 ) .
further , obtained invariants @xmath23 now give spin structure functions from eqs .
( [ cr2 ] ) , ( [ cr3 ] ) in covariant form @xmath40 @xmath41 apparently , according to these relations the structure functions can depend also on mutual orientation of @xmath20 and @xmath42 of course , this dependence is more complicated , apart the factor @xmath43 ahead of the integrals , integration involves the terms @xmath44 and @xmath45 .
this question is being studied and will be discussed in a separate paper . in the present paper
our considerations will be based on the results obtained in @xcite , which follow from eqs .
( [ cr9 ] ) , ( [ cr10 ] ) applied in the proton rest frame for the longitudinal polarization @xmath39 .
obviously , for this case the last two equations are equivalent to eqs .
( q2.13 ) , ( q2.14 ) . the scheme based on the eqs .
( [ ms21 ] ) and ( [ cr8 ] ) with all their implications suggested in @xcite,@xcite can be a priori valid for quasi - free quarks ( on mass shell ) filling up the nucleon volume . in this sense the scheme represents a covariant formulation of the naive qpm .
we have shown that eq .
( [ ms21 ] ) in which the quark internal motion is consistently taken into account imply relations between the structure and distribution functions different from those obtained in the standard procedure relying on the imf , which is based on the approximation @xmath46 ( parton momenta components are @xmath47 fractions of the proton momenta components ) .
in fact this approximation in the covariant formulation is equivalent to the assumption , that the partons are static with respect to the nucleon , therefore there are suppressed not only the transversal momenta , but also longitudinal ones .
of course , this consequence is somewhat obscured just in the imf , where all the relative motion is frozen , since all the processes run infinitely slowly - including the passing of the probing lepton through the nucleon .
let us remark , the standard relations ( e.g. @xmath48 ) obtained in the naive qpm with static quarks are currently applied even in the standard approach based on qcd improved qpm , which is not a consistent procedure , since it means that correct dynamics is combined with incorrect kinematics . in this way we have shown , that the relations between the structure and distribution functions can be , at least on the level of the naive qpm , strongly modified ( particularly for the polarized case ) by the parton internal motion .
this result can be instructive by itself .
the impact of the quark intrinsic motion on the function @xmath49 has been discussed also in some other approaches @xcite-@xcite and necessity of the covariant formulation for the spin structure functions has been pointed out in @xcite . however , the real nucleon is much more complex object , than just a bag of quasi - free fermions .
but we shall try to assume the following .
the relations obtained within the scheme suggested above can be used as a good approximation even for the interacting quarks , but provided that the term mass of quasi - free parton is substituted by the term parton effective mass. by this term we mean the mass , which a free parton would have to have to interact with the probing photon equally as the real , bounded one .
intuitively , this mass should correlate to @xmath2 : a lower @xmath2 allows more time and space for the struck parton to interact with some others , as a result the energy is transferred to a greater system than the parton itself . and on contrary , the higher @xmath2 should mediate interaction with more isolated parton .
moreover , we accept that the value of the effective mass even for a fixed @xmath2 can fluctuate - e.g. in a dependence on the actual qcd process accompanying the photon momentum transfer . this means , that the terms in the relations involving the mass of quasi - free parton @xmath10 will be substituted by their convolution with some mass distribution @xmath50@xmath51 in the following we shall propose a simple , but sufficiently general model for the unknown distributions @xmath52 in which all the dynamics of the system is absorbed . then , these distributions will be used as an input for the calculating of the corresponding structure functions .
construction of the model is based on the following assumptions and considerations : \1 ) parton distribution @xmath53 representing the number of quarks in the energy interval @xmath54 can be formally expressed : @xmath55 where @xmath56 is a probability that the nucleon is in the state with @xmath57 charged partons ( quarks + antiquarks ) of various flavors , and @xmath58 is the corresponding average one - parton distribution , which fulfills @xmath59 \2 ) nucleon consists of the three quarks and partons ( gluons + @xmath60 pairs ) mediating the interaction between them , as sketched in the fig .
[ fg1]_a _ , where the individual pictures represent some terms in the sum ( [ mss31 ] ) . the flavors and spins of all the quarks in each the picture
are mutually cancelled , up to the three quarks giving additively the corresponding nucleon quantum numbers .
these three quarks are in the figure marked by black and in our approach are identified with the _ valence quarks_. the reason , that such identification is quite sensible , is the following .
apparently , the sum ( [ mss31 ] ) can be split into quark and antiquark parts @xmath61 , then our valence term reads @xmath62 which in the @xmath63representation exactly corresponds to the current definition of the valence quarks . correspondingly , the unmarked quarks are identified with the _ sea quarks_. but both the kinds of quarks have the same energy distributions @xmath64 entering the eq .
( [ mss31 ] ) , in this sense they are completely equivalent . on the other hand , it is obvious , that for the valence quarks , in eq .
( [ mss31 ] ) only black quarks from the figure contribute , therefore if @xmath58 is assumed in the first approximation independent on the flavor , then @xmath65 \3 ) the quarks carry only part of the nucleon energy ( mass ) , @xmath66 where the factor @xmath67 equals roughly one half , the rest is carried by the gluons . in the first approximation
we shall assume this factor is valid also for any term in the sum ( [ mss31 ] ) , @xmath68 which in other words means the ratio of the total energies of quarks and gluons , together constituting the nucleon mass , is the same for all possible states sketched in fig .
[ fg1]_a_. \4 ) we assume all the quarks in the nucleon state @xmath57 have approximately the same effective mass @xmath69 and denote @xmath70 .
one can expect , for higher @xmath57 the parameter @xmath71 will drop and so the sum ( [ mss33 ] ) can be substituted by the integral @xmath72 obviously , eq . ( [ mss31 ] ) can be with the use of eq .
( [ mss35 ] ) rewritten in a similar way : @xmath73 where @xmath74 the physical meaning of the distributions @xmath75 is the following .
the distribution @xmath76 represents a probability , that the effective mass of the quark , which the probing lepton interacts with , is @xmath71 or alternatively , @xmath77 is the number of quarks in the interval @xmath78 , which the lepton has chance to interact with . on the other hand ,
the ( normalized ) distribution @xmath79 can be interpreted as a probability , that the exchanging photon distinguishes the quarks with the effective mass @xmath71 - as expressed by the pictures with different granularity in fig .
_ in this sense , each picture in the fig .
[ fg1]_a _ can be labeled by some @xmath71 , equally as the corresponding term @xmath80 in the integral ( [ mss37 ] ) .
obviously , at the same time the @xmath79 represents also the distribution of effective masses corresponding to the valence quark term . intuitively , the probability of different contributions in fig .
[ fg1]_a _ should depend also on _
_ @xmath2 ( higher @xmath81better resolution ) , so we expect @xmath82 in the next , we shall identify these distributions with that suggested on the end of the previous section .
\5 ) the relations ( p3.52 ) , ( p3.56 ) and ( p3.20 ) give the recipe how to obtain the structure function @xmath83 from a given energy distribution of the partons with some fixed value @xmath71 and charge @xmath84@xmath85 @xmath86 where the sign @xmath38 in the second relation refers to the region @xmath87 for the application of this procedure to eqs .
( [ mss37 ] ) , ( [ mss38 ] ) one has to weight the contributions integrated over @xmath71 by the corresponding ( mean ) charge squared .
the charge weight of the valence quarks is the constant @xmath88 = \frac 13\ ] ] for the proton and similarly for the neutron , @xmath89 for the sea we assume in the first approximation the equilibrated mixture of the quarks @xmath90 , so @xmath91 = \frac 29.\ ] ] then for the nucleon with @xmath57 quarks we get @xmath92 or in terms of @xmath71 @xmath93 therefore , the energy distributions ( [ mss37 ] ) , ( [ mss38 ] ) generate the corresponding structure functions : @xmath94 @xmath95 \6 ) now , let us pay attention to the spin structure functions . according to
the concept suggested in item 2 ) , only valence quarks contribute to the nucleon spin .
first , we shall consider the spin functions generated by the valence quarks with some fixed effective mass @xmath96 then we shall easily proceed to the case with the distribution @xmath79 . in @xcite , for sake of simplicity
, we have assumed all the three valence quarks contribute to the proton spin equally [ eq . ( q2.7 ) ] . on the other hand
it is obvious the quark symmetry group can impose an extra constraint on the contributions of different quark flavors as it follows e.g. from the philosophy of the well known bjorken @xcite and ellis - jaffe @xcite sum rules based on the symmetries _
u(6 ) _ and _ su(3)_. in our consideration we shall not strictly assume any particular group of symmetry , but the different spin contributions of _ u- _ and _ d-_quarks will be expressed by a free parameter @xmath97 , having in the notation of eq .
( q2.7 ) , e.g. for the proton , the following sense @xmath98 where @xmath99 is the valence distribution @xmath100 which is not , due to different normalization , identical with the distribution @xmath101 , but the both are simply related @xmath102 in the same way , as the distributions @xmath103 in eq .
( p3.14 ) . in the case of proton , there are the particular cases :
\a ) @xmath104 corresponds to the mutual spin orientation of the three valence quarks @xmath105 \b ) @xmath106 corresponds to the oversimplified scenario studied in @xcite , assuming the equal contribution of all the three quarks ; @xmath107 \c ) @xmath108 corresponds to the non - relativistic _
su(6 ) _ approach . from the wave function@xmath109 @xmath110 one
can easily show the mean value of the spin carried by the @xmath111 quarks is @xmath112 , i.e. @xmath113 , which agrees with @xmath108 in eq .
( [ ms31 ] ) .
\d ) @xmath114 corresponds to the mutual orientation of the three quarks @xmath115 so , the proton spin function @xmath116 entering the master equation ( [ ms21 ] ) and expressed in terms of the functions ( [ ms31 ] ) reads @xmath117 assuming the neutron is isospin symmetric , its corresponding spin function will be @xmath118 therefore the corresponding equations for the nucleon spin structure functions read @xmath119 @xmath120 where , in accordance with eqs . ( q2.17 ) , ( q2.18 ) @xmath121 @xmath122 the function @xmath123 can be with the use of eqs .
( q2.24),(q2.27 ) expanded @xmath124 since eq .
( p3.56 ) implies @xmath125 one can easily show the sum in eq .
( [ ms324 ] ) gives @xmath126 similar manipulation with the function @xmath127 gives the result @xmath128 obviously , for the case with the distribution @xmath129 the corresponding spin structure functions read @xmath130 let us note , the structure functions @xmath131 are not independent , all of them are in the corresponding way generated by the distributions @xmath132 and @xmath133 ( or , equivalently by @xmath134 ) .
\7 ) now , to make the construction suggested above applicable for some quantitative comparison with the experimental data , we have to use some reasonable , simple and sufficiently flexible parameterization for the unknown functions @xmath132 and @xmath133 .
we suggest the following .
a ) _ _ normalized distribution @xmath132 is assumed in the form @xmath135 where the @xmath136dependence is involved in the parameters @xmath137 b ) _ _ for the function @xmath138 we shall use the parameterization suggested in eqs .
( q2.34)-(q2.38 ) @xmath139 ^{\alpha /2x_0},\ ] ] where the upper ( lower ) sign in the first relation refers to the region @xmath140 and @xmath141 ^{-1},\ ] ] which follows e.g. from eqs .
( q2.34 ) , ( p3.25 ) . now , apparently one has to accept the parameter @xmath142 depends on @xmath71 as well .
let us consider the following .
sequence of the pictures in fig .
[ fg1]_a _ can be understood as the pictures of the one and the same nucleon , but
seen with different resolutions as outlined in fig .
[ fg1]_b_. then , it is natural to assume the four - momentum @xmath19 of the parton from some picture can be obtained from four - momenta @xmath143 of @xmath144 partons in a picture more rightwards , representing the parton seen with better resolution
: @xmath145 obviously , the mean values satisfy @xmath146 where @xmath147 equals @xmath148 for the extreme case , when the motion of the partons in the corresponding subset is completely uncorrelated ( correlated ) .
the last relations imply the effective masses and kinetic energies obey @xmath149 which means the quantity @xmath150 is a non - decreasing function of @xmath71 . in this moment
we know nothing more about this function , in the next section we shall show a reasonable agreement with the experimental data can be obtained with @xmath151 since we parameterize the function @xmath152 rather than the function @xmath134 , it will be useful the quantity @xmath153 defined in eq .
( [ mss38 ] ) and afterwards entering the important eq .
( [ mss316 ] ) , to express also in terms of @xmath152 . obviously , using eqs .
( [ mss310 ] ) and ( [ ms330 ] ) one gets @xmath154
@xmath155 now , we can our results shortly summarize .
if there are given some values of the free parameters @xmath156 , then the corresponding proton and neutron structure functions can be directly calculated according to eqs .
( [ mss315 ] ) , ( [ mss316 ] ) , ( [ ms328 ] ) , where the distribution @xmath132 is given by eq . ( [ ms329 ] ) , the function @xmath157 by eq . ( [ ms336 ] ) with the use of eqs . ( [ ms330 ] ) , ( [ ms331 ] ) , ( [ ms335 ] ) and the functions @xmath158
are : @xmath159 @xmath160 @xmath161 where @xmath162 the last expression is calculated from eqs .
( [ ms330 ] ) , ( [ ms331 ] ) and ( [ ms335 ] ) with the use of the step function @xmath163 for @xmath164
now we shall try to compare our formulas for the structure functions with the existing data .
we shall not attempt to make a consistent , global fit of the free parameters based on some rigorous fitting procedure , but only show the set of optimal parameters obtained by their tentative varying on the computer by hand .
moreover , our constraint will be only agreement with the proton structure functions @xmath83 and @xmath1 .
it means that the parameter @xmath165 controlling asymmetry between the proton and neutron spin functions , must be somehow fixed before the fitting . for the first approximation we use the @xmath166 value , @xmath108 [ see item 6c ) in the previous section ] . for a comparison with @xmath83 we use the parameterizations of the world data suggested in @xcite and @xcite ,
both taken for @xmath167 .
the data for @xmath1 are taken over from the recent paper @xcite of the smc collaboration .
after some checking on the computer , the optimal set of the free parameters is considered : @xmath168 results of the calculation of the proton structure functions @xmath1 and @xmath83 with these parameters are shown in figs .
[ fg2 ] , [ fg3 ] together with the data .
let us remark , the experimental points for @xmath1 correspond to the values evolved in @xcite to @xmath167 . in the error bars all the quoted errors ( statistical , systematic and those due to uncertainty of qcd evolution ) are combined .
obviously , the agreement with the experimental data in both the figures can be considered very good , particularly if we take into account that our parameterization of the unknown distributions is perhaps the simplest possible and moreover , the parameters ( [ ms41 ] ) still may not be optimal .
now , having tuned the free parameters by the @xmath1 and @xmath83 , one can predict the remaining functions @xmath0 and @xmath169 .
the results are shown in figs .
[ fg4 ] , [ fg5 ] .
our @xmath170 surely does not contradict the experimental data @xcite , which are compatible with zero - with statistical errors substantially bigger , than the vertical range of the figure .
but instead of the data , the comparison is done with wandzura wilczek @xcite twist-2 term for @xmath171 , which is evaluated in @xcite from the corresponding @xmath1 .
it is obvious , that two completely different approaches give at least qualitatively very similar results .
the proton valence function @xmath169 in fig .
[ fg5 ] is compared with the corresponding combination of the distributions @xmath172 and @xmath173 obtained ( for @xmath174 ) in @xcite by the standard global analysis : @xmath175 apparently , the agreement can be considered good . nevertheless , let us note , a part of the difference between both the curves can be due to our simplification : @xmath176 . on the other hand ,
the distributions of the @xmath177- and @xmath178-valence quarks are not equal and enter the functions @xmath1 and @xmath169 with different weights , see eqs .
( [ ms33 ] ) , ( [ ms42 ] ) . at the same time , our prediction for @xmath169 is based on the combination valid for @xmath1@xmath179 which is related to @xmath169@xmath180 @xmath181 for our value @xmath108 .
the last combination is also shown in the fig . [ fg5 ] and apparently gives a slightly better agreement with our simplified @xmath182 in any case , one can note , that the two different procedures , the standard one ( uses input on @xmath183 + qcd ) and our ( uses input on @xmath184 + our model ) give a very similar picture of the function @xmath185 , which is not directly measurable .
let us make a few comments on the obtained results .
first of all , it should be pointed out , that our structure functions in figs .
[ fg2]-[fg5 ] are calculated on the basis of very simple parameterization of the unknown distributions @xmath76 and @xmath186 , but on the other hand it is essential , that the contributions from the individual components of the quark distribution correctly take into account the intrinsic quark motion , which is particularly important for the spin structure function . the effect of this motion on @xmath1 is demonstrated in fig . q1 and the fact , that we succeeded to achieve a good agreement with the data also in fig .
[ fg2 ] is just thanks to this effect . for a better insight ,
how our structure functions are generated , in the fig .
[ fg6 ] we have displayed the initial distribution function @xmath186 drawn for a few values @xmath96 together with the corresponding structure functions @xmath187 .
the complete structure functions are their superpositions - weighted in the corresponding way with the use of the distribution @xmath79 .
further , also some other assumptions of the model are possibly oversimplified , for a more precise calculation , at least some of them could be rightly modified - but at a price of introducing the additional free parameters .
for example , the constant @xmath188 [ see eq .
( [ mss312 ] ) ] should take into account some suppression of the @xmath189 quarks @xcite and probably should allow a weak dependence on @xmath71 . also for the constant @xmath67 [ see eq .
( [ mss34 ] ) ] some @xmath190dependence should be allowed . concerning this constant ,
let us make one more comment .
the standard global fit @xcite suggests ( at @xmath167 ) the quarks carry @xmath191 of the nucleon energy and our fitted value @xmath67 from the eq .
( [ mss34 ] ) is roughly @xmath192 .
this difference is mainly due to the different relations between the distribution and structure function in both the approaches , see eqs .
( p3.38 ) , ( p3.59 ) .
the second relation ( valid for a subset of quarks with effective mass @xmath71 ) , multiplied by @xmath193 and then integrated by parts gives @xmath194 which for the static quarks [ @xmath195 , see discussion after eq .
( p3.59 ) ] coincides with the standard relation
. nevertheless , generally both the relations imply different rate of the nucleon energy carried by quarks .
one can check numerically that for our @xmath196 in a dominant region of @xmath71 the term @xmath197 in the integral ( [ ms51 ] ) plays a minor role ( see also fig .
[ fg6 ] , positions of the maxima of @xmath83 s are above the corresponding @xmath71 , in particular for lower @xmath71 ) , so as a result we get @xmath198 of the standard estimation of the quark contribution to the nucleon energy .
this ratio agrees with the ratio obtained from the corresponding fits : @xmath199 in the previous section we mentioned the effect of different shape of the @xmath200 and @xmath201quark distributions .
also the result @xmath202 following from eq .
( [ ms34 ] ) is a consequence of the simplifying assumptions @xmath203 and @xmath108 .
a more general approach @xmath204 would give @xmath205 so @xmath204 implies @xmath206 .
in fact , the global fit analysis proves , that @xmath207 is slightly narrower than @xmath208 it means , considering very qualitatively , in accordance with the last equation , in the function @xmath209 the negative term should dominate for smaller @xmath47 , which does not contradict the experimental data .
obviously , a proper accounting for the difference @xmath204 into the model should enable to calculate consistently in a better approximation not only the proton and neutron structure functions @xmath210 , but also the neutrino structure functions .
apparently then one could make a super - global fit covering the both unpolarized and polarized dis data . as a result , the flavor - dependent quark distributions @xmath186 [ or equivalently @xmath211 together with the corresponding effective mass distributions and the parameter @xmath212 controlling the relative spin contribution of the @xmath213 and @xmath201quarks , could be obtained .
finally , let us point out , inclusion the spin structure function into the fit in our model enables to obtain some information about the distribution of the quark effective masses . within our approach there are two distributions , @xmath132 and @xmath214 , relevant for the description of the quark effective masses in the nucleon .
the extrapolation of our parameterization for the @xmath214 distribution with the use of the relations @xmath142 and ( [ mssa38]),([ms329]),([ms335]),([ms41 ] ) give for @xmath215@xmath216 which implies the extrapolated @xmath214 is not integrable in this limit .
on the other hand , the basic distribution @xmath132 , parameterized by eq .
( [ ms329 ] ) with the @xmath217 from the set ( [ ms41 ] ) and with the use of the known relation @xmath218 can give an estimate of the mean value : @xmath219 i.e. @xmath220 for @xmath167 . the corresponding kinetic term calculated as @xmath221 gives a similar number ( @xmath222 ) .
let us recall , these numbers are related to the valence quark term . the @xmath2-dependence is involved only in the distribution @xmath223 , i.e. in our parameterization ( [ ms329 ] ) only via the powers @xmath224 .
it follows , the structure functions , which enhance in a low@xmath225 region for increasing @xmath2 , must be generated by the distribution @xmath223 in which the mean effective mass @xmath226 drops for increasing @xmath2 - in the qualitative agreement with an intuitive expectation .
in the present paper , with the use of the results obtained in the preceding ones @xcite,@xcite , we proposed an alternative covariant formulation of the qpm .
the initial postulates of the standard and our approach are basically the same , despite of that the relations between the structure and distribution functions obtained in both the approaches are not identical .
it is due to the fact , that in the standard approach the intrinsic quark motion is effectively suppressed by the use of the approximation @xmath227 . on the other hand ,
we have shown the master equations can be solved without the use of the this approximation , so in the corresponding solution the quark intrinsic motion is consistently taken into account .
on the basis of the obtained relations ( a priori valid for the version of naive qpm - with non - static quarks on mass shell ) we propose the model , in which the distributions @xmath228 reflecting the parton dynamics are introduced with some free parameters . with the use of this model we calculated simultaneously
the proton structure functions @xmath131 , assuming only the valence quarks term contributes to the proton spin . by a comparison with the data on @xmath184 @xmath229 we fix the free parameters and then on the basis of these parameters the functions @xmath169 and @xmath0 are predicted .
we found out : \1 ) both the unpolarized structure functions are well reproduced by the model .
the comparison is done with the data on @xmath83 and with the @xmath169 obtained from the standard global analysis data .
\2 ) at the same time , the model well agrees with the data on @xmath1 .
the calculated @xmath0 does not contradict the existing experimental data as well , but since in the data there are still rather big statistical errors , it is hard to say more .
\3 ) analysis of the fixed parameters within our approach suggests : @xmath230 the quarks carry less the proton energy ( almost by the factor 3/4 ) , than estimated from the standard analysis .
@xmath231 the average effective mass related to the valence quark term can be roughly @xmath232 and a similar energy can be ascribed to the corresponding motion .
so on the end we can underline the proposed model offers apart of the other results also the consistent explanation , why the experimentally measured proton spin function @xmath1 is less , than it is predicted in the standard approach based on the qcd improved qpm .
in this appendix we shall explain in more detail , that approximation ( [ app0 ] ) is sufficiently correct for the purpose of this paper .
structure functions given by eqs .
( p3.52 ) , ( p3.47 ) and ( q2.13 ) , ( q2.14 ) are given by the integrals of the form @xmath233 in accordance with eqs .
( p2.8)-(p2.13 ) one can express @xmath234 and if one assumes @xmath235 then the second term can be neglected and after introducing the variables @xmath236 @xmath237 the integral ( [ app1 ] ) reads @xmath238 where @xmath239 obviously , our approximation ( [ app0 ] ) used in eqs .
( p3.53 ) and ( q2.15 ) corresponds to @xmath104 i.e. @xmath240 and now we shall try to estimate the effect of the finite @xmath2 in eq .
( [ app4 ] ) .
first , we shall integrate over @xmath241 using the rule @xmath242 which with ( just one @xmath241 ) root in the argument of the @xmath4function in eq .
( [ app4 ] ) gives @xmath243 where @xmath244 , \ ] ] @xmath245 .\ ] ] now the integral ( [ app4 ] ) can be expressed @xmath246 the upper limit of @xmath247 can be with the use of eq .
( p3.11 ) calculated from the equation @xmath248 which implies the equation @xmath249 where @xmath250 and correspondingly @xmath251 since one requires @xmath252 and obviously we have @xmath253 @xmath254 in our region @xmath255 so only the upper sign root is acceptable . for @xmath240 this root reads @xmath256 in accordance with eq .
( p2.19 ) . the integral ( [ app9 ] ) can be now expressed @xmath257 where @xmath258 now , using the rule for differentiation of the integral @xmath259 and taking into account our integrated function vanishes at the boundary of phase space , @xmath260 then the second term in the last equation equals zero and the integral ( [ app15 ] ) can be expanded @xmath261 where @xmath262 at the same time , one can write @xmath263 i.e. the @xmath264odd terms vanish , so the difference @xmath265 can be estimated by an expression of the magnitude of the second order in the parameter @xmath266 quantitatively , this magnitude depends on the actual shape of the function @xmath267 so let us now calculate the integral ( [ app15 ] ) for the functions @xmath268 in accordance with the parameterization used in this and the previous paper .
first of all , one has to replace the parameterized distribution function @xmath269 by the corresponding @xmath270 the equations ( q2.34 ) , ( q2.37 ) and ( q2.38 ) imply @xmath271 ^{\alpha /2x_0},\quad p_0=\frac m2\left ( 1+\frac{x_0 ^ 2}x\right ) , \ ] ] which gives @xmath272 normalization factor is calculated by integration @xmath273 the distribution @xmath274 is related to the distribution @xmath275 by eq .
( p3.14 ) , so we obtain @xmath276 where @xmath150 is for the numerical calculation parameterized in accordance with eqs .
( [ ms335 ] ) and ( [ ms41 ] ) .
now , the integrals corresponding to the functions @xmath268 could be calculated without approximation ( [ app0 ] ) . in an accordance with eqs .
( p3.47 ) , ( p3.52 ) , ( q2.17 ) , ( q2.18 ) one has to compute @xmath277 @xmath278 @xmath279 @xmath280 where @xmath281 is the factor given by eq .
( [ ms35 ] ) .
the last integrals are the special cases of the integral ( [ app1 ] ) , so they can be rewritten in the form of eq .
( [ app15 ] ) allowing an easy numerical calculation @xmath282 @xmath283 @xmath284 @xmath285 where quantities @xmath286 @xmath165 @xmath287 @xmath288 @xmath289 are given by eqs .
( [ app24 ] ) , ( [ app5 ] ) , ( [ appe7 ] ) , ( [ app8 ] ) , ( [ app13 ] ) respectively . these functions @xmath210 could replace the functions ( [ mss310 ] ) , ( [ ms35 ] ) used in the present paper . in the fig .
[ fg7 ] we have shown the results of numerical calculation of the functions ( [ app29 ] ) - ( [ app32 ] ) for @xmath290 which is equivalent to the approximation ( [ app0 ] ) and for @xmath291 , corresponding to @xmath292 and @xmath293 . the curves @xmath104 also correspond to the results of calculation in fig .
[ fg6 ] based on eqs .
( [ ms330 ] ) and ( [ ms337 ] ) - ( [ ms340 ] ) . the figures suggest convincingly , that the effect of finite @xmath2 even for its rather low value , can play some role in our integrals ( [ app29 ] ) - ( [ app32 ] ) only for the higher effective masses ( @xmath294 ) , which are in our approach evidently on the tail of the mass distribution @xmath79 .
it means , that the approximation ( [ app0 ] ) is for the purpose of our model fully acceptable . | the alternative to the standard formulation of the quark - parton model is proposed .
our relativistically covariant approach is based on the solution of the master equations relating the structure and distribution functions , which consistently takes into account the intrinsic quark motion - in contradistinction to the standard infinite momentum approach , in which this motion is latently suppressed .
the model well reproduces the experimental data on the both polarized and unpolarized structure functions , assuming that only the valence quarks term contributes to the nucleon spin .
it is shown , the combined analysis of the polarized and unpolarized data can give an information about the effective masses and intrinsic motion of the quarks inside the nucleon .
simultaneously , it is shown that the rate of the nucleon energy carried by the quarks can be less , than estimated from the standard approach . as an addition
, a prediction for the proton spin function @xmath0 is given .
= -0.5 cm = 0.0 cm = 21.5 cm |
young core - collapse supernova remnants ( snrs ) in the galaxy provide an unique opportunity to study fine details of ejecta from supernova ( sn ) explosions as well as those of the circumstellar medium ( csm ) produced over the final evolutionary stages of massive stars .
an obvious group of such young core - collapse snrs is o - rich snrs that show optical spectra featuring strong o and ne lines , with lines from lighter elements ( e.g. , he ) either weak or absent . since the ejecta from a progenitor and the swept - up csm in the o - rich snrs
are not completely mixed together yet , observation of the o - rich snrs is promising to study the ejecta and csm related to the explosion of a progenitor and its late - stage evolution .
[email protected] ( msh1154 ) , together with cassiopeia a and puppis a , forms a rare group of o - rich snrs in the galaxy .
the o - rich nature of [email protected] was discovered by the optical detection of fast - moving o - rich and ne - rich ejecta @xcite .
recent wide - field observations of the optical line in [email protected] have revealed that the ejecta velocities range from @xmath51400 to @xmath11700 and also that distant ejecta knots are located primarily along the north - south direction @xcite .
the dynamical center of [email protected] , based on the kinematics of the line emission from the ejecta , is roughly coincident with the geometrical center of the radio emission , but shows an apparent offset from its pulsar position discovered at the southeast ( * ? ? ?
* see below ) .
@xcite reported that the mid - infrared ( mir ) spectrum of the ejecta is dominated by ionic lines and a broad bump around 17 @xmath2 m using the @xmath6 space telescope ( @xmath6 ) .
they proposed that the latter is produced either by polycyclic aromatic hydrocarbons ( pahs ) along the line of sight or newly - formed dust within the ejecta . besides there is also an equatorial mir continuum component from the dust associated with the shocked csm partly overlapping with the ejecta emission .
a previous infrared ( ir ) study based on the data obtained with infrared astronomical satellite ( @xmath7 ) , on the other hand , identified enhanced far - ir ( fir ) emission in the southwestern part of the snr , suggesting that the source has encountered adjacent clouds @xcite . in radio , [email protected] mainly consists of bright emission from a central pulsar wind nebula ( pwn ) of @xmath3 4 ( in diameter ) and relatively fainter outer plateau emission of @xmath3 9 @xcite .
the plateau emission declines sharply outward , and there is no apparent surrounding radio shell structure .
the pulsar psr j11245916 , which is located @xmath346 from the dynamical center of the snr in the southeast direction , was discovered by radio timing @xcite , confirming the pwn nature of the central emission of the snr .
the pulsar and pwn were also identified in x - ray emission @xcite , where the pulsar has a nearby jet and a torus of @xmath3 5 scale @xcite .
the jet axis is slightly tilted to the northeast direction @xcite .
there are two clear differences between the soft and hard x - ray emission of [email protected] .
first , while the former is produced by shocked gas of normal composition distributed as equatorial filaments , indicating it is likely from csm , the latter is dominated by bright metallic lines from the ejecta @xcite .
next , the x - ray emission is harder in the northwestern part than the southeastern part , implying density variation or asymmetric sn explosion @xcite .
the distance to [email protected] , on the other hand , was estimated to be 6.2 kpc from h i absorption observations @xcite . in this paper
we present an extensive ir study of [email protected] using multi - band imaging and spectroscopic data obtained with the @xmath0 space telescope @xcite .
we describe the details of our @xmath0 observations and present the results of the observations in 2 and 3 , respectively . in
4 we discuss ir emission in [email protected] from circumstellar dust with a comparison to model calculations of dust destruction by a sn shock .
we also discuss the ir emission from the ejecta and we suggest that there is a negligible amount of dust associated with the ejecta .
then , we make a comparison to the results of @xmath6 obserations .
we finally discuss the sn explosion in [email protected] based on our @xmath0 observations followed by our conclusions in 5 .
the @xmath0 multi - band imaging observations of [email protected] were carried out using its six near - ir ( nir ) and mir bands in the 2.726.5 @xmath2 m range as well as four fir bands in the 50180 @xmath2 m range on 2007 january 17 and 19 .
the nir and mir images were obtained with infrared camera ( irc ) that has three simultaneously operating nir , mir - s , and mir - l channels .
the three channels have comparable field of views of @xmath8 .
the nir and mir - s channels share the same pointing direction , while that of the mir - l channel is 25 away .
the irc observations were conducted in the two - filter mode that produced images of two bands for each channel @xcite .
the total on - source integration times were 178 s for the nir ( n3 , n4 ) observations and 196 s for both the mir - s ( s7 , s11 ) and mir - l ( l15 , l24 ) observations .
the basic calibration and data reduction , including dark subtraction , linearity correction , distortion correction , flat fielding , image stacking , and absolute position determination were performed by the standard irc imaging data reduction pipeline ( version 20071017 ) .
tables [ tab_obssum ] and [ tab_obsimg ] present a journal of our @xmath0 observations , including the spectroscopic observations , and the basic parameters of the @xmath0 imaging bands , respectively . the fir images were obtained with far - infrared surveyor ( fis ) in two round - trip scans using the cross - scan shift mode @xcite .
the scan speed and length were 15 s@xmath9 and 240 , respectively , creating images of 40 @xmath10 12 size elongated along the scanning direction .
all the four fis band ( n60 , wide - s , wide - l , and n160 ) images were obtained simultaneously by a single observing run ( table [ tab_obssum ] ) .
the initial data calibration and reduction such as glitch detection , dark subtraction , flat fielding and flux calibration were processed with fis slow - scan toolkit ( version 20070914 ) , and the final image construction was performed with a refined sampling mechanism .
these irc and fis multi - band imaging observations were followed by nir and mir spectroscopic observations and l18w - band ( 13.925.6 @xmath2 m , combining both l15 and l24 bands ) imaging observations carried out in 2007 july 2022 . the spectroscopic observations were conducted using four ( ng , sg1 , sg2 , and lg2 ) grisms .
similar to the aforementioned imaging observations , the ng , sg1 , and sg2 mode observations were conducted simultaneously , while lg2 mode observations were done separately @xcite .
table [ tab_obssp ] lists the characteristics of the @xmath0 spectroscopic observations .
spectra from the peaks of the equatorial emission and ejecta identified in the ratio between the l15- and l24-band images ( l15/l24 hereafter ) were obtained together with that from a reference background position at the southeastern part of the source ( table [ tab_sppos ] ) .
the data calibration and reduction were processed with irc spectroscopy toolkit ( version 20070913 ) .
the obtained flux was converted to the surface brightness based on the measured slit size .
irc spectroscopy is made at the slits located at the edge of the large imaging field - of - view ( fov ) .
there is internal scattered light in the array of the mir - s and the light diffusing from the imaging fov affects sg1 and sg2 spectra to some extent .
this effect has been corrected for according to the method given in @xcite .
a similar scattered light is also recognized in lg2 spectra and has been corrected for in a similar way .
in addition there is a contribution from the second order light from the edges of the imaging fov in lg2 spectra , but we expect that it is not significant in the background - subtracted spectra .
these spectroscopic observations were simultaneously accompanied by supplemental short ( 49 s @xmath10 3 ) l18w - band imaging observations of a field @xmath3 5 away from the central slit position in the southeast direction in order to check the pointing of the satellite .
this supplemental , short - exposure l18w - band image of the southeast , which fortuitously revealed interesting structures beyond the snr boundary ( see 3.4 ) , was combined with deep ( 442 s ) l18w - band image of the source to generate a final larger - area map .
the l18w - band imaging data were processed with the same procedures that we used for other imaging data sets as described above .
figure [ fig_irc ] presents our @xmath0 irc multi - band images of [email protected] together with the atca 20 cm radio continuum image @xcite , @xmath11 x - ray ( 0.38 kev ) image @xcite , and the background - subtracted @xmath0 s11-band image ( s11s7 ; see below ) for comparison .
the @xmath0 ir emission associated with the snr is most apparent in the l15- and l24-band images as two prominent features : first , there is a ring - like structure composed of two clumpy , narrow , and long filaments crossing the central part of the snr along the east - west direction
we name this equatorial ring " ( er ) ; secondly , the outer boundary of the snr appears as an almost - complete shell structure with its eastern part opened we name this outer elliptical shell " ( oes ) .
the radius of the er is @xmath3 3@xmath4 ( or @xmath3 5 pc ) .
its southern filament is brighter than the northern one by a factor of @xmath12 on average and has the brightest clump at ( ra , decl . )
= ( @xmath13 , @xmath14 ) close to the filament center .
the southern filament also shows prominent x - ray and [ ] emission and has been called equatorial bar " or equatorial belt " @xcite .
the northern filament is @xmath3 1 away from the southern filament and both filaments are elongated roughly parallel to each other .
the oes shows an almost - complete ellipse of @xmath3 7 @xmath10 6 ( @xmath3 12 pc @xmath10 10 pc ) with its major axis aligning roughly with the plane of the er . the center of the ellipse determined by elliptical fits weighted by the 24 @xmath2 m surface brightness of the oes is ( @xmath15 , @xmath16 ) located at the middle of two filaments of the er . in the west the oes appears to be connected with the two filaments of the er and its emission is enhanced there , especially at the southwest near the southern filament of the er . in contrast the eastern part of the oes is open , and also the emission from the southern filament of the er is truncated in the east before it reaches the boundary of the snr .
in addition to the er and oes , there appear to be at least two more clearly identifiable features in the @xmath0 images .
first , the l15-band image shows a separate structure that extends @xmath3 5@xmath4 southward from the er on the southeastern side of the snr .
it is elongated vertically with the bright portion located @xmath3 0.5@xmath4 below the southern equatorial filament .
next , there is faint emission extending beyond the northern and southeastern edges of the oes .
this faint emission is contained within the outermost boundary of radio emission observed in [email protected] @xcite .
as in figure [ fig_irc ] , [email protected] is not detected in the s7- and s11-band images : while some of the s11-band emission appears to arise from locations of strong l15- and l24-band emission , there is almost no corresponding emission in the s7-band image .
this suggests that the s7-band emission is dominated by background emission , probably emission from pahs in the line of sight interstellar medium ( ism ) . in the s11s7 difference image ( figure [ fig_irc ] ) we subtract out scaled s7-band emission from the s11-band image and confirmed that the two filaments of the er and southwestern part of the oes have appreciable s11-band emission .
( note that the three bright point - like sources close to the eastern end of the southern filament of the er in the s11s7 difference image are stellar sources . )
the nir n3- and n4-band images , on the other hand , show only stellar emission without any apparent feature brighter than 16 @xmath2jy at 3 @xmath2 m that might be associated with the snr .
table [ tab_f ] presents ir measurements of [email protected] . in the table
, we list the fluxes and ratios of whole remnant area and ir - ejecta region ( see befow for ir - ejecta ) .
in addition , we also list the peak intensities and ratios of the er , the southweastern oes , and the ir - ejecta . figure [ fig_cmap ] ( left ) , which is a three - color ( 7 , 15 and 24 @xmath2 m ) @xmath0 mir image of [email protected] , shows that most of the prominent features have roughly similar mir colors .
one exception is the feature in the southeastern part of the snr with significant excess in the shorter wavebands we name this feature ir - ejecta " because it is believed to originate in shocked sn ejecta as we describe below .
it is bright near the southern equatorial filament and stretching directly southward beyond the snr boundary .
this feature is clearly seen in figure [ fig_cmap ] ( right ) which shows the l15/l24 emission ratio .
the ratio image was produced by dividing the l15-band image by the l24-band image after background subtraction .
the background was estimated by fitting a slanted plane to the areas surrounding the snr .
the resulting background planes were almost flat , i.e. , the brightness differences over the entire images were only 1.7 % and 2.4 % in l15 and l24 , respectively .
we applied a gaussian convolution to the l15-band image in order to match the final spatial resolution to that of the l24-band image and masked out the pixels with small ( @xmath17 30.2 mjy sr@xmath9 ) l24-band intensities .
the resulting l15/l24 color image is significantly different from the original band images . in figure
[ fig_cmap ] , the l15/l24 ratio is almost uniform ( @xmath3 0.25 ) , and there is no feature corresponding to the prominent er and oes . the most notable feature is the ir - ejecta , where l15/l24 ratio raises to @xmath3 0.8 , which is in fact consistent with what we identified in the three - color image ( left panel of figure [ fig_cmap ] ) .
it appears to be of a triangular shape with one of its vertices towards the direction to the snr center .
the southern vertex passes though the southern boundary of the oes and extends over the snr boundary , while the northeastern vertex is located just above the er . the region of high color ratio of the ir - ejecta covers a large portion of the southeastern area of the snr and positionally coincides with the o- and ne - dominant ejecta identified by previous optical and x - ray observations @xcite .
there is also a wispy patch of emission around ( @xmath18 , @xmath19 ) far ( @xmath20 1 ) beyond the snr boundary , which has the color ratio similar to that of the ir - ejecta .
it , however , is not shown in the [ ] images @xcite . besides the ir - ejecta in the southeastern region , the northwestern region shows extended emission of the high color ratio , although it is not as significant as the southeastern region .
overall the two regions of the high color ratio are roughly symmetric with respect to the center of the oes .
figure [ fig_cplot ] compares the pixel values of the l24- and l15-band emission , where most of the points are distributed along the thick , solid line of a slope of 0.25 .
( note that pixels of stars and pixels with small l24-band intensities are masked out . )
if l24- and l15-band fluxes are from the same region and if they are both thermal dust emission , then the color ratio and the dust temperatures are related as @xmath21 where @xmath22 is the surface brightness at @xmath23 ( @xmath2 m ) in jy sr@xmath9 , @xmath24 is the planck function , and @xmath25 is the dust opacity in @xmath26 g@xmath9 .
the slope of 0.25 corresponds to the dust temperature of 126 k for a mixture of carbonaceous and silicate interstellar grain of @xmath27 = 3.1 @xcite . in figure [ fig_cplot ]
there are two regions , where the data points show a high l15/l24 ratio compared to the thick , solid line : first , the thin line of the slope of 0.88 represents one group whose l24-band surface brightness is less than @xmath3 33 mjy sr@xmath9 ; secondly , there is another group of data points of higher l15-band emission whose l24-band surface brightness lies in the range of 3340 mjy sr@xmath9 .
the inset in the lower - right corner of figure [ fig_cplot ] shows that the data points with high l15/l24 ratios are all from the ir - ejecta as expected .
the ones in the second group , i.e. , ones with high 24 @xmath2 m surface brightness , are superposed on the er and therefore their emission is partly from the swept - up csm while the data points in the first group should represent the emission only from the ejecta . the slope of 0.88 corresponds to a dust color temperature of 240 k ; however the l15-band flux is largely from line emission at this position so that the physical dust temperature should be lower than this ( see 3.4 ) .
figure [ fig_spec ] shows the background - subtracted mir grism spectra of the peak positions of the er and the ir - ejecta .
( note that the lg2 spectra at @xmath20 17 @xmath2 m were obtained from positions slightly shifted from those of the sg spectra of 514 @xmath2 m as we described in 2 . )
the spectrum from the ir - ejecta peak clearly shows the [ ] line at 12.8 @xmath2 m , which confirms the nature of the radiatively shocked ejecta .
no ar lines ( i.e. , [ ] at 7.0 @xmath2 m and [ ] at 9.0 @xmath2 m ) were detected , while neither [ ] line at 15.6 @xmath2 m nor [ ] line at 25.9 @xmath2 m was covered .
table [ tab_line ] lists the flux of the [ ] line and upper limits of the [ ] and [ ] lines .
the flux of the [ ] line at 12.8 @xmath2 m is [email protected]@xmath1010@xmath29 erg @xmath30 s@xmath9 sr@xmath9 , corresponding to the surface brightness of [email protected] mjy sr@xmath9 in the l15-band image . given the l15-band surface brightness of [email protected] mjy sr@xmath9 , @xmath3 15 % of the l15-band emission is due to the [ ] line emission . for the s11 band ,
the observed [ ] line emission flux corresponds to the s11-band surface brightness of [email protected] mjy sr@xmath9 .
this is equivalent to @xmath3 55 % contribution to the total s11-band emission , which is much larger than the case of the l15-band emission .
the difference is because there is no strong emission line other than the [ ] line in the s11 band while there is additional [ ] 15.5 @xmath2 m line in the l15 band ( see the spectral responses in figure [ fig_spec ] ) .
( there could be some dust emission too .
see 4.3 . ) on the other hand , the spectrum at the er peak is dominated by the continuum emission , although the lg2 spectra have a lower signal to noise ratio . figure [ fig_ircw ] presents a combined l18w - band image ( 2 ) covering both the snr and an extended area in the southeast . besides the features of the snr that we already described in previous sections ,
there is a notable feature of the l18-band emission in the south . the elongated narrow tail " is located around ( @xmath31 , @xmath32 ) , @xmath3 7 apart from the center of the snr .
it is close to the ir - ejecta in the southeastern part of the snr and its elongation direction is roughly towards the center of the snr .
its extent is @xmath33 or 2.6 pc .
in addition , there is faint , diffuse emission toward the south and southeast too .
the emission toward the southeast appers distinct - it appears to protrude from the open portion of the oes having similar outer boundary with radio plateau at the east and south , but extends @xmath34 southeast beyond the radio boundary of the snr .
we consider that this wide tail " could be associated with the snr .
there is large , diffuse region superposed on [email protected] on the sky ( rcw 54 ; rodgers et al . 1960 ) , but its h@xmath35 emission is elongated along the northeast - southwest direction without any emission corresponding to the wide tail in the southern h - alpha sky survey atlas @xcite .
therefore , the wide tail is not associated with regions and its association with the snr is likely .
figure [ fig_fis ] presents the @xmath0 fis fir - band images of [email protected] .
the images cover the entire snr with a scan direction of northeast to southwest . the fir emission of the snr associated with the er and oes is clearly detected in the n60- and wide - s - band images with strong concentration in the southwestern part , although fine details are difficult to see because of their low spatial resolutions .
the n160- and wide - l - band images are , however , clearly different from images of the other bands .
they show an elongated feature extended in the northwest - southeast direction which shows no correlation with the emission associated with the snr . also its peak position is located outside the boundary of the snr .
these indicate that the n160- and wide - l - band emission is not directly associated with [email protected] , although it is possible that the enhanced brightness in the sw part of oes is due to the interaction of the remnant with this extended structure ( cf .
the bottom panels of figure [ fig_fis ] present the background - subtracted n60- and wide - s - band images .
the background emission was estimated by calculating scaling factors between the n60- and n160-band images and also between the wide - s- and n160-band images from the areas outside the snr .
compared to the n60- and wide - s - band images of the top panels , the background - subtracted images show more clearly the emission associated with the snr , including the northern part of the oes that is not clear in the orignial images .
the er and oes are the most prominent features in our @xmath0 multi - band ir images ( figures [ fig_irc ] and [ fig_fis ] ) .
the er is composed of two filaments , where the southern one is brighter than the northern counterpart .
the southern filament appears to be of the normal composition in soft x - rays @xcite without any apparent radial motion in the optical @xcite , which led the authors to conclude that it is csm from the progenitor of the sn in [email protected] .
we suggest that the northern filament , which has been clearly found by @xmath0 observations , could be part of the same structure based on its similar distribution to the southern filament .
we note that the bright clump at ( @xmath36 , @xmath37 ) near the eastern end of the northern filament has a counterpart in the [ ] image of @xcite .
its velocity is near zero , which supports that the northern and southern filaments form a single structure .
furthermore , it is most clearly visible at the @xmath5120 km s@xmath9 frame of the rutgers fabry - perot velocity scan images and absent at the 0 km s@xmath9 frame where the southern filament is brightest ( figure 2 of ghavamina et al . 2005 ) .
we interpret this velocity difference as an indication suggesting that the northern and southern filaments are parts of a tilted , expanding ring structure produced by the progenitor of the sn .
the location of the center point of the oes at the middle of the two filaments also reconciles with the interpretation , given the csm - nature of the oes ( see below ) .
we note that such a , but smaller , ring structure of the csm was found in sn 1987a and also possibly in the crab nebula @xcite .
the oes , on the other hand , is relatively fainter in x - rays than the er @xcite . the @xmath0 mir brightness and color of the oes , however , are similar to those of the er ( figure [ fig_cmap ] ) . toward the southern filament of the er , @xcite reported the detection of the optical radiative lines produced by the partially radiative shocks starting to develop cooling zones .
it implies that the x - ray emitting gas in the er is cooler than that of the oes , while they have somewhat similar dust properties .
and there is a possibility that the er in s11-s7 difference image also contains the [ ] line emission produced in the csm region where the shock has started to cool down to @xmath38 100,000 k. however , its contribution might be small , because we have not detected the [ ] 12.8 @xmath2 m line at the er . in addition to the er and oes , there are faint mir emissions beyond the northern and southeastern edges of the oes ( figure [ fig_cmap ] ) .
they have sharp outer boundaries , representing the current locations of the sn blast wave .
faint x - ray emission was detected in those areas too @xcite .
it is possible that the sn blast wave has overtaken the oes and produced those emission features while propagating into a more diffuse medium . on the other hand ,
it is also possible that the remnant has a front - back asymmetry and they are just projected boundaries of the more - extended shell . in any case , their asymmetric spatial distribution suggests that either the ambient density distribution or the mass ejection from progenitor was asymmetric . as in figure
[ fig_spec ] , the mir emission of [email protected] is dominated by continuum emission , not by line emission .
this implies that the ir emission is from shock - heated dust grains in the csm .
figure [ fig_sedtot ] presents the spectral energy distribution ( sed ) of the snr in table [ tab_f ] .
we applied modified blackbody fits composed of two dust components to the observed sed in order to obtain the best - fit dust temperatures .
the dust model based on a mixture of carbonaceous and silicate interstellar grain of @xmath27 = 3.1 @xcite gave dust temperatures of @xmath3 103 k ( warm dust ) and @xmath39 47 k ( cold dust ) , corresponding to the mass of 4.5 @xmath28 0.9 @xmath10 10@xmath40 m@xmath41 and @xmath42 4.8 @xmath10 10@xmath43 m@xmath41 , respectively .
( note that the lower limit of the cold dust temperature comes from the upper limit of the flux at 140 @xmath2 m . ) for the case of the dust model based on the graphite and silicate grain of 0.001 to 0.1
@xmath2 m size @xcite , the derived total mass is in the range of ( 1.03.4 ) @xmath10 10@xmath43 m@xmath41 , comparable to the total mass obtained for the former model .
the derived dust mass corresponds to the dust - to - gas ratio of @xmath42 1.6 @xmath10 10@xmath44 , if we use the 30.5 m@xmath45 swept - up mass ( at distance of 6.2 kpc ) of the gas in the csm obtained in x - ray observations @xcite .
this is lower than the ratio 6.2 @xmath10 10@xmath44 found in the local ism @xcite .
the low dust - to - gas ratios in the swept - up materials by the shock destruction were also obtained with spitzer observations on the core - collapse snrs in the large magellanic cloud @xcite and the kepler @xcite . according to our result , @xmath39 75 % of the dust in [email protected]
might have been destroyed by the sn shock , or the initial dust - to - gas ratio surrounding [email protected] might be lower than the local value .
it is comparable to the fraction derived in other snrs , such as 64 % in cas a , @xcite and 78 % in kepler @xcite .
we perform model simulations for the destruction of dust by sn shock waves and the thermal emission from shock - processed dust .
the physical processes of dust in shocks have been so far discussed in many works ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
once the circumstellar dust grains are swept up by the blast wave , they acquire high velocities relative to the gas and are eroded by kinetic and/or thermal sputtering in the shock - heated gas .
dust grains are also heated by collisions with energetic electrons in the hot gas and radiate thermal emission at ir wavelengths .
dynamics , erosion , and temperature of dust depend on the temperature and density of the gas as well as the chemical composition and size of dust grains .
we adopt the model of dust destruction calculation by @xcite , in which the motion , destruction , and heating of dust in the shocked gas are treated in a self - consistent manner by following the time evolution of the temperature and density of the gas for the spherically symmetric shock wave . as the initial condition of the sn ejecta we consider the freely expanding ejecta with the velocity profile of @xmath46 and the density profile of @xmath47 at @xmath48 and @xmath49 at @xmath50 , where @xmath51 and @xmath52 are the velocity and radius of the outermost ejecta , respectively .
taking the kinetic energy of @xmath53 ergs and the ejecta mass of 8 @xmath54 @xcite for the density profile of core - collapse snr of @xmath55 and @xmath56 ( e.g. , * ? ? ?
* ; * ? ? ?
* ) , we obtain @xmath57 g cm@xmath44 , @xmath59 cm s@xmath9 , and @xmath60 cm at 10 yrs after explosion , when the simulations are started . since most of the dust grains swept up by the forward shock during the later epoch of the evolution , the calculation results are not sensitive to the ejecta structure . for the ambient medium
we consider the constant hydrogen number density of @xmath61 0.1 , 0.5 , 1 , and 10 cm@xmath44 .
the circumstellar dust is assumed to be amorphous carbon or silicate ( forestrite ) with the power - law size distributions ( @xmath62 ) ranging from @xmath63 @xmath2 m to @xmath64 @xmath2 m @xcite .
the optical constants are taken from @xcite and @xcite .
based on the time evolution of the size distribution and temperature of dust given by the simulation and the assumption of the dust grains being in thermal equilibrium , we calculate the ir sed by thermal emission from the shock - heated dust .
the detailed description for calculating the ir sed from shocked dust will be given elsewhere ( t. nozawa et al .
2009 in preparation ) .
figure [ fig_dmodel ] compares the observed fluxes of [email protected] with the calculated ir seds at 3,000 yrs for @xmath61 0.1 , 0.5 , 1 , and 10 cm@xmath44 .
we present the results with the initial dust - to - gas mass ratio to best reproduce parts of the observed sed , for amorphous carbon ( figure [ fig_dmodel]a ) and silicate ( figure [ fig_dmodel]b ) .
the typical temperatures of dust are 3555 , 4565 , 5070 , and 6080 k for @xmath61 0.1 , 0.5 , 1 , and 10 cm@xmath44 , respectively , and the resulting dust masses are in the range of ( 0.35 ) @xmath65 @xmath66 for carbon and ( 0.48 ) @xmath65 @xmath66 for silicate with the higher values for lower ambient density ( thus lower temperature of dust ) .
it can be seen that the results of the silicate grains with @xmath61 0.5 cm@xmath44 , which coincides with the density estimated from x - ray observations @xcite , can reasonably reproduce the overall shape of the ir sed for [email protected] . in this case
, the mass of grains radiating ir emission is @xmath67 @xmath66 .
note that the initial dust - to - gas mass ratio corresponding this result is @xmath68 , which is smaller than that in the local galaxy .
if the ir emission originates in the swept - up materials containing the mass - loss wind of rsg with the solar metallicity , the condensation of silicate is expected , and , according to our results , its condensation efficiency could be low .
it should be noted here that the simulation results for @xmath69 10 cm@xmath44 significantly underestimates the flux at short ( 11 @xmath70 ) wavelength .
however , the sed at shorter wavelengths could be resolved by including the effect of a stochastic heating of small grains ; stochastically heated dust produces more emission at shorter wavelengths than the dust with equilibrium temperature , and it may also allow acceptable fits for lower initial density and/or different size distribution of dust than the current best fit . alternatively , this disagreement may be caused by the difference in the assumed composition of dust .
to gain deeper insights into the properties of dust in [email protected] , we need further investigations by taking account of the stochastic heating and changing the composition and size distribution of dust as well as the density profile in the ambient medium .
we have detected mir emission associated with the sn ejecta : ir - ejecta .
it is prominent in the 15/24 @xmath2 m ratio image by its high l15/l24 ratio ( figure [ fig_cmap ] ) , but is marginally seen in the total intensity maps of the mir ( 1124 @xmath2 m ) and fir ( 6590 @xmath2 m ) too .
the emitting area coincides with the fast - moving ( @xmath71 km s@xmath9 ) , o - rich sn ejecta : the triangle - shaped bright portion near the equatorial filament was called spur " and the extension to the south was called streamers " by @xcite . in the high - resolution [ ] 5007 image , the spur is crescent - shaped with a sharp boundary toward the snr center , while the streamer appears to be composed of clumps embedded in diffuse emission .
recent measurement of their proper motions showed that they are expanding systematically from a point near the geometrical center of the oes ( * ? ? ? * see 4.4 too ) . according to our spectroscopic result , the ir - ejecta shows 12.8 @xmath2 m emission but no [ ] 7.0 or [ ] 9.0 @xmath2 m emission .
note that the latter lines from ar ions were also detected at the metal - rich ejecta in young snrs @xcite .
the lack of ir line emission from ar ions in this area is consistent with the results of optical or x - ray studies which showed that there is no line emission from elements heavier than s in this area ( ghavamian et al .
2005 and references therein ; park et al .
the absence of ar lines supports the claim by @xcite that we are not seeing the ejecta in the inner - most region accelerated by pulsar wind nebula but seeing the he - burning - synthesized , o - rich ejecta swept - up by reverse shock .
the ir - ejecta shows high l15/l24 ratio , i.e. , l15/l24=0.88 compared to 0.25 of the shocked csm .
the high l15/l24 ratio is at least partly due to ne lines .
( there is [ ] 25.9 @xmath2 m line in the l24 band , but it is weak , i.e. , @xmath72% of the [ ] + [ ] lines . see 4.3 . )
according to our estimation , 12.8 line contributes @xmath3 15 % to the l15 flux . in the l15 band
, there is another strong ne line , 15.6 line .
we may estimate the possible contribution of 15.6 line in g292.0 + 1.8 as follows .
the strength of a forbidden ne line is given by @xmath73 where @xmath74 is the frequency of the line , @xmath75 is the column density of ne@xmath76 or ne@xmath77 ions in the upper state along the line of sight , and @xmath78 is the einstein coefficient , so that the 15.6 / 12.8 ratio is given by @xmath79 where @xmath80 and @xmath81 are column densities of ne@xmath77 and ne@xmath82 ions , @xmath83 and @xmath84 are the fractions of ne@xmath77 and ne@xmath82 ions in the upper states of the corresponding emission lines ( see glassgold et al .
2007 for atomic parameters of these lines ) .
assuming statistical equilibrium , the last factor , @xmath85 , can be easily calculated ( e.g. , * ? ? ?
* ) , and varies from 1 to 2 as the electron density increases from a low - density limit to densities much higher than the critical density ( @xmath86 cm@xmath44 at 10,000 k ) .
therefore , the intensity ratio @xmath87 is @xmath88 unless @xmath89 , which happens at temperatures @xmath90 k in collisional equilibrium ( e.g. , * ? ? ?
@xcite suggested that ejecta undergoes radiative shocks with a velocity of 50200 km s@xmath9 from the [ ] 5007 line widths .
the temperature of shocked ejecta gas immediately behind the shock of velocity @xmath91 should be high , i.e. , @xmath92 k , assuming pure neon preshock gas , but since the shocked gas element cools fast due to enhanced heavy elements and the column density of the gas at @xmath93 might vary roughly with @xmath94 , we expect @xmath95 and therefore @xmath96 . toward the sn ejecta in young core - collapse snrs such as crab , cas a , the snr [email protected] in the lmc , or the snr [email protected] in the smc , the 15.6 /12.8 ratio varies from 0.3 to @xmath97 too @xcite .
we may conclude that the contribution of the 15.6 line to the l15 band should be at most comparable to that of the 12.8 line unless the shock is truncated at relatively high temperatures .
we have derived ir fluxes associated with the ejecta and the results are summarized in table [ tab_f ] .
it is difficult to extract the ejecta emission in ir observations , although our results unveiled the ejecta region unambiguously .
the ir - ejecta is clearly divided into two regions ( figures [ fig_cmap ] and [ fig_cplot ] ) ; one that coincides with the ejecta - only region and has a tight correlation in the l15 vs l24 plot and the other superposed on the er and has a loose correlation in the l15 vs l24 plot .
these ir - ejecta regions are part of the structure defined as the [ ] spur " by @xcite , with the ejecta - only region being the bottom of the spur , and the ejecta superimposed on the er being the top portion of the spur .
we measure the flux at the ejecta - only region masking the emission from stars .
the area of the ejecta - only region is more than four times larger than the superposed ejecta region and it contains most of the flux . in order to remove the background emission ,
we subtract the average of nearby ( @xmath3 2@xmath4 ) low intensity region .
the @xmath0 ejecta spectrum in figure [ fig_spec ] shows that the 11 and 15 @xmath2 m fluxes are not from dust continuum .
we believe that the s11-band is mostly dominated by line emission , while the l15-band needs an additional component ( see 3.3 ) . in the next paragraph , we show that some of the l15 flux is due to a bump at 1525 @xmath2 m . on the other hand ,
fir emission is certainly dominated by dust continuum .
the l24-band contains the [ ] 25.9 @xmath2 m line emission , whose wavelength is not covered with our spectroscopic observations .
its contribution in l24-band is expected to be less than 20 % in our estimation using the [ ] 25.9 @xmath2 m line intensity in @xcite . at first
, we fit 24140 @xmath2 m band fluxes using a modified - blackbody for silicate and graphite grain models of 0.0010.1 @xmath2 m size @xcite .
the best fits give temperature of 6488 k and dust mass of 2.08.2 @xmath10 10@xmath29 m@xmath41 .
if we exclude the l24-band data point , the temperature drops to 4452 k and the dust mass increases by a factor of 5 .
this ir - emitting dust mass is much smaller ( @xmath38 4 @xmath10 10@xmath40 ) than the theoretically predicted dust mass ( 0.11 m@xmath98 ) to be formed in the core - collapse sn explosion ( e.g. , * ? ? ?
recently the results from mir observations of [email protected] with the @xmath6 satellite have been published @xcite based on low - resolution spectroscopy of the ejecta and the southern filament of the er where the ejecta emission is superimposed on the emission from the csm . the former is located @xmath3 0.5 north from our ir - ejecta slit position , while the latter , which has a high l15/l24 ratio , is @xmath3 1 apart in the northeast direction .
their ejecta spectrum shows strong emission from the [ ] 12.8 @xmath2 m , [ ] 15.6 @xmath2 m , [ ] 25.9 @xmath2 m lines and relatively weak [ ] 24.4 @xmath2 m and [ ] 36.0 @xmath2 m lines , but no lines of mg , si , s , ar or fe are identified .
( note that the si and s lines in their spectrum are not from the ejecta but from the background . )
the observed ratio of [ ] 12.8 @xmath2 m to [ ] 15.6 @xmath2 m is 2.12.7 .
the non - detection of ar lines and the observed ratio of ne lines are consistent with our results and the prediction in 4.2 .
the observed flux of [ ] 25.9 @xmath2 m line is @xmath72% of the [ ] plus [ ] line fluxes @xcite .
if the @xmath0 l15 and l24 fluxes are entirely due to lines , we estimate that the l15/l24 ratio should be 2.32.6 using the line fluxes in @xcite .
( it becomes higher if we consider the 15 - 25 @xmath2 m bump in the next paragraph . )
the observd ratio , however , is 0.25 for the er and oes and 0.88 for the ejecta .
therefore , the contribution of [ ] 25.9 @xmath2 m line flux to the observed l24-band flux should be small , particularly toward the er where @xmath99 ( 3.2 ) .
we , however , note that the @xmath6 irs spectrum toward the ejecta in @xcite does not show any obvious dust continuum emission between 24 and 36 @xmath2 m .
we consider that it could be because there was dust continuum emission in the irs background . in the @xmath0 24 @xmath2 m image ,
there is faint filament coincident with the irs background region ( ll1 sky in ghavamian et al .
this filament is associated with the oes and its 24 @xmath2 m emission might be dominated by dust continuum , so that , by subtracting its spectrum from the ejecta spectrum , the continuum feature could have been removed .
an interesting feature in the @xmath6 ejecta spectrum is a weak 1525 @xmath2 m bump , which was suggested to be produced by newly - formed dust or swept - up pahs along the line of sight .
first of all , this bump explains the high l15/l24 ratio at the ejecta region described in
the peak of the bump feature appears in the l15 band .
therefore , it contributes mainly to the @xmath0 15 @xmath2 m band .
secondly , our @xmath0 observations show that the areas with a high l15/l24 ratio are coincident with those of the optical o - rich ejecta region .
this supports the interpretation that the bump feature is related to the newly - formed dust in association with the sn ejecta , not to the swept - up pahs . according to @xcite , the @xmath6 spectrum from the southern filament of the er is consistent with the emission from two dust components : a warm ( or hot ) component of @xmath100 114 k and a cold component of
@xmath100 35 k. the temperature of the warm dust is higher than what we estimated with the @xmath0 data by @xmath3 10 k , while that of the cold dust is lower than the @xmath0 temperature by a comparable amount .
it is possible that the presence of the bump feature in the @xmath6 spectrum resulted in a higher temperature for the warm dust . for the cold dust temperature ,
as explained in @xcite , the absence of the longer wavelength ( @xmath20 30 @xmath2 m ) data in the @xmath6 spectrum is likely the primary reason of the lower temperature compared to the @xmath0 results .
ir spectroscopic observations covering a broad band are necessary to clearly resolve the issues .
one interesting result that we obtained in this study is the difference ( @xmath3 42 @xmath101 1 pc ) between the center of the csm and the dynamical center of the ejecta ( figure [ fig_cmap ] ) .
the former corresponds to the center of the er and oes that we determined with the @xmath0 results ; the latter was determined by the distribution of the o - rich ejecta in the optical , which is close to the center of the snr in the radio emission .
in addition , the position of the pulsar is different from the both positions it is shifted in the southeast direction from the dynamical center by @xmath3 46 .
one possibility is that the progenitor star was at the center of oes during its rsg phase but exploded at the position of the dynamical center of optical knots .
this is possible because the rsg wind could be confined by external pressure while the central star is moving .
according to @xcite , the rsg wind from a 2535 m@xmath41 star , which explodes as sn iil / b , would be pressure confined while its outer radius becomes @xmath102 pc .
the radius of oes ( 6 pc ) is comparable to what the theory predicts .
if the progenitor star was moving at @xmath3 10 km s@xmath9 and exploded after @xmath3 10@xmath103 yrs of the pressure confinement of the shell , then the explosion center would be close to the dynamical center of optical knots .
another possibility is that the sn exploded at the center of er and oes , not at the dynamical center of the ejecta or the center of the radio emission .
this is motivated by the fact that the oes shows a very well - defined shell structure surrounding the snr , thereby the center of the oes may pinpoint to the real location of the progenitor , which later exploded as a sn . on the other hand ,
the centers of the radio emission and the o - rich ejecta motion could have been weighted toward the southeast : firstly , the geometrical center of the radio nebula could be significantly weighted toward the southeast because of the existence of the bright pwn . secondly
, the dynamical center of optical knots is also weighted to the southeast because the bright optical knots in the southeastern area are moving relatively slowly and the center is derived using an assumption of an unhindered constant velocity since the explosion @xcite .
if the sn explosion in [email protected] indeed occurred at the center of the oes , then the tangential velocity of the pulsar needs to be @xmath3 1,000 km s@xmath9 .
although 1,000 is somewhat large as a pulsar kick velocity , it is still within the acceptable range ( e.g. , * ? ? ?
* ) , and the pulsar in another o - rich snr puppis a may also have such a high velocity @xcite . the optical and x - ray studies of the sn ejecta in [email protected] have shown that their spatial distribution and physical properties are not symmetric , including the northwest - southeast concentration of the o and ne ejecta , and the distinctive difference of the x - ray plasma temperature between the northwestern and southeastern areas @xcite .
also the ejecta in the northern and southern boundaries move faster than those in the eastern and western areas @xcite .
it is worth to note that the pulsar jet axis is also along the northeast - southwest direction @xcite .
our @xmath0 result on the ne - line emitting ejecta material identified by their high l15/l24 ratio is consistent with the spatial distribution seen in [ ] 5007 : most of the @xmath104/@xmath105 ions are distributed in the southeastern area called spur and streamers , several isolated ones coincide with the optical knots , and the other group of ne - rich knots is distributed in the northwestern area and coincides with [ ] optical emission knots ( fast - moving knots , or fmks , as reported by @xcite and @xcite ) .
the northwest ejecta is less obvious in optical but it is easily recognized in the x - ray image of ionized o and ne elements @xcite .
this type of bipolar ejecta distribution is also found in other young core - collapse snrs , e.g. , cas a and [email protected] @xcite .
cas a shows a similar ne ejecta distribution , which is almost perpendicular to the well - known narrow northeast - southwest jet , but is roughly aligned to the bipolar ionic ejecta , which sometimes is suggested be the major direction of explosion @xcite . in [email protected] , which is the remnant of the historical sn ad 386 , the iron ejecta
is found to be distributed mainly along northwest - southeast direction @xcite . the symmetry axis in the spatial and kinematical distribution of ejecta in [email protected] ,
therefore , is either along northwest - southeast or north - south , which is perpendicular to the plane of the er . we consider that either the explosion was asymmetric and/or the cs wind was denser in the equatorial plane so that the ejecta expanding in this plane was slowed down more compared to those expanding to the other directions .
an interesting feature is the narrow tail in the wide l18w - band image that extends from the end of the streamers to outside the remnant .
a possible explanation is that the narrow tail is a part of the o - rich streamer .
note that the narrow tail is connected to the streamers by a southern patch of emission near the boundary of l15-band image .
the distance from the center to the end of the narrow tail is 7@xmath4 ( 13 pc ) , which is @xmath3 2@xmath4 ( 3 pc ) farther than the outermost o - rich clump in this area @xcite .
it , however , does not has an optical counterpart in the [ ] image @xcite . if we assume a constant velocity and adopt an age of 3,000 yrs @xcite , the transverse velocity of the narrow tail is 4,000 km s@xmath9 .
this gives a possibility that its radial velocity is also very large , beyond the velocity range of previous optical narrow - band imaging observations ( e.g. , @xmath3 2,000 km s@xmath9 in * ? ? ?
an alternative explanation is that the narrow tail is a reradiated ir light echo similar to the echoes identified in cas a @xcite .
the lack of counterpart in the optical [ ] and x - ray images , together with the faint feature in the fir images ( figure [ fig_fis ] ) , could be consistent with the continuum origin of the ir emission . in case of a light echo
, the location of the echo can be derived from the geometrical equation of ellipse whose two focuses are the sn and the observer ( e.g. , * ? ? ?
* ; * ? ? ?
applying a projected radius of 13 pc and a time delay of 3,000 yrs , we obtain a location of the ir echo at 450 pc behind [email protected] and its angular offset of @xmath3 2@xmath106 from the line of sight .
it appears that this alignment is too tight at a first glance , but it can be a selection effect caused by our limited imaging area .
for example , @xcite discovered ir echoes along the scan direction in cas a for the first time , but @xcite reported that the echoes were distributed in many positions in large ( @xmath3 2@xmath106 ) area .
more observations are definitely necessary to inspect the nature of the narrow tail . on the other hand , the sharp boundary of the sn blast wave was detected only toward the north and southwest ( figure [ fig_cmap ] ) .
the absence of sn blast wave in the other directions implies that either the shock is trapped within the shell because that part of the shell has a larger column density or the sn blast wave has propagated far beyond because the ambient density is lower toward that direction . there is no indication in the @xmath0 images that the oes is denser , where the sn blast wave is missing . instead those parts are fainter in the fir images which indicates a lower column density . in this regard ,
the faint mir emission that extends far beyond the bright shell to the southeast ( figure [ fig_ircw ] ) is interesting because if it is part of the snr , it indicates that the sn blast wave has propagated much further out toward this direction probably due to the lower ambient density .
a deep radio observation could reveal faint features associated with this ir structure .
we have presented the nir to fir imaging and mir spectroscopic observations of the o - rich snr [email protected] using the irc and fis instruments aboard @xmath0 satellite .
the almost continuous multiband imaging capability of @xmath0 covering wide ir wavelengths together its wide field of view enabled us to clearly see the distinct ir emission from the entire snr and to derive its ir charactersitics .
we derive the physical parameters of ir - emitting dust grains in the swept - up circumstellar medium and compare the result with the model calculations of dust destruction by a sn shock .
the overall shape of the observed sed can be explained by a simple model using characteristic snr parameters with a bit lower initial dust - to - gas ratio . at 11 @xmath2 m ,
the model flux is significantly smaller , which may indicate the importance of stochastic heating .
we have not detected any signficant amount of freshly - formed dust associated with the sn ejecta .
the akari results in this paper give new insights into the explosion dynamics of [email protected] .
we have discovered an almost symmetric ir shell probably produced by the circumstellar wind from the progenitor star in the rsg phase .
its center is significantly offset from the previously suggested explosion centers .
we consider that either oes represents the circumstellar shell pressure - confined by external medium or the sn exploded close to the center of the oes . in the latter case
, the pulsar in [email protected] may be traveling at a speed of @xmath107 . at the same time
, the ejecta distribution is unveiled by their high 15 to 24 @xmath2 m ratio .
the ejecta are mainly distributed along the northwest - southeast direction .
this symmetric pre - supernova structure and asymmetric ejecta distribution appear to be rather common in the remnants of sn iil / b , which suffer strong mass - loss like cas a @xcite .
there is also a narrow tail outside the snr shell , which might be similar to the feature observed in cas a. a detailed study is necessary to understand the nature of this feature .
we suggest that multi - band ir imaging observations are powerful tools to explore both the ejecta and csm emission in young core - collapse snrs . especially the 15 and 24 @xmath2 m images are useful to reveal the detailed structure of ir features , which leads to better understanding of environments of the progenitor and the sn explosion . this work is based on observations with @xmath0 , a jaxa project with the participation of esa .
we wish to thank all the members of the @xmath0 project .
we also thank b. gaensler for providing the atca 20 cm image and s. park for providing the @xmath11 x - ray image .
this work was supported by the korea research foundation grant funded by the korean government ( krf-2008 - 357-c00052 ) and the korea science and engineering foundation ( r01 - 2007 - 000 - 20336 - 0 ) .
this work was also supported in part by a grant - in - aid for scientific research for the japan society of promotion of science ( 18204014 ) .
t.n . has been supported in part by world premier international research center initiative ( wpi initiative ) , mext , japan , and by the grant - in - aid for scientific research of the japan society for the promotion of science ( 19740094 ) .
allen , j. w. , & dupree , a. k. 1969 , , 155 , 27 blair , w. p. , ghavamian , p. , long , k. s. , williams , b. j. , borkowski , k. j. , reynolds , s. p. , & sankrit , r. 2007 , , 662 , 998 bouchet , p. , de buizer , j. m. , suntzeff , n. b. , danziger , i. j. , hayward , t. l. , telesco , c. m. , & packham , c. 2004 , , 611 , 394 braun , r. , goss , w. , m. , caswell , j. l. , & roger , r. s. 1986 , , 162 , 259 camilo , f. , manchester , r. n. , gaensler , b. m. , lorimer , d. r. , & sarkissian , j. 2002 , , 567 , l71 chevalier , r. a. 1982 , , 259 , 302 chevalier , r. a. 2005 , , 619 , 839 couderc , p. 1939 , ann .
dastrophys . , 2 , 271 douvion , t. , lagage , p. o. , cesarsky , c. j. , dwek , e. 2001 , , 373 , 281 draine , b. t. , & lee , h. m. 1984 , , 285 , 89 draine , b. t. , 2003 , , 598 , 1017 dwek , e. , petre , r. , szymkowiak , a. , & rice , w. l. 1987 , , 320 , 27 dwek , e. , foster , s. m. , & vancura , o. 1996 , , 457 , 244 dwek , e. & arendt , r. g. 2008 , , 685 , 976 edo , o. 1983 , phd dissertation , dept . of physics , university of arizona fesen , r. a. & gunderson , k. s. 1996 , , 470 , 967 gaensler , b. m. & wallace , b. j. 2003 , , 594 , 326 gaustad , j. e. , mccullough , p. r. , rosing , w. , & van buren , d. 2001 , pasp , 113 , 1326 ghavamian , p. , hughes , j , p. & williams , t. b. 2005 , , 635 , 365 ghavamian , p. , raymond , j. c. , blair , w. p. , long , k. s. , tappe , a. , park , s. , winkler , p. f. 2009 , , in press glassgold , a. e. , najita , j. r. , & igea , j. 2007 , , 656 , 515 gonzalez , m. , & safi - harb , s. 2003 , , 583 , 91 goss , w. m. , shaver , p. a. , zealey , w. j. , murdin , p. , & clark , d. h. 1979 , , 188 , 357 green , d. a. , tuffs , r. j. , & popescu , c. c. 2004 , , 355 , 1315 hines , d. c. , et al .
2004 , , 154 , 290 hughes , j. p. , et al .
2003 , , 591 , l139 hui , c. y. , & becker , w. 2006 , , l457 , 33 hwang , u. , et al .
2004 , , 615 , 117 jones , a. p. 2004 in asp conf .
309 , astrophysics of dust , ed .
a. n. witt , g. c. clayton , & b. t. draine ( san francisco : asp ) , 347 kawada , m. , et al .
2007 , pasj , 59 , 389 koo , b .-
c . , moon , d .- s . , lee , h .-
lee , j .- j .
, & matthews , k. 2007 , , 657 , 308 krause , o. , et al .
2005 , science , 308 , 1604 laor , a. , & draine , b. t. 1993 , , 402 , 441 lockhart , i. a. , gross , w. m. , caswell , j. l. , & mcadam , w. b. 1977 , , 179 , 147 mathis , j. s. , rumpl , w. , & nordsieck , k. h. 1977 , , 217 , 425 matzner , c. d. , & mckee , c. f. 1999 , , 510 , 379 moon , d .- s . , et al . 2004 , , 610 , l33 moon , d .- s . , et al .
2009 , , 703 , l81 murakami , h. , et al .
2007 , pasj , 59 , 369 murdin , p. & clark , d. h. 1979 , , 189 , 501 ng , g .- y . & romani , r. w. 2007 , , 660 , 1357 nozawa , t. , kozasa , t. , umeda , h. , maeda , k. & nomoto , k. 2003 , , 598 , 785 nozawa , t. , kozasa , t. , & habe , a. 2006 , , 648 , 435 ohyama , y. , et al .
2007 , pasj 59 , 411 onaka , t. , et al .
2007 , pasj 59 , 401 park , s. , roming , p. w. a. , hughes , j. p. , slane , p. o. , burrows , d. n. , garmire , g. p. , & nousek , j. a. 2002 , , 564 , l39 park , s. , et al .
2004 , , 602 , l33 park , s. , hughes , j. p. , slane , p. o. , burrows , d. n. , gaensler , b. m. , & ghavamian , p. 2007
, , 670 , l121 pittard , j. m. , dyson , j. e. , falle , s. a. e. g. , & hartquist , t. w. 2001 , , 375 , 827 rho , j. , kozasa , t. , reach , w. t. , smith , j. d. , rudnick , l. , delaney , t. , ennis , j. a. , gomez , h. , & tappe , a. 2008 , , 673 , 271 rho , j. , reach , w. t. , tappe , a. , hwang , u. , slavin , j. d. , kozasa , t. , & dunne , l. , 2009 , , 700 , 579 rodgers , a. w. , campbell , c. t. , & whiteoak , j. b. 1960 , , 121 , 103 sakon , i. , et al .
2008 , proc . of spie , 7010 , 88 semenov , d. , henning , th . ,
helling , ch . ,
ilgner , m. , & sedlmayr , e. 2003 , , 410 , 611 smith , j. d. t. , rudnick , l. , delaney , t. , rho , j. , gomez , h. , kozasa , t. , reach , w. , & isensee , k. 2009 , , 693 , 713 temim , t. , et al .
2006 , , 132 , 1610 tielens , a. g. g. m. , mckee , c. f. , seab , c. g. , & hollenbach , d. j. 1994 , , 431 , 321 vancura , o. , raymond , j. c. , dwek , e. , blair , w. p. , long , k. s. , & foster , s. 1994 , , 431 , 188 wheeler , j. c. , maund , j. r. , & couch , s. m. 2008 , , 677 , 1091 williams , b. j. , et al . 2006 , , 652 , l33 williams , b. j. , et al .
2008 , , 687 , 1054 winkler , p. f. & long , k. s. 2006 , , 132 , 360 winkler , p. f. , & petre , r. 2007 , , 670 , 635 winkler , p. f. , twelker , k. , reith , c. n. , & long , k. s. 2009 , , 692 , 1489 zeiger , b. r. , brisken , w. f. , chatterjee , s. , & goss , w. m. 2008 , , 674 , 271 zubko , v. , dwek , e. , & arendt , r. g. 2004 , , 152 , 211 l ccl 2007 . 1 .
& 1400748 & irc imaging & n3 , n4 , s7 , s11 + 2007 . 1 .
& 1400749 & irc imaging & l15 , l24 + 2007 . 1 . 19 . &
1400751 & fis imaging & n60 , wide - s , wide - l , n160 + 2007 .
& 1402801 & irc imaging & l18w + 2007 .
7 . 20 . & 1402839 , 41 , 43 & irc spectroscopy & nir , mir - s + 2007 .
21.-22 . & 1402840 , 42 , 44 & irc spectroscopy & mir - l lc cccccc cccc band & n3 & n4 & s7 & s11 & l15 & l24 & n60 & wide - s & wide - l & n160 + reference wavelength ( @xmath2 m ) & 3.2 & 4.1 & 7.0 & 11.0 & 15.0 & 24.0 & 65 & 90 & 140 & 160 + effective bandwidth ( @xmath2 m ) & 0.87 & 1.53 & 1.75 & 4.12 & 5.98 & 5.34 & 21.7 & 37.9 & 52.4 & 34.1 + pixel size ( @xmath108 ) & 1.46 & 1.46 & 2.34 & 2.34 & 2.51@xmath109 & 2.51@xmath109 & 26.8 & 26.8 & 44.2 & 44.2 + fwhm ( @xmath108 ) & 4.0 & 4.2 & 5.1 & 4.8 & 5.7 & 6.8 & 37 & 39 & 58 & 61 l cccc grism & ng & sg1 & sg2 & lg2 + slit size ( @xmath110 ) & 5 @xmath10 0.8 & 5 @xmath10 0.8 & 5 @xmath10 0.8 & 7 @xmath10 0.4 + wavelength coverage ( @xmath2 m ) & 2.55.0 & 4.69.2 & 7.213.4 & 17.526.5 + resolving power & 120 & 53 & 50 & 48 + exposure time ( s ) & 400 & 196 & 245 & 442 l ccc cc + s11 & 0.30 ( 0.02 ) & 0.53 ( 0.26 ) & 0.53 ( 0.24 ) & 0.013 ( 0.001 ) & 0.37 ( 0.24 ) + l15 & 2.81 ( 0.14 ) & 5.10 ( 0.36 ) & 5.04 ( 0.36 ) & 0.066 ( 0.009 ) & 1.96 ( 0.26 ) + l24 & 11.98 ( 0.72 ) & 19.09 ( 1.37 ) & 18.94 ( 1.26 ) & 0.072 ( 0.015 ) & 2.60 ( 0.36 ) + n60 & 30.6 ( 6.1 ) & 8.81 ( 2.53 ) & 7.65 ( 1.83 ) & 0.173 ( 0.035 ) & 5.32 ( 1.47 ) + wide - s & 30.0 ( 9.0 ) & 6.76 ( 2.14 ) & 6.97 ( 2.20 ) & 0.250 ( 0.075 ) & 6.05 ( 1.95 ) +
wide - l & @xmath38 6.8 & @xmath38 6.50 & @xmath38 6.58 & @xmath38 0.048 & @xmath38 6.87 + + s11/l15 & 0.11 ( 0.01 ) & 0.10 ( 0.05 ) & 0.11 ( 0.05 ) & 0.20 ( 0.03 ) & 0.19 ( 0.13 ) + l15/l24 & 0.23 ( 0.02 ) & 0.27 ( 0.03 ) & 0.27 ( 0.03 ) & 0.92 ( 0.23 ) & 0.75 ( 0.14 ) + n60/wide - s & 1.02 ( 0.37 ) & 1.30 ( 0.56 ) & 1.10 ( 0.43 ) & 0.69 ( 0.25 ) & 0.88 ( 0.37 ) + wide - s / wide - l & @xmath20 4.42 & @xmath20 1.04 & @xmath20 1.06 & @xmath20 5.25 & @xmath20 0.88 | we present the results of @xmath0 observations of the o - rich supernova remnant [email protected] using six irc and four fis bands covering 2.726.5 @xmath2 m and 50180 @xmath2 m , respectively .
the @xmath0 images show two prominent structures ; a bright equatorial ring structure along the east - west direction and an outer elliptical shell structure surrounding the remnant .
the equatorial ring structure is clumpy and incomplete with its western end opened .
the outer shell is almost complete and slightly squeezed along the north - south direction .
the central position of the outer shell is @xmath3 1@xmath4 northwest from the embedded pulsar and coincides with the center of the equatorial ring structure . in the northen and southwestern regions
, there is also faint emission with a sharp boundary beyond the bright shell structure . the equatorial ring and the elliptical shell structures were partly visible in optical and/or x - rays , but they are much more clearly revealed in our @xmath0 images .
there is no evident difference in infrared colors of the two prominent structures , which is consistent with the previous proposition that both structures are of circumstellar origin .
however , we have detected faint infrared emission of a considerably high 15 to 24 @xmath2 m ratio associated with the supernova ejecta in the southeastern and northwestern areas .
our irc spectra show that the high ratio is at least partly due to the emission lines from ne ions in the supernova ejecta material .
in addition we detect a narrow , elongated feature outside the snr shell .
we derive the physical parameters of the infrared - emitting dust grains in the shocked circumstellar medium and compare the result with model calculations of dust destruction by a sn shock .
the @xmath0 results suggest that the progenitor was at the center of the infrared circumstellar shell in red supergiant stage and that the observed asymmetry in the sn ejecta could be a result of either a dense circumstellar medium in the equatorial plane and/or an asymmetric explosion . |
recent results from longitudinally polarized lepton - nucleon scattering experiments @xcite and , in particular , for single - inclusive pion and jet production in helicity - dependent proton - proton ( @xmath0 ) collisions at the relativistic heavy ion collider ( rhic ) @xcite have started to put significant limits on the amount of gluon polarization in the nucleon @xcite .
this is best quantified in a `` global qcd analysis '' , which treats all available experimental probes simultaneously and consistently at a given order in the strong coupling @xmath1 in perturbative qcd ( pqcd ) .
it allows one to extract the set of universal , spin - dependent parton distribution functions , defined as @xmath2 that yields the optimum theoretical description of the combined data . in ( [ eq : pdfdef ] ) , @xmath3 ( @xmath4 ) denotes the probability of finding a parton of flavor @xmath5 at a resolution scale @xmath6 with light - cone momentum fraction @xmath7 and helicity @xmath8 ( @xmath9 ) in a proton with helicity @xmath8 .
our current understanding of the spin structure of the nucleon is derived from eq .
( [ eq : pdfdef ] ) by taking the first moments of the densities @xmath10 .
these quantities enter the helicity sum rule of the nucleon along with the contributions from the orbital angular momenta of quarks and gluons @xcite .
specifically , the total gluon polarization is given by @xmath11 and the challenge is to precisely map the gluon helicity density @xmath12 in a wide range of @xmath7 in order to minimize extrapolation uncertainties in the first moment @xmath13 .
a first global qcd analysis of polarized parton densities @xmath10 at next - to - leading order ( nlo ) accuracy was completed recently @xcite .
it was based on the world - data on polarized inclusive and semi - inclusive deep - inelastic scattering , which are pivotal in constraining the quark and antiquark densities @xcite , as well as on the latest rhic @xmath0 measurements @xcite mentioned above .
the conclusion is that available results from all lepton - nucleon scattering experiments and the rhic spin program are in nice agreement .
this underpins the notion of factorization also for spin - dependent hard scattering processes , which is the foundation for most pqcd calculations and their predictive power .
the polarized gluon density @xmath12 turns out to be compatible with zero in the range of momentum fractions , @xmath14 , accessible to experiments so far .
however , it is still impossible to give a reliable estimate for the total gluon polarization @xmath13 @xcite .
a significant contribution to the integral in eq .
( [ eq : firstmom ] ) can still come from the unexplored small @xmath7 region .
hence , the fundamental question of what constitutes the proton spin still remains largely unanswered , despite the fact that impressive progress , both theoretically and experimentally , was made in the past two decades .
back then , it was discovered that only an unexpectedly small fraction , about a quarter , of the proton s spin can be attributed to the intrinsic spin of quarks and antiquarks @xcite . narrowing down the uncertainties on @xmath12 and , at the same time , extending
the range in @xmath7 continues to be the main objective of experimental efforts in the years to come , utilizing both longitudinally polarized lepton - nucleon and proton - proton scattering . with higher luminosities becoming available at rhic , less inclusive final - states like jet - jet correlations
will be instrumental in achieving this goal as they give a much better handle on the @xmath7 range probed in experiment @xcite .
also , rare probes like prompt photons and heavy quarks come into focus .
both will not be able to compete with single - inclusive pion or jet measurements with respect to statistical precision , but they follow rather different underlying qcd hard scattering dynamics .
therefore , such measurements are crucial for further testing and establishing the universality of helicity - dependent parton densities and hence for our understanding of the spin structure of the nucleon and qcd in general . in this paper , we present a comprehensive phenomenological analysis of open heavy flavor production in longitudinally polarized @xmath0 collisions at rhic . until vertex detector upgrades are in place , rhic experiments will identify heavy quarks through their semi - leptonic decay electron or muon spectra , which receive contributions from both charm and bottom hadron decays , or by direct reconstructions of hadronic @xmath15 meson decays .
since heavy flavors are a versatile probe of high - density medium effects in nucleus - nucleus collisions , such as modifications of the transverse momentum spectra @xcite , various reference data have been taken at rhic in unpolarized @xmath0 collisions @xcite .
similar measurements are intended with longitudinally polarized protons @xcite . to reduce the uncertainties from deconvoluting experimental results for decay lepton spectra back to the heavy quark level
, all theoretical calculations should be done as close as possible to the observational level .
this was achieved , e.g. , in a recent phenomenological study of unpolarized charm and bottom production at rhic @xcite . for the simplest example of a single - inclusive electron spectrum from semi - leptonic decays of a heavy meson @xmath16 , the corresponding invariant cross section takes schematically the following form @xmath17 where the symbol @xmath18 denotes a convolution .
the cross section @xmath19 for the production of a heavy quark @xmath20 with mass @xmath21 , energy @xmath22 , and momentum @xmath23 in ( polarized ) @xmath0 collisions can be evaluated within pqcd .
nlo qcd corrections , which are essential for any meaningful , quantitative analysis , are known , both in the unpolarized @xcite and polarized @xcite case , for quite some time .
we note that the longitudinally polarized hadronic cross section is defined as the combination @xmath24 \ , , \label{eq : polxsec - def}\ ] ] where the subscripts @xmath25 label the helicity states of the colliding hadrons .
the result for @xmath19 depends on the choice of non - perturbative ( helicity - dependent ) parton densities , the value of @xmath21 , and on the unphysical factorization ( @xmath26 ) and renormalization ( @xmath27 ) scales .
the sensitivity of the cross section to variations of @xmath28 can be taken as a rough estimate of the theoretical uncertainty due to the truncation of the perturbative series at a certain order .
likewise , variations of @xmath21 contribute to the theoretical ambiguities as well .
we will assess all these sources of uncertainties in our detailed numerical studies .
the other two ingredients to eq .
( [ eq : xsec - generic ] ) , are the fragmentation @xmath29 of the heavy quark @xmath20 into a heavy meson @xmath16 and the semi - leptonic decay @xmath30 of @xmath16 into the experimentally observed electrons @xmath31 . since @xmath21 cuts off final - state collinear singularities associated with the heavy quark , its hadronization @xmath29 is fundamentally different from those for light quarks and gluons . in the latter case ,
scale - dependent parton - to - hadron fragmentation functions @xcite have to be introduced by virtue of the factorization theorem .
the non - perturbative transition @xmath32 is described by various phenomenological models for a scale independent function @xmath29 , whose parameters are determined from fits to @xmath33 data @xcite . for our phenomenological studies ,
we use the functional form proposed in ref .
@xcite , with its single parameter taken in the range given in @xcite .
in addition , a fixed order pqcd calculation of @xmath19 in eq .
( [ eq : xsec - generic ] ) can be supplemented by all - order resummations of quasi - collinear logarithms of the form @xmath34 @xcite , which can be large if the transverse momentum @xmath35 of the produced heavy quark is much larger than its mass . for the time being , we do not pursue similar resummations for the polarized hadroproduction of heavy quarks since @xmath36 for all phenomenologically relevant applications at rhic .
we adopt the parameterization of the semi - leptonic decay spectrum @xmath30 obtained in ref .
@xcite from a fit to babar and cleo data @xcite and used in unpolarized analyses of heavy quark production at rhic @xcite .
the expression in eq .
( [ eq : xsec - generic ] ) can be easily generalized to the important case of heavy quark or decay lepton correlations .
as will be demonstrated in some detail below , such measurements appear to be more promising for accessing the gluon polarization at rhic than single - inclusive decay electron or muon spectra .
to make theoretical calculations for such observables feasible at nlo accuracy , we develop a flexible parton - level monte carlo program to perform all phase - space integrations numerically . for the subtraction of soft and collinear divergences present at intermediate stages ,
we follow closely the methods devised in ref .
@xcite for the computation of heavy flavor correlations in unpolarized hadron - hadron collisions .
our monte carlo code is capable of computing any infrared safe heavy flavor cross section at @xmath37 , including correlations of the @xmath38 pair and control of the accompanying jet , with the same kinematic cuts as used in experiment .
the hadronization of the @xmath38 pair into heavy mesons and their subsequent semi - leptonic decays are modeled as outlined above .
our results complement and significantly extend previous spin - dependent nlo calculations of single - inclusive heavy ( anti)quark yields and of the heavy quark charge asymmetry based on largely analytical methods @xcite , where any information on the partonic recoil system was lost .
the required spin - dependent matrix elements squared at @xmath37 for producing a @xmath38 pair plus a light parton , @xmath39 as well as the virtual corrections to the @xmath40 born cross sections , @xmath41 can be taken from ref .
expressions obtained in @xmath42 dimensional regularization are required only in the singular regions of phase - space , and @xmath43 can be set to zero otherwise .
after appropriate modifications , the results obtained in this work can be used also as the `` resolved '' photon contribution to the spin - dependent photoproduction of heavy quarks at nlo .
a similar parton - level monte carlo program including both direct and resolved photon processes will be presented in a forthcoming publication @xcite .
it will allow one to include existing data on spin - dependent charm photoproduction @xcite into future global qcd analyses of polarized parton densities .
nlo expressions for the point - like , `` direct '' photon part of the cross section can be adapted from @xcite ; see also @xcite .
the outline of the paper is as follows : in sec .
ii we briefly review some of the technical aspects of setting up a parton - level monte carlo program for heavy flavor production in polarized hadron - hadron collisions at nlo accuracy .
some additional technical details can be found in the appendix . in sec .
iii we present a detailed phenomenological study of heavy flavor production in polarized @xmath0 collisions at rhic , focussing on experimentally relevant decay lepton , electron or muon , distributions and correlations within the kinematic acceptance of the phenix and star experiments at rhic .
we assess theoretical uncertainties due to variations of @xmath26 and @xmath27 , the heavy quark mass @xmath21 , and parameters related to modeling the fragmentation process . finally , we give predictions for double - spin asymmetries and discuss their potential in further constraining helicity - dependent parton densities , in particular , @xmath12 .
we summarize our results in sec . iv .
we consider heavy quark hadroproduction in longitudinally polarized @xmath0 collisions at @xmath37 in qcd .
all phase - space integrations are performed numerically with monte carlo techniques .
this enables us to compute any observable involving heavy quarks within experimental acceptance cuts , including single - inclusive distributions , correlations among the heavy quark - antiquark pair , and , although not pursued in this work , with the associated hard jet present for the first time at @xmath37 .
this significantly extends available calculations @xcite based on largely analytical integrations over the variables characterizing the partons recoiling from the observed single - inclusive heavy quark or antiquark . in the integration of the fully exclusive partonic cross sections for the processes in eqs .
( [ eq : nlo - proc ] ) and ( [ eq : born - proc ] ) at @xmath37 , one has to deal with ultraviolet , infrared , and collinear divergences , which have to be eliminated before any numerical approach can be used . to this end , we follow closely the subtraction method devised and used in ref .
@xcite to compute the unpolarized production of a @xmath38 pair .
the subtraction method is based on adding and subtracting counter terms which approximate the real emission processes in ( [ eq : nlo - proc ] ) in the singular regions of phase - space and are integrable with respect to the momentum of an unresolved parton . in this section ,
we briefly review the technical aspects relevant to extend and apply the subtraction method of @xcite to heavy quark hadroproduction in longitudinally polarized @xmath0 collisions and discuss the numerical implementation .
for further details , we refer the reader to ref .
we note that a general formulation of the dipole subtraction method for nlo calculations with massive partons in qcd and supersymmetric qcd has been developed in @xcite . assuming , as usual , factorization , the cross section ( [ eq : polxsec - def ] ) for producing a heavy @xmath38 pair in longitudinally polarized @xmath0 collisions at a center - of - mass system ( c.m.s . ) energy @xmath44 can be written as a convolution , @xmath45 where the @xmath46 denote the spin - dependent parton distribution functions of flavor @xmath47 at momentum fraction @xmath48 and scale @xmath26 , as defined in eq .
( [ eq : pdfdef ] ) .
the sum in ( [ eq : xsec - fact ] ) is over all contributing partonic processes @xmath49 to @xmath37 with @xmath50 the associated polarized hard scattering cross sections .
they are defined in complete analogy to eq .
( [ eq : polxsec - def ] ) and can be computed perturbatively as a series in the strong coupling @xmath1 .
parton @xmath51 is either a gluon or a light ( anti)quark producing the associated jet possible at @xmath37 . @xmath52
denote the momenta of the heavy quark @xmath20 and antiquark @xmath53 with mass @xmath21 , i.e. , @xmath54 .
the required spin - dependent matrix elements squared at @xmath37 in @xmath42 dimensional regularization for the processes in ( [ eq : nlo - proc ] ) and ( [ eq : born - proc ] ) can be taken from ref .
starting from the nlo level , the subprocess cross sections @xmath50 in ( [ eq : xsec - fact ] ) depend explicitly on the renormalization and factorization scale @xmath27 and @xmath26 , arising from the subtraction of ultraviolet and collinear singularities , respectively .
infrared ( soft gluon ) divergences cancel among real emission and virtual loop corrections . in eq .
( [ eq : xsec - fact ] ) , @xmath55 is the `` measurement function '' used to define the observable one is interested in .
one can think of @xmath55 as being a set of step functions implementing the experimental cuts imposed on the final - state particles and selecting a certain bin in a histogram .
as mentioned in the introduction , charm and bottom quarks are currently detected only indirectly at rhic , mainly through the semi - leptonic decays of the produced heavy @xmath15 and @xmath56 mesons .
thus , the cross section ( [ eq : xsec - fact ] ) at the heavy quark - level is not yet sufficient for comparing theory with experimental results . as indicated in eq .
( [ eq : xsec - generic ] ) , one needs to convolute the parton - level results for @xmath57 with additional phenomenological functions @xmath58 and @xmath59 describing the hadronization into a heavy meson @xmath16 and the semi - leptonic decay of @xmath16 into the observed lepton , respectively .
our flexible parton - level monte carlo program not only performs the phase - space integrations for arbitrary @xmath55 for any infrared safe observable but can also account for the semi - leptonic decays of the heavy quark pair into electrons and muons .
we specify our choice for @xmath58 and @xmath59 in sec .
[ sec : sec3-a ] . for the implementation of eq .
( [ eq : xsec - fact ] ) in a numerically efficient integration it is convenient to express the three - body phase - space and the matrix elements squared for the @xmath60 processes listed in ( [ eq : nlo - proc ] ) in terms of variables in which soft and collinear singularities can be identified easily . instead of choosing the usual set of five independent scalar products ( or mandelstam variables ) of the parton momenta in @xmath49 ,
this is achieved by introducing @xmath7 , @xmath61 , @xmath62 , @xmath63 , and @xmath64 @xcite .
they are defined as follows : @xmath65 , the invariant mass of the @xmath38 pair scaled by the available partonic c.m.s .
energy squared , i.e. , @xmath66 , and @xmath61 is the cosine of the angle between the @xmath67-direction , aligned with the spatial direction of parton @xmath68 , and @xmath69 , the momentum of parton @xmath51 , in the c.m.s . of the incoming partons ,
i.e. , @xmath70 . soft and collinear regions of phase - space are associated with @xmath71 and @xmath72 , respectively .
both @xmath62 and @xmath63 do not matter for this discussion .
they range between @xmath73 and @xmath74 and are used to parameterize the spatial orientation of @xmath52 with respect to the plane span by the other three momenta in the c.m.s . of the @xmath38 pair ,
see @xcite for an explicit parameterization of the momenta . the @xmath75-dimensional three - body phase - space expressed in terms of the variables @xmath7 , @xmath61 , @xmath62 , @xmath63 , and @xmath76 reads @xmath77 which agrees with the result in @xcite and where we have introduced @xmath78^{1/2}$ ] .
@xmath79 represents the gamma function .
the contribution of the @xmath60 real emission processes in ( [ eq : nlo - proc ] ) is then given by @xmath80 where the spin - dependent amplitude squared , @xmath81 , includes the partonic flux factor @xmath82 and is summed over final - state color and spin degrees of freedom and averaged over the color of the interacting partons @xmath47 @xcite .
soft ( @xmath71 ) and collinear ( @xmath83 ) singularities in @xmath81 appear as @xmath84 where @xmath85 is regular for @xmath71 and @xmath83 .
the @xmath86 process in ( [ eq : nlo - proc ] ) can have only collinear singularities at nlo .
due to the finite mass @xmath21 , there can be no collinear gluon radiation from a heavy quark ( `` dead cone '' ) . upon inserting ( [ eq : dps3 ] ) and ( [ eq : m - sing ] ) into ( [ eq : m - ps3 ] )
, one can proceed by expanding the resulting @xmath87 and @xmath88 for small @xmath43 as shown in ref .
@xcite , @xmath89 \frac{(2\omega)^{-\varepsilon}}{2\varepsilon } \\
\nonumber & + & \frac{1}{2 } \left [ \left ( \frac{1}{1-y } \right)_{\omega}+ \left ( \frac{1}{1+y } \right)_{\omega } \right]+ { \cal{o}}(\varepsilon)\,,\\\end{aligned}\ ] ] where @xmath90 .
explicit expressions for the distributions @xmath91_{\tilde{\rho}},\ldots$ ] in ( [ eq : xy - expand ] ) are collected in eq .
( [ eq : distrib ] ) of the appendix .
the choice of the parameters @xmath92 and @xmath93 is to some extent arbitrary and will be discussed at the end of this section . using eq .
( [ eq : xy - expand ] ) , the subprocess cross sections for @xmath49 at @xmath37 can be decomposed as @xmath94 here , @xmath95 and @xmath96 denote the @xmath40 born contribution and the @xmath37 one - loop corrections to the @xmath97 and @xmath98 scattering processes in ( [ eq : born - proc ] ) , respectively .
analytic expressions for the virtual contributions in @xmath75 dimensions , with ultraviolet divergences being subtracted at a renormalization scale @xmath27 , have been obtained in ref .
@xcite . in eq .
( [ eq : xsec - decomp ] ) , @xmath99 is the soft component of the @xmath97 or @xmath98 scattering cross section , which can be either evaluated by explicitly taking the soft gluon limit of the full @xmath75-dimensional matrix elements squared computed in ref .
@xcite or constructed using general properties of soft gluon emission in qcd , see , e.g. , @xcite . in the limit @xmath100 ,
the kinematics simplifies , and phase - space integrations can be performed analytically .
the relevant integrals are the same as for unpolarized heavy flavor hadroproduction and can be found , e.g. , in app .
a of ref .
one obtains @xmath101 explicit expressions for @xmath102 and , for completeness , the standard two - body phase - space factor @xmath103 in @xmath75 dimensions are given in the appendix .
all @xmath60 processes in ( [ eq : nlo - proc ] ) exhibit singularities related to collinear splittings off the incoming partons . again , for such configurations the kinematics collapses to the simpler case of @xmath104 scattering , and these contributions , summarized by @xmath105 in eq .
( [ eq : xsec - decomp ] ) , can be evaluated analytically . as for the soft contribution
, one can either start by taking the collinear @xmath106 limit of the full , @xmath75-dimensional @xmath60 matrix elements taken from ref .
@xcite or by deriving the expressions from scratch . after combining the relevant matrix elements with @xmath107 in eq .
( [ eq : dps3 ] ) , taking the limit @xmath108 , and integrating over @xmath63 one obtains @xmath109 \left(\frac{2}{\omega}\right)^{\varepsilon } \frac{s^{-1-\varepsilon}}{4\varepsilon } d{\mathrm{ps}}_2^x \\ \nonumber & \times & \left[\left ( \frac{1}{1-x } \right)_{\tilde{\rho } } -2\varepsilon \left ( \frac{\log(1-x)}{1-x } \right)_{\tilde{\rho } } \right ] \\ & \times & \delta f_{ab}^{(c\pm)}(s , m_q , x,\theta_1)\,,\end{aligned}\ ] ] where @xmath110 .
the superscript @xmath25 in @xmath105 distinguishes the two configurations with @xmath111 and @xmath112 , where parton @xmath51 is emitted collinearly to the momentum of parton @xmath68 and @xmath113 , respectively .
the relevant @xmath114 are again collected in the appendix . the last term in eq .
( [ eq : xsec - decomp ] ) , @xmath115 , contains all the finite contributions after using the expansions ( [ eq : xy - expand ] ) for @xmath87 and @xmath88 , and the phase - space integration can be performed numerically in four dimensions , i.e. , with @xmath116 .
one obtains @xmath117\\ \nonumber & \times & \beta_x \sin\theta_1 d\theta_1\ , d\theta_2\ , dx\ , dy\\ & \times & \delta f_{ab}(s , m_q , x , y,\theta_1,\theta_2)\,.\end{aligned}\ ] ] as can be seen , all soft and collinear singularities are regulated by the @xmath92- and @xmath93-prescriptions defined in ( [ eq : distrib ] ) . the resulting @xmath118 divergence in ( [ eq : collinear ] ) assumes the form dictated by the factorization theorem , i.e. , a convolution of @xmath75-dimensional helicity - dependent lo splitting functions @xmath119 and born matrix elements @xmath81 .
due to the collinear splitting , the latter have to be evaluated at a `` shifted kinematics '' where parton @xmath68 ( or @xmath113 ) carries only a fraction @xmath7 of its original momentum , i.e. , @xmath120 and @xmath121 ; see eqs .
( [ eq : coll1])-([eq : coll6 ] ) in the appendix , where , for convenience , also the born cross sections and the lo @xmath119 are listed .
collinear singularities are factorized into the bare parton distribution functions at a scale @xmath26 by adding an appropriate `` counter cross section '' to ( [ eq : collinear ] ) which to @xmath37 schematically reads @xmath122\,,\end{aligned}\ ] ] where @xmath123 + \delta g_{ij}(x).\ ] ] the sum in ( [ eq : xsec - counter ] ) is over all possible collinear configurations involving one of the initial - state partons @xmath47 . the argument @xmath124 of the born cross sections @xmath125 in ( [ eq : xsec - counter ] ) indicates that they have to be evaluated at the shifted kinematics as discussed above . in eqs .
( [ eq : xsec - counter ] ) and ( [ eq : trans - func ] ) , the euler constant @xmath126 and @xmath127 , both , like the scale @xmath6 , artifacts of dimensional regularization , are subtracted along with the @xmath118 singularity . the factorization scheme is fully determined by the choice of @xmath128 , for which we take @xmath129 with @xmath130 and @xmath131 otherwise .
this guarantees helicity conservation when the hvbm prescription for @xmath132 in @xmath75 dimensions is adopted to project onto definite helicity states @xcite and defines the @xmath133 scheme in the polarized case @xcite , which we use throughout our calculations . as a consequence of factorization ,
both the hard scattering cross sections and the parton distribution functions in eq .
( [ eq : xsec - fact ] ) depend on the scale @xmath26 which is arbitrary .
@xmath26 can be chosen differently than the renormalization scale @xmath27 at which ultraviolet singularities are absorbed into the bare coupling and heavy quark mass .
note that @xmath134 given in eq .
( [ eq : xsec - soft ] ) receives an additional singular contribution from the soft gluon parts of the diagonal splitting functions @xmath135 and @xmath136 in the factorization procedure , which is proportional to @xmath137 . only then , all remaining singularities cancel in the sum of @xmath134 and @xmath138 , and the full expression for the subprocess cross section @xmath50 in eq .
( [ eq : xsec - decomp ] ) is finite in the limit @xmath116 .
the numerical evaluation of ( [ eq : xsec - fact ] ) for different measurement functions @xmath55 can now be done in parallel with standard monte carlo techniques by randomly generating a large sample of final - state configurations characterized by @xmath139 and @xmath63 .
the @xmath92 and @xmath93-distributions regulating the singularities in @xmath140 and the sum of @xmath141 and @xmath142 in eqs .
( [ eq : xsec - finite ] ) , ( [ eq : collinear ] ) , and ( [ eq : xsec - counter ] ) , respectively , need special attention . to this end ,
one inserts the definitions of distributions , given in eq .
( [ eq : distrib ] ) of the appendix , into eqs .
( [ eq : collinear])-([eq : xsec - counter ] ) and computes for each phase - space point a set of six correlated weights to account for all possible configurations with @xmath71 and @xmath83 .
the values of the measurement functions @xmath55 one is interested in are then multiplied by the appropriate weights and accumulated in different histograms . in principle , the choice for the parameters @xmath143 and @xmath1440,2[$ ] in ( [ eq : distrib ] ) does not matter as it only leads to different values for each of the individual , unphysical contributions at @xmath37 on the right - hand - side of eq .
( [ eq : xsec - decomp ] ) but not for their sum .
large cancellations among the different terms in ( [ eq : xsec - decomp ] ) can take place , however , if @xmath92 is chosen too close to 1 or @xmath93 too close to 0 @xcite . comparison of the results of our monte carlo code , @xmath145 , with the analytical calculation @xmath146 of refs .
@xcite for single - inclusive charm production as a function of transverse momentum @xmath147 and integrated over all rapidities.,scaledwidth=52.0% ] to validate the numerical implementation of ( [ eq : xsec - fact ] ) , we compare the results obtained with the monte carlo techniques outlined above to those of the largely analytical code developed and used in refs . @xcite .
since the analytical calculation is only applicable for single - inclusive heavy quark ( or antiquark ) hadroproduction , the comparison is done for charm production in longitudinally polarized @xmath0 collisions at a c.m.s .
energy of @xmath148 , without any experimental acceptance cuts .
figure [ fig : figure1 ] shows the difference of the numerical results obtained with both codes , labelled as @xmath145 and @xmath149 , as a function of the transverse momentum @xmath147 of the charm quark , normalized to the analytical calculation @xmath149 . as can be seen , deviations are at a level of a few per cent at most , well within the precision of the monte carlo integration for the relatively small sample of phase - space points used in fig .
[ fig : figure1 ] . needless to say that all discussions in this section also apply in the unpolarized case , and the corresponding expressions are obtained by appropriately replacing all helicity - dependent terms by their spin - averaged counterparts .
we fully agree with the results given in @xcite .
having laid out the technical framework in the previous section , we now turn to a detailed phenomenological study of heavy flavor hadroproduction in longitudinally polarized @xmath0 collisions and their semi - leptonic decays at rhic . for comparison and to compute experimentally relevant double - spin asymmetries , defined as @xmath150 we also present results for the corresponding unpolarized quantities .
we study the impact of the nlo qcd corrections on the polarized and unpolarized cross sections and quantify the theoretical uncertainties from different choices for unphysical factorization and renormalization scales , heavy quark masses , and parameters describing the hadronization of the heavy quarks .
we concentrate on observables of immediate relevance for the rhic spin program with collisions of longitudinally polarized protons at a c.m.s .
energy of @xmath148 .
these are single - inclusive transverse momentum distributions of electrons and muons from semi - leptonic decays of charm and bottom quarks , and , in particular , invariant mass spectra for two leptons observed in coincidence .
such measurements have been already carried out in spin - averaged @xmath0 collisions at rhic @xcite and are intended with longitudinally polarized beams once sufficient statistics has been accumulated @xcite .
we note that the leptons can stem from both charm and bottom quark decays which can not be separated experimentally until displaced vertex detector upgrades have been installed .
therefore , our results always refer to the sum of charm and bottom production , their hadronization into @xmath15 and @xmath56 mesons , including @xmath151 , @xmath152 , and `` cascade '' @xmath153 contributions , and the subsequent semi - leptonic decays of the heavy mesons into the observed leptons .
we assume that electrons and muons are detected at central and forward rapidities , @xmath154 and @xmath155 , respectively , which corresponds to the acceptance of the phenix experiment @xcite .
the fragmentation of the heavy quarks into @xmath15 and @xmath56 mesons , i.e. , @xmath156 and @xmath157 , is modeled by phenomenological functions extracted from fits to @xmath33 data @xcite . @xmath15 and
@xmath56 indicate a generic admixture of charm and bottom mesons .
contrary to fragmentation functions for light quarks and gluons into light mesons @xcite , the non - perturbative functions describing the hadronization of heavy quarks are very hard , i.e. , charm and bottom quarks only lose very little momentum when hadronizing .
the main effect of the fragmentation functions is to introduce a shift in the normalization of the heavy meson spectra .
it depends mainly on the average momentum fraction @xmath67 taken by the meson , while the details of the shape of @xmath158 have a negligible effect @xcite .
one can expect that ratios of cross sections , like in the experimentally most relevant double - spin asymmetry ( [ eq : all ] ) , are much less affected by the actual choice of @xmath158 .
we use the functional form of kartvelishvili - likhoded - petrov @xcite with a single parameter @xmath159 controlling the hardness of @xmath160 where @xmath161 to normalize the integral of @xmath158 to one .
we take @xmath162 and @xmath163 from tab . 4 in ref .
@xcite as the default values in eq .
( [ eq : kart ] ) and vary them in the range @xmath164 and @xmath165 , respectively , to estimate the uncertainties associated with the choice of @xmath159 . as in ref .
@xcite , the fragmentation is numerically performed by rescaling the heavy quark s three - momentum by @xmath67 at a constant angle in the laboratory frame , i.e. , @xmath166 . the uncertainty introduced by this particular choice for the `` scaling variable '' @xmath67 ,
which is not uniquely defined for @xmath58 , was shown to be not larger than scale and mass uncertainties @xcite and will be not considered further .
the subsequent semi - leptonic decay of the @xmath15 and @xmath56 mesons into leptons is controlled by another set of phenomenological functions @xmath167 which need to be extracted from data as well .
here we use the spectra obtained in ref .
@xcite based on babar and cleo data @xcite .
we note that we do _ not _ normalize our cross sections with the appropriate branching ratios for @xmath168 , @xmath169 , etc .
, which are all close to 10@xmath170 @xcite . of course , branching ratios drop out of experimentally relevant double - spin asymmetries ( [ eq : all ] ) . the main motivation
to study heavy flavor production with polarized beams at rhic is the expected sensitivity to the helicity - dependent gluon density through the tree - level gluon - gluon fusion process , @xmath171 , which is known to be dominant for unpolarized collisions up to the largest values of the heavy quark s transverse momentum currently accessible at rhic @xcite .
we will show , however , that the fractional contribution of gluon - gluon fusion to the spin - dependent cross section depends crucially on the assumed set of polarized parton densities .
our default choice is the dssv set @xcite , obtained in a global qcd analysis of the latest spin - dependent data , including those from rhic on single - inclusive pion and jet production @xcite . due to the smallness of @xmath172 in the dssv set and a node in the @xmath7-shape near @xmath173 @xcite
, the @xmath98 annihilation subprocess , @xmath174 , turns out to be the dominant mechanism for charm and bottom production in polarized @xmath0 collisions at rhic .
this is in sharp contrast to naive expectations based on unpolarized results . for comparison and to study the sensitivity to @xmath172
, we adopt also two alternative sets of spin - dependent parton densities , grsv(std ) @xcite and dns(kre ) @xcite , both characterized by a positive gluon polarization of moderate size . in general , for gluon polarizations from current qcd fits @xcite , the double - spin asymmetries for leptons from heavy flavor decays all turn out to be small , often well below the one percent level , making their measurement very challenging .
this is in particular true for single - inclusive lepton observables ; see below .
heavy flavor production at rhic can not compete with the statistical precision achievable for more abundant probes of the nucleon s spin structure , like pions and jets @xcite , which are already used in global fits @xcite .
nevertheless , measurements of double - spin asymmetries related to heavy flavor production will be crucial for further testing and establishing the all important concept of factorization and universality for helicity - dependent scattering processes and parton densities , respectively .
the underlying dynamics of the partonic scattering processes , i.e. , @xmath171 and @xmath174 , is very much different as compared to the multitude of qcd processes driving the production of light hadrons @xcite or jets @xcite . in the computation of the nlo
unpolarized cross sections in ( [ eq : all ] ) , which proceeds along similar lines as outlined in sec .
ii , for details , see ref .
@xcite , we use the nlo cteq6 m parton densities @xcite and values for the strong coupling @xmath1 .
since the dssv analysis @xcite does not provide a lo set of spin - dependent parton distributions , our lo results always refer to the born part of the full nlo calculation , i.e. , they are computed with nlo parton densities and values for @xmath1 .
strictly speaking this is , of course , inconsistent as it introduces some unwanted scheme dependence into a tree - level quantity .
nevertheless , the lo results should give a faithful estimate of the relevance of nlo corrections .
as will be demonstrated below , they turn out to be sizable and rather different for unpolarized and polarized cross sections such that they do not cancel in experimentally relevant double - spin asymmetries .
we take @xmath175 and @xmath176 as reference values for the charm and bottom quark mass and vary them in the range @xmath177 and @xmath178 , respectively , to estimate the resulting mass uncertainties . for the factorization and renormalization scale
we take @xmath179/2)^{1/2}$ ] with @xmath180 as the central value . as usual , the sensitivity of the cross section to missing higher order corrections is estimated by varying @xmath26 and @xmath27 simultaneously in the range @xmath181 .
following the procedure used for unpolarized charm and bottom production at rhic in ref .
@xcite , we also vary @xmath26 and @xmath27 independently in the same range of @xmath182 and combine the ensuing uncertainty with the one stemming from variations of @xmath183 in quadrature . unless stated otherwise , we use the central values for @xmath26 , @xmath27 , @xmath183 , and @xmath184 given above .
scale dependence of the single - inclusive transverse momentum spectrum of electrons from heavy quark decays at central rapidities @xmath185 in unpolarized ( upper panel ) and polarized ( middle panel ) @xmath0 collisions at rhic .
all scales are varied simultaneously , i.e. , @xmath179/2)^{1/2}$ ] , in the range @xmath186 ( shaded bands ) .
the solid lines correspond to the default choice @xmath180 .
note that the lo results are rescaled by a factor of 0.1 , and all polarized cross sections are multiplied by @xmath187 .
the lower panel shows the ratio of nlo to lo polarized and unpolarized cross sections ( @xmath188-factor).,scaledwidth=52.0% ] we begin our detailed numerical studies with a discussion of unpolarized and polarized cross sections for various decay lepton distributions accessible at rhic .
figure [ fig : figure2 ] shows the single - inclusive transverse momentum spectrum of electrons from charm and bottom decays in lo and nlo accuracy , integrated over the angular acceptance of the phenix detector , i.e. , @xmath154 @xcite .
similar results are obtained for the star experiment @xcite with its larger acceptance for electrons at central rapidities , @xmath189 , and hence not shown .
the transverse momentum @xmath190 is limited to a region which should be accessible with luminosities envisaged in longitudinally polarized @xmath0 collisions at rhic .
photon conversion , @xmath191 , and @xmath192 dalitz decays are the dominant source of electron background for such measurements and may require an additional cut @xmath193 @xcite .
recall that the branching ratios of about @xmath194 are not included in the cross sections shown in fig .
[ fig : figure2 ] .
the solid lines are obtained with the default values of the heavy quark masses , scales , parameters , and parton densities stated in the previous subsection .
the shaded bands indicate the theoretical uncertainty from varying the factorization and renormalization scale simultaneously in the range @xmath181 specified above .
note that the lo results are rescaled by a factor of 0.1 , and all polarized cross sections are multiplied by -1 to display them on a logarithmic scale .
the bottom panel of fig .
[ fig : figure2 ] gives the resulting unpolarized and polarized `` @xmath188-factors '' , defined as usual by the ratio @xmath195 one notices that the nlo corrections are sizable in the unpolarized case , @xmath196 , but moderate for polarized @xmath0 collisions , except for the region @xmath197 .
here , the polarized cross section approaches a node , and perturbative corrections are artificially enhanced .
same as in fig .
[ fig : figure2 ] but for the single - inclusive transverse momentum spectrum of muons from heavy quark decays at forward rapidities @xmath198.,scaledwidth=52.0% ] less pronounced nlo corrections for polarized cross sections are a rather generic feature and have been observed already for other hadronic processes such as single - inclusive pion @xcite and jet @xcite production . to some extent
this behavior can be traced back to the less singular scale evolution of polarized parton densities at small momentum fractions @xmath7 @xcite .
this has the effect that the partonic threshold region , which is the source of large logarithmic corrections associated with the emission of soft gluons , is less emphasized in the convolution ( [ eq : xsec - fact ] ) than in the unpolarized case .
specifically for heavy flavor production , it was noticed in @xcite that large nlo corrections to the gluon - gluon fusion process related to amplitudes with a gluon exchange in the @xmath199-channel are independent of the helicities of the interacting gluons and hence do not contribute to the polarized cross section .
substantially different @xmath188-factors for unpolarized and polarized cross sections immediately imply that born level estimates for double - spin asymmetries ( [ eq : all ] ) can serve only as very rough estimates . in general , they are insufficient for any quantitative analysis such as a global qcd extraction of spin - dependent parton densities .
same as in fig .
[ fig : figure2 ] but for the invariant mass spectrum of electron - muon correlations from heavy quark decays .
electrons are detected at central rapidities , @xmath185 , and muons are detected at forward rapidities , @xmath198 .
the transverse momenta of both electrons and muons are required to be larger than @xmath200.,scaledwidth=52.0% ] as can be inferred from fig .
[ fig : figure2 ] , theoretical ambiguities due to the choice of @xmath27 and @xmath26 in ( [ eq : xsec - fact ] ) are reduced in the polarized but not in the unpolarized case . along with the observation of large qcd corrections , this indicates the need for next - to - next - to - leading order corrections for the unpolarized cross section to better control the dependence on @xmath26 and @xmath27 .
we note that substantial progress toward this direction has already been made recently @xcite , mainly to allow for precision studies with the large amount of top quarks expected to be produced at the cern - lhc . in addition , fixed order calculations need to be amended by all - order resummations if @xmath201 becomes large .
this was achieved in @xcite but is not really relevant for our discussions here since we are mainly interested in the region where @xmath202 .
we postpone a discussion of theoretical uncertainties due to the choice of @xmath21 and @xmath159 in eq .
( [ eq : kart ] ) , as well as the effect of varying @xmath27 and @xmath26 independently , until the end of this subsection . the single - inclusive transverse momentum spectrum of muons from heavy quark decays is shown in fig . [ fig : figure3 ] in lo and nlo accuracy .
the pseudorapidity @xmath203 of the muon is integrated in the range @xmath198 corresponding to the angular acceptance of the phenix experiment .
all observations made in fig . [ fig : figure2 ] regarding the relevance of nlo corrections , the behavior of the @xmath188-factor , and the dependence on @xmath28 apply also here .
the polarized @xmath188-factor stays even closer to one than in fig .
[ fig : figure2 ] as @xmath204 develops no node in the @xmath205 range shown .
same as in fig .
[ fig : figure2 ] but for the invariant mass spectrum of muon - muon correlations from heavy quark decays . both muons are detected at forward rapidities , @xmath198 , but in different hemispheres .
the transverse momentum of each muons is required to be larger than @xmath200.,scaledwidth=52.0% ] by glancing at the relative sizes of the unpolarized and polarized single - inclusive transverse momentum spectra for electrons and muons shown in fig . [
fig : figure2 ] and [ fig : figure3 ] , respectively , it becomes immediately obvious that the corresponding double - spin asymmetries @xmath206 and @xmath207 , to which we turn to in subsec .
[ sec : all ] , are very small if the most up - to - date dssv parton densities @xcite are used .
expected asymmetries of the order of a few tenths of a percent are extremely challenging experimentally as systematic uncertainties , like from the determination of the relative beam luminosities at rhic , are of similar size @xcite . at higher @xmath208 , where double - spin asymmetries are largest , the single - inclusive cross sections in figs .
[ fig : figure2 ] and [ fig : figure3 ] have dropped already several orders of magnitude from their peak values , and measurements require substantial integrated luminosities .
more promising appear to be observables where both the heavy quark and the heavy antiquark decay semi - leptonically , and both leptons are observed in coincidence .
this is also where our numerical phase - space integration and the flexible monte carlo code for polarized heavy flavor hadroproduction introduced in sec .
[ sec : tech ] become truly essential .
particle correlations are hard , and often impossible , to compute at nlo with largely analytical methods , see , e.g. , @xcite .
figure [ fig : figure4 ] shows our results for the invariant mass spectrum of electron - muon correlations from semi - leptonic decays of @xmath15 and @xmath56 mesons within the angular acceptance of the phenix detector , i.e. , @xmath185 and @xmath198 .
in addition , we require a minimum transverse momentum for both leptons of @xmath200 as required by experiment . as in figs . [ fig : figure2 ] and [ fig : figure3 ] , results shown as solid lines are obtained with the default choice of parameters .
again , shaded bands indicate the theoretical uncertainty from varying the factorization and renormalization scales simultaneously in the range @xmath181 . scale and mass uncertainties for the polarized invariant mass spectrum of electron - muon correlations at nlo accuracy , using the same cuts as in fig .
[ fig : figure4 ] .
the upper and lower panel shows the contribution from @xmath209 and @xmath210 decays , respectively , multiplied by @xmath187 .
the dashed lines indicate the range of uncertainties for @xmath211 , @xmath212 , and central values of @xmath183 as in fig .
[ fig : figure4 ] .
the solid curves are for @xmath180 .
the effect of varying @xmath26 and @xmath27 independently in the same range of @xmath182 for fixed @xmath213 , @xmath214 and for @xmath215 , @xmath216 is illustrated by hatched and solid bands , respectively . in the latter case ,
scale and mass uncertainties are combined in quadrature ( see text).,scaledwidth=52.0% ] the @xmath188-factor , shown in the lower panel of fig .
[ fig : figure4 ] , is smaller than what was found for the single - inclusive observables in figs .
[ fig : figure2 ] and [ fig : figure3 ] in the unpolarized case .
still , nlo corrections differ considerably for the unpolarized and polarized invariant mass spectra .
again , the corrections are such that the corresponding double - spin asymmetry is reduced at nlo accuracy .
the scale uncertainty is significantly smaller for the spin - dependent cross section with nlo corrections included .
the improvement in the helicity - averaged case is much less pronounced .
compared to the single - inclusive results in figs .
[ fig : figure2 ] and [ fig : figure3 ] , the cross sections obtained for the electron - muon invariant mass spectrum are smaller , but @xmath217 drops much less with increasing @xmath218 than @xmath219 with increasing @xmath190 or @xmath205 .
this makes measurements of @xmath220 at comparatively large values of @xmath218 feasible .
we do not consider here correlations with back - to - back electrons at central rapidities .
electron - muon correlations are phenomenologically more interesting due to their asymmetric kinematics with respect to rapidity , probing the interacting partons at different momentum fractions @xmath7 as will be demonstrated below .
the corresponding invariant mass spectrum for two muons from @xmath15 and @xmath56 meson decays observed in coincidence is shown in fig .
[ fig : figure5 ] .
both muons are required to have @xmath198 and @xmath221 , with one muon detected at forward ( positive ) and one muon detected at backward ( negative ) pseudorapidities .
again , the cross sections @xmath222 decrease rather slowly with increasing invariant mass @xmath223 .
this observable is very demanding in terms of required monte carlo statistics as can be seen by the still fairly pronounced fluctuations , most noticeable in the unpolarized @xmath188-factor .
the general trend and features of the cross sections are , however , reliable . as before ,
nlo corrections are more significant in the unpolarized case , where @xmath224 .
unfortunately , the reduction of the theoretical ambiguities related to the choice of @xmath28 is only marginal at nlo .
we now turn to a more detailed discussion of theoretical uncertainties for the observables discussed in this section , taking the phenomenologically interesting invariant mass spectrum for electron - muon correlations , presented in fig .
[ fig : figure4 ] , as an example .
qualitatively very similar results are obtained for the other cross sections given in figs .
[ fig : figure2 ] , [ fig : figure3 ] , and [ fig : figure5 ] and hence not shown here .
the impact of varying @xmath26 and @xmath27 independently is shown in fig .
[ fig : figure5a ] for the electron - muon invariant mass spectrum in polarized @xmath0 collisions at nlo accuracy .
since we are also interested in variations of @xmath183 , the contribution from @xmath209 and @xmath210 decays are shown in separate panels and add up to @xmath225 discussed in fig .
[ fig : figure4 ] . following ref .
@xcite , we compute our results for seven different settings of scales @xmath226/2)^{1/2}$ ] , using @xmath227 and keeping @xmath228 fixed to their central values @xmath213 and @xmath214 . the envelope of all resulting curves defines the scale uncertainty and is show as hatched bands in fig . [
fig : figure5a ] .
for comparison , the dashed lines indicate the range of uncertainties for standard choice @xmath211 used in fig .
[ fig : figure4 ] .
as can be seen , taking @xmath229 does not significantly enlarge the scale ambiguities for the polarized cross section , in particular , for the contribution from bottom quarks shown in the lower panel .
dependence of the polarized ( solid lines ) and unpolarized ( dashed lines ) invariant mass spectra for electron - muon correlations at nlo accuracy on the choice of fragmentation parameters @xmath230 ( upper panel ) and @xmath231 ( lower panel ) defined in eq .
( [ eq : kart ] ) . displayed
are the relative deviations for @xmath209 and @xmath210 decays using @xmath232 and @xmath233 with respect to the cross sections obtained for our default values @xmath162 and @xmath163 @xcite , respectively .
@xmath26 , @xmath27 , and heavy quark masses are taken at their central values , and the same experimental cuts as in fig .
[ fig : figure4 ] are adopted.,scaledwidth=52.0% ] the solid bands in fig .
[ fig : figure5a ] take also variations of @xmath234 and @xmath235 into account .
the recipe we follow here is again similar to the one used to estimate theoretical uncertainties for unpolarized heavy flavor production @xcite . in practice ,
we add scale and mass uncertainties in quadrature , and the envelope of all results is defined by @xmath236^{1/2}$ ] and @xmath237^{1/2}$ ] . here
, @xmath238 denotes the results obtained for central values of scales and masses .
@xmath239 ( @xmath240 ) are the maximum ( minimum ) cross sections computed for @xmath229 , @xmath213 , and @xmath214 , as depicted by the hatched bands .
correspondingly , @xmath241 ( @xmath242 ) denote the maximum ( minimum ) cross sections for @xmath243 and varying @xmath234 and @xmath235 in the range @xmath215 and @xmath216 , respectively . in general
, the combined uncertainties are much smaller for @xmath210 than for @xmath209 production and decays , which is not too surprising . in both cases , variations of @xmath21
add noticeably to the theoretical uncertainties .
the dependence of the cross sections on the choice of @xmath159 in the non - perturbative function @xmath158 describing the hadronization of the heavy quarks into @xmath15 and @xmath56 mesons , see eq .
( [ eq : kart ] ) , is illustrated in fig .
[ fig : figure5b ] .
again , we take the invariant mass spectrum for electron - muon correlations as an representative example .
we vary @xmath230 and @xmath231 in the range @xcite @xmath244 and @xmath245 , respectively , and show the impact on the invariant mass spectrum as relative uncertainty with respect to the results obtained for the central values @xmath162 and @xmath163 used in fig . [
fig : figure4 ] .
it turns out that polarized and unpolarized invariant mass spectra are affected very much in the same way by variations of @xmath184 . for charm production and taking @xmath244
, it roughly amounts to a shift in the normalization of the cross sections by @xmath246 .
the impact of varying @xmath231 on the contribution to @xmath217 from bottom decays is significantly smaller , up to about @xmath247 deviation from the results for @xmath163 , but is less uniform with @xmath218 .
these observations have the important implication that theoretical uncertainties associated with the actual choice of @xmath184 drop out to a large extent for experimentally relevant double - spin asymmetries @xmath248 discussed in sec .
[ sec : all ] below .
fractional amount of different partonic subprocesses at nlo accuracy ( left column ) and of charm , bottom , and cascade ( @xmath249 ) decays ( right column ) contributing to the single - inclusive transverse momentum spectrum of electrons shown in fig .
[ fig : figure2 ] . results are shown for unpolarized ( upper row ) and polarized ( middle and lower rows ) @xmath0 collisions at rhic using the cteq6 @xcite , dssv @xcite , and grsv @xcite set of parton densities , respectively.,scaledwidth=52.0% ] we now take a detailed look at the fractional contributions of the different partonic hard scattering processes to the cross sections shown in figs . [
fig : figure2 ] - [ fig : figure5 ] .
this will help to understand the dependence of the double - spin asymmetries on different sets of polarized parton densities , to be discussed in the next subsection .
since charm and bottom decays both contribute to the lepton spectra , we also present their fractional contributions .
this includes also the `` cascade '' decay @xmath250 , which is modeled following the procedure discussed in @xcite .
it is found to be negligible for all observables we are interested in . the left - hand - side of fig .
[ fig : figure6 ] shows the contributions of the three different subprocesses at nlo accuracy , with @xmath97 , @xmath98 , and @xmath86 initial - states , to the single - inclusive decay electron spectra shown in fig .
[ fig : figure2 ] . in the unpolarized case ( upper panel ) ,
gluon - gluon fusion is the by far dominant subprocess for heavy flavor production at rhic energies , with @xmath98 annihilation becoming somewhat more relevant at larger values of transverse momentum @xmath190 .
interestingly enough , the genuine nlo , i.e. , @xmath1 suppressed , @xmath86 scattering process also contributes very significantly at larger @xmath190 , even exceeding the @xmath98 annihilation cross section .
this observation can be linked to the abundance of gluons at all momentum fractions @xmath7 @xcite .
this implies that the partonic flux relevant for @xmath86 scattering , i.e. , @xmath251 , is much larger than the corresponding flux for @xmath98 annihilation , in particular , at the medium - to - large momentum fractions @xmath252 relevant for rhic .
this compensates for the @xmath253 suppression in the @xmath86 hard scattering channel . in @xmath254 scattering , e.g. , at the tevatron , where antiquarks are `` valence '' quarks in the antiproton beam , this is different , and the @xmath98 flux is much enhanced .
a similar observation concerning the relevance of the @xmath98 annihilation channel was made also for fixed - target experiments in ref .
@xcite , where it is expected to contribute very significantly to charm hadroproduction in proposed @xmath254 collisions at the gsi - fair facility , but not in @xmath0 scattering of similar c.m.s .
energy planned at j - parc . in general
, the situation is much more involved in the polarized case , where both hard scattering cross sections and parton densities are not positive definite and can contribute with either sign , depending on the kinematics relevant for a particular process . in the vicinity of sign changes , large cancellations are to be expected .
as we shall demonstrate below , depending on the chosen set of polarized parton densities , the subprocess fractions can differ considerably from each other and often gluon - gluon fusion does not dominate , in contrast to the unpolarized case . the middle panel of fig .
[ fig : figure6 ] shows our results for the polarized subprocess fractions obtained with the dssv set @xcite , our default choice of parton densities used in figs .
[ fig : figure2]-[fig : figure5 ] . at small @xmath190 ,
the cross section is entirely dominated by @xmath98 annihilation , contrary to the unpolarized case . towards larger @xmath190 , both @xmath97 and
@xmath86 processes contribute significantly but with opposite sign , leading to strong cancellations .
this happens , however , in a kinematic region close to a sign change of the cross section at @xmath255 .
a rather different pattern of fractional subprocess contributions can be found in the bottom panel of fig .
[ fig : figure6 ] , where the grsv(std ) @xcite parton densities were used .
note that the cross section has a sign change near @xmath256 .
this explains the complicated behavior of the ratios in this region and makes it very awkward to display them properly .
like in the unpolarized case , gluon - gluon fusion is the most important contribution to the cross section . at larger
@xmath190 , @xmath98 and @xmath86 subprocesses become more relevant , both contributing with the opposite sign than @xmath97 scattering , leading again to fairly significant cancellations .
same as in fig .
[ fig : figure6 ] but now for the single - inclusive transverse momentum spectrum of muons shown in fig .
[ fig : figure3].,scaledwidth=52.0% ] the gross features of the results in fig .
[ fig : figure6 ] obtained with dssv and grsv parton densities can be readily understood by comparing the size and sign of the individual quark , antiquark , and gluon densities in both sets , see , e.g. , fig . 2 in ref .
@xcite . since the decay electrons stem from heavy ( anti)quarks produced at central rapidities , the interacting partons have very similar momentum fractions , i.e. , @xmath257 .
therefore , @xmath258 , irrespective of the node in the dssv gluon distribution , and the sign of the @xmath97 contribution follows the sign of the hard scattering cross section , which changes from positive at small @xmath259 to negative at larger @xmath259 values .
same as in fig .
[ fig : figure6 ] but now for the invariant mass spectrum of electron - muon correlations shown in fig .
[ fig : figure4].,scaledwidth=52.0% ] due to helicity conservation , @xmath260 @xcite , and the sign of the @xmath98 contribution depends on the individual parton densities for each quark and antiquark flavor . all sets of polarized parton densities have @xmath261 and @xmath262 , resembling the features of the naive quark model .
the grsv(std ) set @xcite assumes an @xmath263 symmetric sea , with all antiquark polarizations being negative , such that for the dominant @xmath264-quarks one has @xmath265 , resulting in a net positive contribution to the cross section .
this is exactly opposite in the dssv set @xcite , where @xmath266 , unless @xmath267 gets very large .
the genuine nlo @xmath86 subprocess cross section , as well as the sum of all quark and antiquark polarizations , @xmath268 $ ] , are both positive .
this implies that the sign of the @xmath86 contribution depends on the sign of @xmath172 in the relevant region of @xmath7 , which turns out to be positive for both grsv(std ) and dssv .
overall , the fractional contributions of the individual subprocesses to the single - inclusive decay electron spectrum are essentially controlled by the modulus of the polarized gluon density , @xmath269 , which is much larger for the grsv(std ) set , i.e. , @xmath270 .
the bigger the gluon density , the closer the result is to what we have found in the unpolarized case .
it turns out that even for the moderate gluon polarization of the grsv(std ) set , the gluon - gluon channel prevails for all @xmath190 values shown in fig .
[ fig : figure6 ] . the right - hand - side of fig .
[ fig : figure6 ] shows the fractional contributions of the charm , bottom , and `` cascade '' @xmath249 decays to the single - inclusive transverse momentum spectrum of electrons . for @xmath271 ,
almost all electrons originate from charm decays , but above the bottom contribution catches up , yielding about @xmath272 at @xmath273 in the unpolarized case shown in the upper panel of fig .
[ fig : figure6 ] . eventually , at somewhat larger values of @xmath190 , it becomes dominant , as was shown in @xcite .
same as in fig .
[ fig : figure8 ] but now for the invariant mass spectrum of muon - muon correlations shown in fig .
[ fig : figure5].,scaledwidth=52.0% ] as for the subprocess fractions , the corresponding results in the polarized case depend on the choice of parton densities , and cancellations and possible sign changes of the individual hard scattering subprocesses @xmath50 further complicate their interpretation . the only feature common to both the unpolarized and the polarized inclusive electron spectra is the smallness of the @xmath274 cascade decay contribution .
the results obtained with the dssv set ( middle panel ) show a very sizable bottom contribution , exceeding the @xmath275 decay above @xmath276 .
however , this is due to a sign change of the @xmath275 cross section at @xmath255 , and above , @xmath275 and @xmath277 contribute on equal footing . choosing the grsv(std ) distributions instead ( lower panel ) , both the @xmath275 and the @xmath277 cross sections change from positive to negative at 2 and @xmath278 , respectively , with @xmath277 starting to be the dominant contribution above @xmath279 . the fractional contributions of the different partonic hard scattering processes ( left - hand - side ) and heavy flavor decays ( right - hand - side ) to the single - inclusive muon cross section shown in figs .
[ fig : figure3 ] are given in fig .
[ fig : figure7 ] .
the results are qualitatively very similar to the ones depicted in fig .
[ fig : figure6 ] and discussed above .
again , the polarized subprocess fractions very much depend on the choice of parton densities , and the interpretation is obscured by sign changes and large cancellations among the different contributions .
compared to the single - inclusive electron spectrum at central rapidities , bottom decays contribute less to the muon transverse momentum spectrum at @xmath198 , even up to @xmath280 .
its contribution is rather flat with respect to @xmath205 and amounts to about @xmath281 ( @xmath282 ) in the ( un)polarized case .
double - spin asymmetry @xmath206 for single - inclusive electrons from charm and bottom decays at rhic , computed at nlo accuracy for three different sets of polarized parton densities : dssv @xcite ( solid line ) , grsv(std ) @xcite ( dashed line ) , and dns(kre ) @xcite ( dot - dashed line ) .
electrons are restricted to central rapidities @xmath154.,scaledwidth=52.0% ] corresponding results for the electron - muon and muon - muon invariant mass distributions , shown in fig .
[ fig : figure4 ] and [ fig : figure5 ] , can be found in fig .
[ fig : figure8 ] and [ fig : figure9 ] , respectively . as before
, subprocess fractions can be found on the left - hand - side and contributions from different heavy flavor decays on the right - hand - side of the plots .
note that in all panels of figs .
[ fig : figure8 ] and [ fig : figure9 ] , the bin corresponding to the smallest invariant mass only has a small number of entries due to the cuts @xmath283 , and the numerical results for that bin should be taken with caution . in general ,
cancellations among different subprocesses are found to be less pronounced in figs .
[ fig : figure8 ] and [ fig : figure9 ] , except for small invariant masses , say , below @xmath278 , where a sign change in the polarized cross section occurs .
gluon - gluon fusion is even more dominant for electron - muon and muon - muon correlations than for single - inclusive decay lepton observables , with both @xmath98 and @xmath86 subprocesses being negligible in the unpolarized case ( upper row ) .
this is also the case for the polarized cross section if the grsv ( std ) parton distributions are chosen . for the dssv set , @xmath98 annihilation remains dominant for electron - muon correlations , but gluon - gluon fusion contributes significantly to muon - muon correlations for @xmath284 . as will be shown in the next subsection , back - to - back muon - muon correlations with @xmath198 probe on average fairly large momentum fractions , @xmath285 , where the dssv @xmath172 is positive and larger than the sea quark polarizations , such that @xmath286 . since the dssv @xmath287 turns negative at large @xmath7 , there are additional cancellations among the different quark flavors in the @xmath98 annihilation channel , as @xmath265 and @xmath288 .
same as in fig .
[ fig : figure10 ] but for the invariant mass spectrum for electron - muon correlations .
electrons and muons are restricted to @xmath154 and @xmath155 , respectively .
in addition , we demand @xmath289.,scaledwidth=52.0% ] same as in fig . [ fig : figure10 ] but for the invariant mass spectrum for electron - muon correlations . electrons and muons are restricted to @xmath154 and @xmath155 , respectively .
in addition , we demand @xmath289.,scaledwidth=52.0% ] as in figs .
[ fig : figure6 ] and [ fig : figure7 ] , the contribution from the cascade @xmath290 decay is found to be negligible for lepton - lepton correlations . in particular ,
the muons in back - to - back correlations originate mainly from charm decays , as can be seen on the right - hand - side of fig .
[ fig : figure9 ] . at the same invariant lepton - lepton mass ,
bottom quark decays contribute more significantly to electron - muon correlations than to muon - muon correlations for @xmath291 .
same as in fig .
[ fig : figure10 ] but for the invariant mass spectrum for muon - muon correlations .
the muons are restricted to forward rapidities @xmath155 and have to be in opposite hemispheres .
in addition , we demand @xmath292.,scaledwidth=52.0% ] the quantities of actual interest in experiments exploiting polarized beams and targets are the double - spin asymmetries @xmath248 defined in eq .
( [ eq : all ] ) .
experimental normalization uncertainties conveniently cancel to a large extent in the ratio ( [ eq : all ] ) . in general
, this does not happen for higher order qcd corrections or the various sources of theoretical ambiguities as we shall demonstrate below .
nevertheless , it is often erroneously assumed that lo estimates for @xmath248 give reliable results which can be used in quantitative qcd analyses .
we start by giving expectations for various double - spin asymmetries at nlo accuracy in figs .
[ fig : figure10 ] - [ fig : figure13 ] , based on the polarized and unpolarized cross sections for decay lepton transverse momentum and invariant mass distributions presented in figs .
[ fig : figure2 ] - [ fig : figure5 ] . apart from our default choice of dssv
polarized parton densities @xcite , which leads to very small asymmetries throughout , we adopt also two alternative sets , grsv(std ) @xcite and dns(kre ) @xcite . unlike dssv
, both sets are characterized by a positive gluon polarization of moderate size and an almost @xmath263 symmetric sea .
we refrain from using outdated models with a large , but strongly disfavored gluon polarization in the @xmath7 range already probed by rhic @xmath0 and fixed target data @xcite . with the exception of the double - spin asymmetry @xmath206 for single - inclusive electrons from charm and bottom
decays shown in fig .
[ fig : figure10 ] , differences in the results obtained with grsv(std ) and dns(kre ) parton distributions are readily explained by the slightly larger @xmath172 in the grsv set . for @xmath206 , the result based on the dns(kre ) set is strongly affected by cancellations between the @xmath97 subprocess on the one hand , and the @xmath98 , @xmath86 processes on the other hand , leading to an essentially zero spin asymmetry in the @xmath190 range shown .
cancellations among the different subprocesses contributing with different sign are less pronounced for the grsv set , see the lower left panel of fig .
[ fig : figure6 ] , due to a significantly less negative @xmath293 density at @xmath173 .
cancellations in conjunction with the smallness of @xmath206 for all @xmath190 make this observable not really suited for studies of the nucleon s spin structure . the double - spin asymmetry @xmath207 for single - inclusive decay muons at @xmath155 , presented in fig .
[ fig : figure11 ] , shows a much better correlation of the size of @xmath172 and @xmath207 . as was demonstrated in fig .
[ fig : figure7 ] , cancellations among the different subprocesses are less pronounced than for @xmath206 . for the same value of transverse momentum , the obtained spin asymmetries are about a factor of two larger for @xmath294 than for @xmath295 . however , they are still significantly smaller than projections based on very large positive , but outdated gluon polarizations like grsv(max ) @xcite , see , e.g. , fig . 3 in ref .
@xcite . nevertheless
, with sufficient statistics accumulated , spin asymmetries of @xmath296 should be measurable . based on the cross sections given in figs .
[ fig : figure2 ] and [ fig : figure3 ] , we estimate that an integrated luminosity of about @xmath297 is required . the best suited observables related to heavy flavor hadroproduction in longitudinally polarized @xmath0 collisions at rhic are double - spin asymmetries for invariant mass spectra of electron - muon and muon - muon correlations shown in fig .
[ fig : figure12 ] and [ fig : figure13 ] , respectively . at small invariant mass ,
the corresponding cross sections are smaller than for single - inclusive transverse momentum distributions at similar values of @xmath208 , but fall off much slower with increasing invariant mass , cf . figs .
[ fig : figure2 ] - [ fig : figure5 ] .
this should allow for measurements of @xmath220 and @xmath298 up to @xmath299 , where spin asymmetries can be up to the @xmath300 level for the grsv(std ) and dns(kre ) parton densities .
an integrated luminosity of a few hundred @xmath301 should be sufficient .
in addition , the size of @xmath302 and @xmath172 in the relevant range of momentum fractions @xmath7 , see below , are nicely correlated . in the absence of shifted vertex detectors at the rhic experiments , @xmath298 is the observable with the cleanest sample of charm decays for all @xmath223 shown in fig .
[ fig : figure13 ] and irrespective of the set of polarized parton densities used in the calculations , cf . fig .
[ fig : figure9 ] .
impact of nlo qcd corrections on the double - spin asymmetries shown in figs .
[ fig : figure10 ] - [ fig : figure13 ] .
depicted is the @xmath188-factor , i.e. , @xmath303 , computed in each case using the dssv polarized and cteq6 unpolarized parton densities.,scaledwidth=50.0% ] impact of variations of the scales @xmath179/2)^{1/2}$ ] on the double - spin asymmetry @xmath220 for electron - muon correlations at nlo accuracy .
the upper panel shows @xmath220 computed with the dssv parton densities for @xmath180 ( solid line ) and in the range @xmath186 ( shaded band ) .
the lower panel gives the relative deviation of @xmath220 for @xmath304 with respect to the result obtained for our default value @xmath180.,scaledwidth=52.0% ] figure [ fig : figure14 ] clearly illustrates the inadequacy of computing double - spin asymmetries based on lo estimates of heavy flavor cross sections .
depicted is the @xmath188-factor , i.e. , @xmath303 , for all double - spin asymmetries presented in figs .
[ fig : figure10 ] - [ fig : figure13 ] , computed in each case using the dssv polarized and cteq6 unpolarized parton densities . other sets of polarized parton densities yield qualitatively very similar results . on average , lo estimates for @xmath248 are about a factor of two larger than corresponding calculations at nlo accuracy and depend , in case of the single - inclusive observables @xmath206 and @xmath207 , strongly on @xmath208 .
this reflects the difference of @xmath188-factors for the polarized and unpolarized cross sections found in figs .
[ fig : figure2 ] - [ fig : figure5 ] and invalidates any approximation based on constant @xmath188-factors or the idea that higher order qcd corrections cancel in @xmath248 . as was already illustrated in fig .
[ fig : figure5b ] , theoretical uncertainties associated with the actual choice of the parameters @xmath184 in the non - perturbative function @xmath158 given in eq .
( [ eq : kart ] ) cancel to a large extent in double - spin asymmetries .
unfortunately , this is not the case for ambiguities related to the choice of scales @xmath28 . as we have discussed in sec .
[ sec : scale ] , the dependence of unpolarized heavy flavor cross sections on variations of @xmath28 is in general more pronounced than in the polarized case , see , e.g. , fig .
[ fig : figure4 ] for electron - muon correlations .
this can cause sizable ambiguities also for ratios of cross sections , like double - spin asymmetries . as a representative example
, we show in fig .
[ fig : figure15 ] the dependence of the double - spin asymmetry @xmath220 for electron - muon correlations on variations of @xmath28 .
the shaded band in the upper panel of fig . [ fig : figure15 ] illustrates the uncertainty on @xmath220 if @xmath179/2)^{1/2}$ ] are varied simultaneously in the range @xmath186 .
the lower panel gives the relative deviation of @xmath220 for @xmath304 with respect to the result obtained for our default value @xmath180 .
the scale uncertainties are quite substantial and not uniform as a function of the invariant mass @xmath218 . nevertheless , the asymmetries obtained with the dssv parton densities are still much smaller than for sets with larger gluon polarizations , like dns(kre ) or grsv(std ) , as can be inferred by comparing with the results given in fig .
[ fig : figure12 ] .
qualitatively similar effects as in fig .
[ fig : figure15 ] are found for the other double - spin asymmetries discussed in this subsection .
we refrain from varying @xmath26 and @xmath27 independently , which increases the uncertainties only slightly for the specific observables and kinematics we are interested in , cf .
[ fig : figure5a ] .
similar observations apply to variations of the heavy quark masses @xmath183 , which , in principle , need to be considered as well , see fig .
[ fig : figure5a ] .
finally , we estimate the range of momentum fractions @xmath7 at which the ( un)polarized parton densities in eq .
( [ eq : xsec - fact ] ) are predominantly probed for the different single - inclusive and lepton - lepton correlation observables discussed in this paper .
figure [ fig : figure16 ] shows the corresponding cross sections differential in @xmath7 . in each case , we have integrated over the angular acceptance for detecting electrons and/or muons with the phenix experiment at rhic as well as over all transverse momenta of the decay leptons . as before , an additional cut @xmath283 is imposed for lepton - lepton correlations .
all @xmath7 distributions are normalized to the respective integrated cross section @xmath305
. typical range of momentum fractions @xmath7 at which the parton densities in eq .
( [ eq : xsec - fact ] ) are probed for the integrated unpolarized ( upper row ) and polarized ( lower row ) cross sections for leptons from heavy flavor decays at rhic shown in figs . [ fig : figure2]-[fig : figure5 ] .
the left panels give the results for the single - inclusive electron ( solid lines ) and muon ( dashed lines ) cross sections .
results for the electron - muon ( dashed lines ) and muon - muon ( solid lines ) correlations are presented in the right panels.,scaledwidth=50.0% ] since small transverse momenta probe also the smallest possible @xmath7 values , fig .
[ fig : figure16 ] gives a rough idea of the lowest possible momentum fractions accessible in heavy quark hadroproduction at rhic with a c.m.s .
energy of @xmath148 .
as can be seen , single - inclusive observables , shown in the panels on the left , receive contributions from @xmath7 values as low as @xmath306 , but the majority of events has @xmath307 . both , electron - muon and muon - muon correlations , displayed on the right - hand - side of fig .
[ fig : figure16 ] , probe on average larger values of @xmath7 . in particular , the latter observable can be used to study parton densities at @xmath308 or higher
the entire suite of possible observables related to heavy flavor hadroproduction at rhic can cover a wide range in @xmath7 and has the potential to provide novel information on the spin structure of the nucleon and the applicability of perturbative qcd in polarized hard scattering processes .
we have presented a flexible parton - level monte carlo program to compute heavy flavor distributions and correlations at nlo accuracy in longitudinally polarized @xmath0 collisions .
experimental acceptance cuts , the hadronization of the produced heavy quark pair into @xmath15 and @xmath56 mesons , and their subsequent semi - leptonic decays can be included in phenomenological applications .
heavy flavor hadroproduction receives its importance for the field of spin physics from its partonic hard scattering processes , which differ from their counterparts for light hadron and jet production utilized in global qcd analyses so far .
upcoming measurements of heavy flavor production at rhic will further our current understanding of factorization in the presence of spin and test the notion of universality for polarized parton densities .
we have performed a comprehensive phenomenological study of various observables where heavy quarks are identified through their semi - leptonic decays into electrons or muons .
such measurements can be done once sufficient statistics has been collected in polarized proton - proton collisions at rhic and do not require the completion of vertex detector upgrades .
decay lepton correlations turned out to be particularly suited probes for the spin structure of the nucleon , and experimentally relevant double - spin asymmetries of about one percent can be expected even for present - day gluon polarizations of rather moderate size .
contrary to naive expectations based on unpolarized results , gluon - gluon fusion is not necessarily the dominant channel for heavy flavor production in longitudinally proton - proton collisions at rhic .
next - to - leading order qcd corrections are in general more pronounced for unpolarized than for polarized heavy flavor production cross sections , such that they do not cancel in the ratio defining double - spin asymmetries .
also , theoretical uncertainties estimated by varying factorization and renormalization scales are usually smaller for spin - dependent observables .
the technical methods and their implementation into a parton - level monte carlo program can be straightforwardly extended to deal also with the spin - dependent photoproduction of heavy quarks .
this will allow one to analyze already existing data for charm production consistently at nlo accuracy within future global qcd analyses of polarized parton densities .
in addition , one can assess the physics impact of heavy flavor distributions and correlations obtained at a possible future polarized lepton - nucleon collider like the eic .
we thank ming xiong liu for useful discussions and ramona vogt for providing us with the electron spectra used in ref .
is supported by a grant of the `` cusanuswerk '' , bonn , germany .
. acknowledges partial support of the initiative and networking fund of the helmholtz association , contract ha-101 ( `` physics at the terascale '' ) . this work was supported in part by the `` bundesministerium fr bildung und forschung '' ( bmbf ) , germany
in the appendix we collect for completeness some additional details of the calculation and some explicit expressions which were omitted in @xcite but might be useful for the reader .
first of all , we recall the lo partonic cross sections for open heavy flavor hadroproduction , which are needed for the factorization of collinear singularities .
they also emerge in the soft gluon limit of the nlo @xmath60 real emission contributions .
the spin - dependent , color - averaged matrix elements squared for the tree - level processes in ( [ eq : born - proc ] ) read in @xmath42 dimensions : @xmath309 \delta b_{qed}\,,\\ % \label{eq : mqqlo } \delta|m_{q\bar{q}}|^2 & = & \frac{1}{2s } ( 4\pi\alpha_s)^2 \frac{c_f}{n_c } \delta a_{qed}\,,\end{aligned}\ ] ] where @xmath310 here , @xmath311 denotes the number of colors , @xmath312 , and @xmath313 .
contrary to the unpolarized case @xcite , @xmath314 receives no @xmath315 contributions .
the mandelstam variables used in eqs .
( [ eq : mgglo ] ) - ( [ eq : aqed ] ) are defined by @xmath316 where @xmath317 .
@xmath318 are the momenta of the incoming partons , and @xmath319 and @xmath320 are the momenta of the produced heavy quark and antiquark , respectively . together with the appropriate two - body phase - space factor @xmath321 in @xmath75 dimensions @xcite , @xmath322^{-1 }
\left(\frac{t_1 u_1-m_q^2 s}{s}\right)^{-\varepsilon}\\ & \times & \delta(s+t_1+u_1)dt_1 du_1\,,\end{aligned}\ ] ] the @xmath97 and @xmath98 born cross sections can be written as @xmath323 where @xmath324^{-1 } \left(\frac{t_1 u_1-m^2 s}{\mu^2 s}\right)^{-\varepsilon}\ ] ] collects all phase - space factors given in eq .
( [ eq : dps2 ] ) and the flux factor @xmath82 included in eqs .
( [ eq : mgglo ] ) and ( [ eq : mqqlo ] ) .
the mass parameter @xmath6 is introduced to keep the strong coupling dimensionless in @xmath75 dimensions . in the limit @xmath116 , @xmath325 reduces to @xmath326 . in the c.m.s .
frame of the incoming partons , @xmath321 and @xmath50 can be conveniently expressed in terms of the scattering angle between @xmath327 and @xmath328 by using the relation @xmath329 where @xmath330 .
the corresponding unpolarized born cross sections in @xmath75 dimensions can be found in ref .
@xcite .
next , we give explicit expressions for the @xmath92 and @xmath93 prescriptions introduced in eq .
( [ eq : xy - expand ] ) to regularize soft and collinear regions of phase - space .
the distributions are defined as follows @xcite : @xmath331\frac{\log(1-x)}{1-x } dx,\\ % \nonumber & \int_{1-w}^1h(y)\left(\frac{1}{1-y}\right)_{w } dy = \int_{1-w}^1\frac{h(y)-h(1)}{1-y } dy,\\ % \nonumber & \int_{-1}^{-1+w}h(y)\left(\frac{1}{1+y}\right)_{w } dy\\ & \quad\quad\quad\quad=\int_{-1}^{-1+w}\frac{h(y)-h(-1)}{1+y } dy.\end{aligned}\ ] ] @xmath332 denotes an arbitrary `` test function '' , which is sufficiently regular in the limits @xmath100 and @xmath108 . in a numerical implementation ,
the distributions ( [ eq : distrib ] ) need to be used only if @xmath7 and @xmath61 are sampled in the ranges @xmath333 and @xmath334 or @xmath335 , respectively , in the monte carlo integration . following ref .
@xcite , the soft , @xmath100 limit of the @xmath60 matrix elements squared is derived by applying the eikonal approximation , i.e. , by taking the limit when the momentum of the emitted gluon gets soft .
the obtained soft matrix elements squared for @xmath98 and @xmath97 scattering have the same form as in the unpolarized case @xcite but with the born results replaced by their polarized counterparts @xmath336 and @xmath314 , given in eq .
( [ eq : aqed ] ) and ( [ eq : bqed ] ) , respectively .
the expressions can be straightforwardly integrated over @xmath7 and @xmath61 in ( [ eq : dps3 ] ) , and we obtain for @xmath337 @xmath338 \bigg\}\delta(s+t_1+u_1)\\ \label{eq : soft - qqba } \frac{d\delta\hat{\sigma}_{q\bar{q},a}^{(s)}}{dt_1du_1}&= \frac{c_a c_f}{n_c } f_\varepsilon g_\varepsilon \alpha_s^3 \tilde{\beta}^{-4\varepsilon } \delta a_{qed } \bigg\{\frac{2}{\varepsilon}\ln\frac{sm^2_q}{t_1u_1 } + \frac{6}{\varepsilon}\ln\frac{t_1}{u_1}-\ln^2\frac{sm^2_q}{t_1u_1 } + 6\ln\varkappa\ln\frac{t_1}{u_1 } + \ln^2\frac{t_1}{u_1}-\ln^2\varkappa \notag\\ & \qquad { } -2\mathrm{li}_2\left(1-\frac{sm^2_q}{t_1u_1}\right)-6\mathrm{li}_2\left(1- \frac{t_1}{\varkappa u_1}\right ) % \notag\\ % & \qquad { } + 6\mathrm{li}_2\left(1-\frac{u_1}{\varkappa t_1}\right ) + \frac{2(2m^2_q - s)}{s\beta}s(\varkappa ) \bigg\}\delta(s+t_1+u_1)\,,\end{aligned}\ ] ] where we have used @xmath339 } ( 1-\frac{3}{2}\zeta(2)\varepsilon^2 ) \left(\frac{m_q^2}{\mu^2}\right)^{-\varepsilon}\,,%\\ % \end{aligned}\ ] ] and @xmath340 .
the dilogarithm function @xmath341 is defined as in ref .
@xcite , and @xmath342 denotes the riemann zeta function .
@xmath343 in ( [ eq : gepsilon ] ) parameterizes the difference of the @xmath60 and @xmath104 phase - space factors , the latter given by @xmath344 in eq .
( [ eq : feps ] ) . as in ref .
@xcite , we have split up the result for @xmath337 into contributions from different color structures .
the results for @xmath345 and @xmath346 in eq .
( [ eq : soft - qqbf ] ) and ( [ eq : soft - qqba ] ) , respectively , agree with the corresponding unpolarized expressions in ref .
@xcite after replacing @xmath336 by @xmath347 . with the help of ( [ eq : t1-cos ] )
, @xmath348 can be easily transformed into @xmath349 used in eq .
( [ eq : xsec - soft ] ) .
@xmath351\right\}\delta(s+t_1+u_1)\,,\\ % \label{eq : soft - gga } \frac{d\delta\hat{\sigma}^{(s)}_{gg , a}}{dt_1du_1 } & = \frac{c_a^2}{n_c^2 - 1 } f_\varepsilon g_\varepsilon \alpha_s^3 \tilde{\beta}^{-4\varepsilon } \delta b_{qed } \bigg\ { -\frac{2t_1u_1}{s^2}\left(\frac{1}{\varepsilon}+1\right ) + \left(1-\frac{2t_1u_1}{s^2}\right)\notag\\ & \quad\times \left[\frac{2}{\varepsilon^2 } -\frac{1}{\varepsilon}\ln\frac{m^2_qs}{t_1u_1 } + \frac{1}{2 } ( \ln^2\frac{m^2_qs}{t_1u_1}+\ln^2\frac{t_1}{u_1}-\ln^2\varkappa ) + \mathrm{li}_2\left(1-\frac{m^2_qs}{t_1u_1}\right)\right]\notag\\ & \quad{}+\frac{t_1 ^ 2-u_1 ^ 2}{s^2}\left [ \ln\frac{t_1}{u_1}\left(-\frac{2}{e}+\ln\varkappa\right ) -\mathrm{li}_2\left(1-\frac{t_1}{u_1\varkappa}\right ) + \mathrm{li}_2\left(1-\frac{u_1}{t_1\varkappa}\right)\right ] \notag\\ & \quad{}-\frac{2m^2_q - s}{s\beta}\left[\frac{2t_1u_1}{s^2}\ln\varkappa - s(\varkappa)\right ] \bigg\}\delta(s+t_1+u_1)\,,\\ % \label{eq : soft - gg1 } \frac{d\delta\hat{\sigma}^{(s)}_{gg,1}}{dt_1du_1 } & = \frac{1}{n_c^2 - 1 } f_\varepsilon g_\varepsilon \alpha_s^3 \tilde{\beta}^{-4\varepsilon } \delta b_{qed } \bigg\ { -\frac{2}{\varepsilon^2 } + \frac{2t_1u_1}{s^2}\left(\frac{1}{\varepsilon}+1\right ) -\ln^2\frac{t_1}{u_1}+\ln^2\varkappa \notag\\ & \quad{}-\frac{2(2m^2_q - s)}{s\beta } \left[-\frac{t_1u_1}{s^2}\ln\varkappa + \left(1+\frac{t_1u_1}{s^2}\right)s(\varkappa)\right ] \bigg\}\delta(s+t_1+u_1)\,.\end{aligned}\ ] ] note that we have chosen a slightly different way of organizing the above results according to their color structure than in ref .
@xcite , but the sum of eqs .
( [ eq : soft - ggqed])-([eq : soft - gg1 ] ) agrees with their expression after replacing @xmath314 by its unpolarized counterpart @xmath352 @xcite .
the expressions for the virtual corrections @xmath96 and the finite contributions @xmath115 in ( [ eq : xsec - finite ] ) are too long to be presented here .
they are available upon request .
turning to the collinear , @xmath108 limit of the @xmath60 processes in ( [ eq : nlo - proc ] ) , we give explicit expressions for the functions @xmath114 appearing in eq .
( [ eq : collinear ] ) .
they read @xmath353 where the born matrix elements squared in eqs .
( [ eq : mgglo ] ) and ( [ eq : mqqlo ] ) are to be evaluated with `` shifted '' kinematics .
this is due to the collinear emission off one of the incoming partons such that only a fraction @xmath7 of their original momenta @xmath318 is available in the subsequent hard scattering .
the @xmath42 dimensional lo polarized splitting functions @xmath354 in eqs .
( [ eq : coll1])-([eq : coll6 ] ) can be found in @xcite and read : @xmath355,\\ % \nonumber \delta p_{qg}(x)&=&\frac{1}{2}\left[2x-1 - 2\varepsilon(1-x)\right],\\ % \nonumber \delta p_{gq}(x)&=&c_f\left[2-x+2\varepsilon(1-x)\right],\\ % \nonumber \delta
p_{gg}(x)&=&2 c_a \big[\frac{1}{(1-x)_{\tilde{\rho } } } - 2x+1 + 2\varepsilon(1-x)\\ & + & \delta(1-x)\left(\frac{\beta_0}{2}+\varepsilon\frac{c_a}{6}+2\log\tilde{\beta}\right)\big]\,,\end{aligned}\ ] ] with @xmath356 , @xmath357 as the number of active flavors , and where we have expressed the standard @xmath358 distributions in @xmath359 and @xmath360 by the corresponding @xmath92-prescriptions defined in ( [ eq : distrib ] ) . this amounts to introducing an additional @xmath361 term in the soft @xmath137 parts of @xmath359 and @xmath360 in ( [ eq : pol - split ] ) .
note that in eqs .
( [ eq : coll1])-([eq : coll6 ] ) , contributions proportional to @xmath137 do not show up as they are already included as @xmath362 poles in the soft cross sections listed in eqs .
( [ eq : soft - qqbf])-([eq : soft - gg1 ] ) . in the factorization counter term @xmath363 in ( [ eq : xsec - counter ] ) ,
only four - dimensional splitting functions are needed , i.e. , @xmath116 in ( [ eq : pol - split ] ) .
99 a. airapetian _ et al . _
[ hermes collaboration ] , phys .
lett . *
84 * , 2584 ( 2000 ) ; p. liebing [ hermes collaboration ] , aip conf .
proc . * 915 * , 331 ( 2007 ) .
b. adeva _ et al .
_ [ spin muon collaboration ( smc ) ] , phys .
d * 70 * , 012002 ( 2004 ) .
e. s. ageev _ et al .
_ [ compass collaboration ] , phys .
b * 633 * , 25 ( 2006 ) .
m. alekseev _ et al . _ [ compass collaboration ] , arxiv:0802.3023 [ hep - ex ] ; phys .
b * 676 * , 31 ( 2009 ) .
s. s. adler _ et al .
_ [ phenix collaboration ] , phys .
lett . * 93 * , 202002 ( 2004 ) ; phys .
d * 73 * , 091102 ( 2006 ) ; a. adare _ et al .
_ [ phenix collaboration ] , phys .
d * 76 * , 051106 ( 2007 ) ; phys .
103 * , 012003 ( 2009 ) ; phys .
d * 79 * , 012003 ( 2009 ) .
b. i. abelev _ et al . _ [ star collaboration ] , phys .
lett . * 97 * , 252001 ( 2006 ) ; phys .
lett . * 100 * , 232003 ( 2008 ) ; m. sarsour [ star collaboration ] , aip conf .
* 1149 * , 389 ( 2009 ) .
see , e.g. , m. stratmann and w. vogelsang , j. phys .
ser . * 69 * , 012035 ( 2007 ) and references therein .
d. de florian , r. sassot , m. stratmann , and w. vogelsang , phys .
* 101 * , 072001 ( 2008 ) ; phys .
d * 80 * , 034030 ( 2009 ) .
g. bunce _
_ , `` plans for the rhic spin physics program '' , june 2008 , http://spin.riken.bnl.gov / rsc/. see , e.g. , a. d. frawley , t. ullrich , and r. vogt , phys .
* 462 * , 125 ( 2008 ) .
a. adare _ et al .
_ [ phenix collaboration ] , phys .
lett . * 97 * , 252002 ( 2006 ) ; phys .
b * 670 * , 313 ( 2009 ) ; phys . rev .
lett . *
103 * , 082002 ( 2009 ) ; s.s .
adler _ et al .
_ [ phenix collaboration ] , phys .
d * 76 * , 092002 ( 2007 ) .
j. adams _ et al . _ [ star collaboration ] , phys . rev .
* 94 * , 062301 ( 2005 ) ; b. i. abelev _ et al . _
[ star collaboration ] , phys .
* 98 * , 192301 ( 2007 ) ; phys .
d * 79 * , 112006 ( 2009 ) . for a review of the current status , see , e.g. , x. wang [ phenix collaboration ] , in _ 18th international symposium on spin physics , charlottesville , va , 2008
_ , eds . d.g .
crabb _ et al .
_ , aip conf .
1149 , p. 393 .
m. cacciari , p. nason and r. vogt , phys .
* 95 * , 122001 ( 2005 ) .
w. beenakker , h. kuijf , w. l. van neerven , and j. smith , phys .
d * 40 * , 54 ( 1989 ) ; w. beenakker , w. l. van neerven , r. meng , g. a. schuler , and j. smith , nucl .
b * 351 * , 507 ( 1991 ) .
p. nason , s. dawson , and r. k. ellis , nucl .
b * 327 * , 49 ( 1989 ) [ erratum - ibid . * b335 * , 260 ( 1990 ) ] . i. bojak and m. stratmann , phys .
d * 67 * , 034010 ( 2003 ) .
d. de florian , r. sassot , and m. stratmann , phys .
d * 75 * , 114010 ( 2007 ) ; phys . rev .
d * 76 * , 074033 ( 2007 ) . for a review , see , e.g. , j. baines _
_ , _ summary report of the heavy quarks working group for the hera - lhc workshop proceedings _ , arxiv : hep - ph/0601164 and references therein . v. g. kartvelishvili , a. k. likhoded , and v. a. petrov , phys .
b * 78 * , 615 ( 1978 ) .
b. mele and p. nason , nucl .
b * 361 * , 626 ( 1991 ) ; m. cacciari and m. greco , nucl .
b * 421 * , 530 ( 1994 ) ; m. cacciari , m. greco , and p. nason , jhep * 9805 * , 007 ( 1998 ) . b. aubert _ et al . _ [ babar collaboration ] , phys . rev .
d * 69 * , 111104 ( 2004 ) ; a. h. mahmood _ et al .
_ [ cleo collaboration ] , phys . rev .
d * 70 * , 032003 ( 2004 ) ; j. wiss , in _ proceedings of 4th flavor physics and cp violation conference ( fpcp 2006 ) , vancouver , canada , 2006 _ , arxiv : hep - ex/0605030 . m. l. mangano , p. nason , and g. ridolfi , nucl . phys .
b * 373 * , 295 ( 1992 ) .
j. riedl , a. schfer , and m. stratmann , eur .
j. c * 52 * , 987 ( 2007 ) .
j. riedl , a. schfer , and m. stratmann , work in progress .
i. bojak and m. stratmann , phys .
b * 433 * , 411 ( 1998 ) ; nucl .
b * 540 * , 345 ( 1999 ) [ erratum - ibid .
b * 569 * , 694(e ) ( 2000 ) ] .
a. p. contogouris , z. merebashvili , and g. grispos , phys .
b * 482 * , 93 ( 2000 ) ; phys .
d * 62 * , 114509 ( 2000 ) [ erratum - ibid .
d * 69 * , 019901 ( 2004 ) ] .
l. phaf and s. weinzierl , jhep * 0104 * , 006 ( 2001 ) ; s. catani , s. dittmaier , m. h. seymour , and z. trocsanyi , nucl .
b * 627 * , 189 ( 2002 ) .
g. t hooft and m. j. g. veltman , nucl .
b * 44 * , 189 ( 1972 ) ; p. breitenlohner and d. maison , commun .
phys . * 52 * , 11 ( 1977 ) .
r. mertig and w. l. van neerven , z. phys .
c * 70 * , 637 ( 1996 ) ; w. vogelsang , phys .
d * 54 * , 2023 ( 1996 ) ; nucl .
b * 475 * , 47 ( 1996 ) .
m. cacciari , s. frixione , m. l. mangano , p. nason , and g. ridolfi , jhep * 0407 * , 033 ( 2004 ) .
r. vogt , private communication . c. amsler
_ et al . _
[ particle data group ] , phys .
b * 667 * , 1 ( 2008 ) .
m. glck , e. reya , m. stratmann , and w. vogelsang , phys .
d * 63 * , 094005 ( 2001 ) .
d. de florian , g.a .
navarro , and r. sassot , phys .
d * 71 * , 094018 ( 2005 ) .
d. de florian , phys . rev . * d67 * , 054004 ( 2003 ) ; b. jger _
et al . _ ,
phys . rev . * d67 * , 054005 ( 2003 ) .
b. jger , m. stratmann , and w. vogelsang , phys . rev . *
d70 * , 034010 ( 2004 ) . j. pumplin , d. r. stump , j. huston , h. l. lai , p. m. nadolsky , and w. k. tung , jhep * 0207 * , 012 ( 2002 ) .
m. czakon , a. mitov , and s. moch , phys .
b * 651 * , 147 ( 2007 ) ; nucl .
b * 798 * , 210 ( 2008 ) ; m. czakon , phys .
b * 664 * , 307 ( 2008 ) ; b. kniehl , z. merebashvili , j. g. korner , and m. rogal , phys .
d * 77 * , 094011 ( 2008 ) ; phys .
d * 78 * , 094013 ( 2008 ) ; c. anastasiou and s. m. aybat , phys . rev .
d * 78 * , 114006 ( 2008 ) . c. hendlmeier , a. schfer , and m. stratmann , eur .
j. c * 55 * , 597 ( 2008 ) .
l. lewin , _ polylogarithms and associated functions _ , north holland , amsterdam , 1983 . | we present a comprehensive phenomenological study of heavy flavor distributions and correlations in longitudinally polarized proton - proton collisions at bnl - rhic .
all results are obtained with a flexible parton - level monte carlo program at next - to - leading order accuracy and include the fragmentation into heavy mesons , their subsequent semi - leptonic decays , and experimental cuts .
next - to - leading order qcd corrections are found to be significant for both cross sections and double - spin asymmetries .
the sensitivity of heavy flavor measurements at bnl - rhic to the gluon polarization of the nucleon is assessed .
electron - muon and muon - muon correlations turn out to be the most promising observables .
theoretical uncertainties are estimated by varying renormalization and factorization scales , heavy quark masses , and fragmentation parameters . |
chains of interacting spins are important models for investigating properties of quantum many - body systems .
the fact that the dimension of the hilbert space grows exponentially with the length of the chains does , however , limit the applicability of brute force numerical computations .
this makes analytical insights particularly valuable , both as a direct source of information and as a starting point for constructing and testing approximate numerical methods .
important analytical techniques include the bethe ansatz @xcite and the construction lying behind the aklt model @xcite .
the former was introduced by bethe in 1931 to solve the antiferromagnetic heisenberg model with nearest - neighbour interactions and has turned out to be a more generally applicable tool to solve particular spin models , and the latter has led to a number of models for which both the hamiltonian and the corresponding ground state(s ) are known .
solvable models with long - range interactions are more rare , but the haldane - shastry ( hs ) model @xcite is an example . in this model ,
@xmath4 spin @xmath5 particles are positioned uniformly on a circle , and each spin interacts with all other spins through a two - body exchange interaction whose strength is inversely proportional to the square of the distance between the spins measured along the chord . in the present paper , we propose a new method to find matching hamiltonians and ground state wave functions , which naturally leads to models with long - range interactions .
the approach , which is presented in detail in section [ sec : hamiltonians ] , relies on the existence of null vectors in conformal field theories ( cft ) , and we shall specifically consider the @xmath0 wzw models , where @xmath1 is a positive integer known as the level of the algebra .
the vacuum expectation value of a product of primary chiral conformal fields is zero if one of the fields is a null vector , and we use this fact and the ward identity to derive equations of the form @xmath6 , where @xmath7 is a hermitian and positive semidefinite operator and @xmath8 is a chiral correlator ( i.e. , the vacuum expectation value of a product of primary chiral fields ) .
explicit analytical expressions for several of the relevant chiral correlators have been found recently in @xcite .
in addition to deriving new spin models , the null vectors enable us to derive a set of simple algebraic linear equations , which relate different @xmath9-point spin correlators within each model ( i.e. , expectation values of the form @xmath10 , where @xmath11 is the @xmath12 component of the spin operator acting on spin number @xmath13 ) .
the approach turns out to be particularly valuable for @xmath2 , where it leads to a closed set of equations for the two - point spin correlators , a closed set of equations for the two - point and the four - point spin correlators and so on . as detailed below , this allows us to obtain a number of results for spin correlators either analytically or through a very simple numerical computation . after explaining the general approach , which is valid for all @xmath1
, we turn to a more detailed investigation of models obtained at levels @xmath2 and @xmath3 in sections [ sec : keq1 ] and [ sec : keq2 ] , respectively . in the @xmath14 wzw model , there are only spin 0 and spin 1/2 chiral conformal fields , and since the fusion properties of the spin 0 field are trivial , we find the explicit expression for a hamiltonian which has the chiral correlator of @xmath4 spin @xmath5 fields as its ground state in section [ sec : keq1hw ] .
the hamiltonian involves only two - body interactions , and in section [ sec : keq1hs ] , we show that it is essentially a nonuniform generalization of the hs model with long - range exchange interactions .
we also compute the spectrum for a number of choices of distributions for @xmath15 and discuss symmetries and the origin of degeneracies .
the hamiltonian of the nonuniform hs model was put forward recently in @xcite , where it was shown analytically that the chiral correlator of @xmath4 spin @xmath5 fields is an eigenstate of it .
this was done by using the knizhnik - zamolodchikov ( kz ) equation @xcite , and we provide the details of that derivation in [ sec : alternative ] .
in contrast to the present work , however , this approach can not be straightforwardly generalized to higher values of @xmath1 , and it does not automatically guarantee that the chiral correlator is , indeed , the ground state .
the nonuniform hs model was compared to other spin models through exact numerical computations for small @xmath4 in @xcite , and the finite chain renyi entropy and the two - point spin correlation function was studied numerically through monte carlo simulations . in the present paper , we instead concentrate on the analytical and semi - analytical expressions for the spin correlators that follow from the properties of the null vectors through the aforementioned linear equations .
we rederive the expression for the two - point spin correlator of the uniform hs model in the thermodynamic limit and furthermore demonstrate that the two - point spin correlator can also be computed analytically for the finite chain uniform hs model .
these results are presented in section [ sec : keq1cf ] , where we also provide further numerical results for two - point and four - point spin correlators .
section [ sec : keq2 ] is devoted to an investigation of the case @xmath3 . in this model , there are spin 0 , spin 1/2 and spin 1 chiral conformal fields , and we shall , in particular , analyze the spin chain model in which the wave function is the chiral correlator of a product of @xmath4 spin 1 fields . in section [ sec : ham1 ] , we derive the hamiltonian and discuss the spectrum . in section [ sec : bilinear ] , we find that the wave function has an overlap with the ground state of the bilinear - biquadratic spin 1 chain close to unity in a region within the haldane phase for @xmath15 . in section [ sec : entcor1 ] , we compute the renyi entropy and the two - point spin correlation function by monte carlo simulations . we find that the renyi entropy grows logarithmically with the length of the considered subchain , while the two - point spin correlation function decays algebraically with the separation between the spins .
the hamiltonian for the chiral correlator of a product of @xmath4 spin 1/2 fields is given in [ sec : k2spinhalf ] together with a discussion of the spectrum . finally , concluding remarks are given in section [ sec : conclusion ] .
the wzw models , together with the minimal models of belavin , polyakov and zamolodchikov , are the best known examples of rational conformal field theories ( rcft ) . in a rcft
the number of primary fields is finite , and they form a closed operator algebra characterized by the fusion rules . in this paper , we shall restrict ourselves to the wzw model based on the kac - moody ( km ) algebra @xmath0 .
there are @xmath16 primary fields in this rcft , which are denoted by @xmath17 , where @xmath18 labels the total spin .
each field consists of @xmath19 components corresponding to the @xmath19 possible values of the projection of the spin on the quantization axis . to refer explicitly to one of these components
, we use the notation @xmath20 , where @xmath21 . the virasoro central charge @xmath22 , conformal weights @xmath23 and fusion rules are given by @xmath24 the simplest nontrivial wzw model corresponds to @xmath2 and has @xmath25 and two primary fields @xmath26 and @xmath27 with @xmath28 and @xmath29 .
this model can be easily constructed using a free massless boson whose central charge is @xmath25 .
the next model in this series corresponds to @xmath3 and has @xmath30 and three primary fields @xmath26 , @xmath27 and @xmath31 with @xmath28 , @xmath32 and @xmath33 .
this model can be constructed by combining three ising models ( @xmath34 ) , or alternatively it can be built using a free boson and an ising model ( @xmath35 ) . in this paper , we shall focus on the @xmath2 and @xmath36 models , but the general formalism will be valid for any integer level @xmath37 .
the _ rationality _ of the wzw models lies in the existence of null vectors in the representation spaces of km algebras .
the decoupling of these vectors provides the basic equations that allow for a complete solution of these models .
null vectors will be the key for the construction of hamiltonians associated to wzw models , and for that reason we shall describe them in detail below .
the @xmath0 wzw model is characterized by the chiral currents @xmath38 , whose laurent expansion @xmath39 contains the modes @xmath40 that satisfy the km algebra @xmath41 = \rmi \sum_c \ ; \vep_{abc } \ ; j^c_{n+m } + \frac{k}{2 } n \ ; \delta_{a b } \ , \delta_{n+m,0 } , \qquad n , m \in \mathbb{z},\ ] ] where @xmath42 is the levi - civita symbol .
notice that the zero modes @xmath43 form a closed @xmath44 algebra . in the standard spin basis
the algebra becomes @xmath45 = \frac{k}{2 } n \
, \delta_{n+m,0 } , \\ \left [ j_n^0 , j_m^{\pm } \right ] = \pm j^{\pm}_{n+m } ,
\\ \left [ j_n^+ , j_m^- \right ] = 2 j^0_{n+m } + k \ , n \ , \delta_{n+m,0},}\end{aligned}\ ] ] where @xmath46 . to construct the representation theory of this km algebra
, we first note that the primary fields introduced above effectively factorize into a product of a chiral field and an anti - chiral field .
we shall only consider the former , which we denote by @xmath47 , where @xmath48 is a complex parameter . in the context of conformal field theory
, @xmath48 represents a point in two - dimensional space - time , but later on , these coordinates will turn into free parameters in the hamiltonians we construct .
to each chiral primary field @xmath47 , one associates a primary state @xmath49 satisfying the following conditions @xmath50 is the vacuum of the theory ( i.e. , @xmath51 ) and @xmath52 are the @xmath19 dimensional matrices of the spin @xmath53 representation of @xmath44 ( the @xmath54th component of @xmath55 is @xmath56 ) .
the representation spaces of the km algebra are modules @xmath57 obtained by acting on the primary states with products of the negative mode operators , i.e. @xmath58 .
these states are called descendants and can be organized in towers having a common value of the sum @xmath59 that is called the virasoro level , not to be confused with the level @xmath1 of the km algebra .
a null vector @xmath60 is a descendant state which is also primary , i.e. , satisfies the condition ( recall ( [ prim ] ) ) @xmath61 each module @xmath57 contains infinitely many null vectors . here
, we shall consider the null vector @xcite @xmath62 which is the null vector occurring at the lowest virasoro level .
it is illustrative to verify ( [ null ] ) for the case @xmath63 . using the commutator @xmath64 = \ell \left ( j^+_{-1 } \right)^{\ell -1 } \left ( k + 1 - \ell - 2 j^0_0 \right ) , \qquad \ell \geq 1,\ ] ] that follows from ( [ km2 ] ) by induction , one obtains , @xmath65 \ ;
| \phi_{jj } \rangle \\ & =( k + 1 - 2 j ) \left ( j^+_{-1 } \right)^{k - 2 j } ( 2 j - 2 j^0_0 ) | \phi_{jj } \rangle = 0,}\ ] ] where we have used that @xmath66 is a primary state , @xmath67 , and that its spin projection is @xmath53 , @xmath68 .
one can easily prove that @xmath69 so that @xmath70 is the highest weight vector of a representation with spin @xmath71 . in summary , the module @xmath57 contains at level @xmath72 a multiplet of @xmath73 null vectors with spin @xmath71 . since a null vector is also a primary field ,
a null vector and its descendants comprise a module on their own , and this module can be shown to decouple from @xmath57 @xcite .
the number of states @xmath74 appearing at virasoro level @xmath9 after removing all the decoupled modules can be determined by considering the character @xmath75 this character is given by @xcite @xmath76 } } { \eta(q)^3 } \sum_{n = - \infty}^\infty [ 2 j + 1 + 2 n ( k+2 ) ] \ ; q^ { n [ 2 j + 1 + n ( k+2)]},\ ] ] where @xmath77 is the dedekin eta function @xmath78 we collect the values of @xmath74 for the lowest spin modules @xmath57 and various values of @xmath9 and @xmath1 in table [ table1 ] .
an inspection of the rows of this table shows that , for given @xmath9 , the number of states remains constant beyond a certain value of @xmath1 .
this can be understood from the fact that states are removed whenever there is a null vector . for @xmath79 ( recall ( [ gw ] ) ) ,
one or more null vectors are present for @xmath80 , whereas no null vectors are present for @xmath80 if @xmath81 .
the number of states removed at @xmath82 and @xmath83 is the number of null vectors in the multiplet @xmath84 , i.e. @xmath73 .
this is expressed in the relation @xmath85 in the module @xmath86 , for example , there are @xmath87 states at @xmath88 and @xmath2 , which turn into @xmath89 states for all @xmath90 .
the difference @xmath91 is the number of null vectors that form the @xmath92 spin multiplet at level @xmath93 and @xmath2 . equation ( [ gw ] ) corresponds to the clebsch - gordan ( cg ) decomposition @xmath94 where @xmath95 is the vector space for the spin @xmath96 irreducible representation ( irrep ) .
this decomposition consists in the symmetrized product of @xmath72 copies of the spin 1 irrep tensored with the spin @xmath53 irrep to yield the highest possible spin @xmath97 .
a basis of @xmath98 is given by ( @xmath99 ) @xmath100 the constants @xmath101 can be constructed from the product of @xmath102 cg coefficients and satisfy the condition @xmath103 which ensures the orthonormality of the vectors @xmath84 .
we adopt the condon - shortley convention that all cg coefficients are real .
we note that ( [ cg2 ] ) could equally well be written in terms of the fields @xmath104 and @xmath105 , and we use this notation in the following .
let us next define the linear map @xmath106 which allows us to write ( [ cg3 ] ) as @xmath107 where @xmath108 is the identity matrix in @xmath98 .
multiplying @xmath109 and @xmath110 in reverse order @xmath111 yields the map @xmath112 that , using ( [ cg5 ] ) and ( [ cg6 ] ) , satisfies @xmath113 so that @xmath114 is a projector with components @xmath115 using this equation , one can parameterize the null vectors ( [ cg2 ] ) with indices of the vector space @xmath116 as follows @xmath117 since , by use of ( [ cg3 ] ) , @xmath118 we see that @xmath119 fulfil the eigenvalue equation @xmath120 as a result of the above construction , @xmath121 must fulfil the following symmetry constraints @xmath122 and , indeed , ( [ cg10c1]-[cg10c3 ] ) together with ( [ cg8 ] ) are sufficient to uniquely determine @xmath121
. the null vectors at levels @xmath2 and @xmath36 are collected in table [ table2 ] , and one can check that the following @xmath123-tensors fulfil the conditions , @xmath124 \delta_{m m'}\\ -\frac{\rmi}{5 } \sum_c \left ( \vep_{a_1 b_1 c } \ , \delta_{a_2 b_2 } + \vep_{a_2 b_2 c } \ , \delta_{a_1 b_1 } + \vep_{a_1 b_2 c } \ , \delta_{a_2 b_1 } + \vep_{a_2 b_1 c } \ , \delta_{a_1 b_2 } \right ) \ , t^c_{m m'},}\\ \fl\left ( { \bf k } _ { 1 , 1 } \right)^{a m } _ { b m ' } = \frac{1}{2 } ( \delta_{a b } \delta_{m m ' } + \delta_{a m ' } \delta_{m b } ) - \frac{1}{3 } \delta_{a m } \delta_{b m ' } , \label{cg16}\end{aligned}\ ] ] where @xmath125 ( @xmath126 ) are the pauli matrices . to simplify the notation , we shall drop the @xmath54 and @xmath127 indices in the following , considering that @xmath123 is a matrix , and we shall let the summation over repeated spin component indices ( e.g. , @xmath128 ) be implicit .
@lllll @xmath1 & @xmath53 & @xmath72 & @xmath71 & @xmath129 + 1 & 0 & 2 & 2 & @xmath130 + 1 & @xmath131 & 1 & @xmath132 & @xmath133 + 2 & 0 & 3 & 3 & @xmath134 + 2 & @xmath131 & 2 & @xmath135 & @xmath136 + 2 & 1 & 1 & 2 & @xmath137 + a chiral correlator is the vacuum expectation value of the product of a set of primary chiral fields : @xmath138 in our construction , @xmath139 is the wave function . written more explicitly , the state of the spin chain is @xmath140 where the coefficients are given by @xmath141 we note that a chiral correlator is not necessarily one unique function .
if the fusion rules ( [ su2 ] ) allow more than one possibility for how two fields can fuse ( i.e. , if there is more than one field on the right hand side of ( [ su2 ] ) ) , there will be more different fusion channels , and each fusion channel corresponds to one wave function .
all of these wave functions are ground states of the hamiltonian that we shall construct .
there are more approaches available to find explicit expressions for the chiral correlators ( [ dec1 ] ) . in some simple cases it is possible to solve the kz equations @xcite , which are a system of partial differential equations in the coordinates @xmath142 fulfilled by the chiral correlators .
the kz equations can be derived from the null vector @xmath143 , where @xmath144 is one of the virasoro generators .
another possibility is to use the feigin - fucks or coulomb gas representation of the wzw model @xcite .
the aim of this section is to derive a set of algebraic equations satisfied by the chiral correlators that follow from the decoupling of null vectors in the modules @xmath57 .
these equations are algebraic , and not differential , as in the kz case , and this fact will allow us to construct quantum spin hamiltonians whose ground states are the chiral correlators of the wzw model . before presenting this construction ,
let us notice that ( [ dec1 ] ) is invariant under global @xmath44 rotations generated by the total spin operator @xmath145 @xcite @xmath146 where @xmath147 denotes the spin matrix acting at the site @xmath148 , in the representation @xmath149 corresponding to the field @xmath150 .
we shall suppose for simplicity that @xmath151 , @xmath152 , but the results can be easily generalized when the primary fields have different spins .
let us start by replacing the primary field @xmath153 in the correlator ( [ dec1 ] ) by the null field ( [ cg10 ] ) in the module @xmath154 .
one can show from the definition of a null field that a chiral correlator is zero if it contains a null field @xcite .
this is referred to as the decoupling of the null field . therefore , @xmath155 i.e. , @xmath156 for all @xmath157 , where the superindex @xmath148 in @xmath158 implies that this matrix acts on the spin degrees of the operator @xmath17 representing the @xmath148th spin in the chain .
the modes @xmath159 act on the field located at the point @xmath142 , but their action can be transferred to the primary fields located at the other positions by means of the ward identity @xcite @xmath160 here , and in all that follows , we use the convention that the sum over @xmath161 means that the indices outside ( inside ) the brackets should ( should not ) be summed over , and only terms for which all @xmath162 indices are different contribute to the sum .
similarly , we write @xmath163 when all @xmath9 indices are required to be different . applying ( [ dec5 ] ) @xmath72 times to ( [ dec4 ] ) yields
@xmath164 which can be written in the compact form @xmath165 where @xmath166 and @xmath167 .
these are the algebraic equations mentioned at the beginning of this subsection .
finally , we shall transform ( [ dec7 ] ) into the following equation @xmath168 where @xmath169 and @xmath170 . to show this relation ,
let us apply the identity @xmath171 to @xmath172 where we have used ( [ cg10c3 ] ) in the final step .
next , we want to move the @xmath173 matrices to the right in this expression .
this produces extra terms coming from the commutator @xmath174 = \rmi \vep _ { b b_k b'_k } \sum _ { k ( \neq i ) } w_{i k } \ , t^{b'_k}_k,\ ] ] but the antisymmetry in the @xmath175 indices causes these terms to cancel out when multiplied by the tensor @xmath123 whose indices are totally symmetric .
once the @xmath176 matrices are brought to the right , they annihilate the correlator @xmath8 , which proves that @xmath177 and thus ( [ dec9 ] ) is equivalent to ( [ dec7 ] ) .
we can use the operators in ( [ dec10 ] ) to define the matrices @xmath178 which are hermitian , positive semidefinite and rotationally invariant @xmath179 = 0.\ ] ] they annihilate the correlator @xmath8 , @xmath180 and their sum can be taken as the hamiltonian of a spin chain with spin @xmath53 at each site , i.e. , @xmath181 it follows immediately that @xmath8 is the ground state of this hamiltonian .
if the fusion rules give rise to several chiral correlators , then @xmath182 will have several ground states in one - to - one correspondence with the chiral correlators .
finally , we note that hamiltonians that are translationally invariant by one site can be achieved by choosing the coordinates @xmath183 to be equally spaced on the unit circle , i.e. , @xmath184 we shall refer to this choice as the uniform case .
the decoupling equations ( [ dec9 ] ) also imply certain relations between the different @xmath9-point spin correlators .
let us write @xmath185 as the sum of an anti - hermitian and a hermitian operator , @xmath186 and define @xmath187 the decoupling equations ( [ dec9 ] ) imply @xmath188 and adding and subtracting these equations , we find @xmath189\right ) |\psi\rangle = 0 , \label{ss20a } \\
\left(\left [ t_{i_1\ldots i_n}^{b_1\ldots b_n } , a_{n _ * , j}^{i , a_1 \dots a_{n _ * } } \right ] + \left\ { t_{i_1\ldots i_n}^{b_1\ldots b_n } , q_{n _ * , j}^{i , a_1 \dots a_{n _ * } }
\right\}\right ) |\psi\rangle = 0 . \label{ss20b}\end{aligned}\ ] ] inserting the actual expressions for @xmath190 and @xmath191 and choosing different values of @xmath9 , ( [ ss20a ] ) and ( [ ss20b ] ) provide a set of linear equations between different spin correlators .
we now turn to a more detailed investigation of the case @xmath2 .
first , we provide explicit expressions for the wave function and the hamiltonian for @xmath4 spin 1/2 fields .
we then show that the model is a generalization of the hs model and discuss the degeneracies of the spectrum .
as mentioned in the introduction , the overlap with other spin models , the renyi entropy and the two - point spin correlator have already been computed numerically in @xcite , and consequently we here concentrate on deriving analytical and semi - analytical expressions for the spin correlators , which is done in the last part of this section . at level @xmath2
there are two primary fields @xmath26 and @xmath27 whose fusion rules are @xmath192 these rules tell us the number of chiral correlators that arise from a product of primary fields on a given geometry , which in our case is the sphere . for @xmath2 ,
the nontrivial correlators are obtained by multiplying @xmath4 fields @xmath27 .
equation ( [ fuse1 ] ) implies that the fusion of these fields produces the field @xmath26 if @xmath4 is even and @xmath27 if @xmath4 is odd .
since the vacuum expectation value of @xmath27 is zero , and that of @xmath26 is not , @xmath4 must be even to get a nonvanishing correlator @xmath193 here , we have made a slight change of notation compared to the previous section , such that @xmath194 is the third component of the @xmath195 spin .
in addition , @xmath196 can only be nonzero for @xmath197 because @xmath8 is a singlet .
note that the correlator ( [ psi1 ] ) is unique since the fusion decomposition only generates one field at a time
. the explicit form of ( [ psi1 ] ) can easily be obtained from the free field representation of the @xmath14 wzw model in terms of a free boson @xmath198 whose two - point correlator is @xmath199 . in the cartan - weyl representation ,
the current algebra operators @xmath200 are given by @xmath201 and the primary fields @xmath202 ( @xmath203 ) by @xmath204 where @xmath205 denotes normal ordering .
the two - point correlators of these fields are @xmath206 where @xmath207 as above . plugging ( [ phi1 ] ) into ( [ psi1 ] ) , one gets @xcite @xmath208
the marshall sign factor @xmath209 ensures that @xmath8 is a singlet state , as required by ( [ dec2 ] ) .
the hamiltonian @xmath210 , whose ground state is given by ( [ psi2 ] ) , can be constructed using the results of the previous section . for @xmath211 , @xmath152 , we find @xmath212 . \label{htot22}\end{aligned}\ ] ] in 1988 , haldane and shastry considered the state obtained by the gutzwiller projection of the one dimensional fermi state at half filling @xcite ( see @xcite for a review ) @xmath213 where @xmath214 are the occupation operators of an electron in a periodic chain with @xmath4 sites , and @xmath215 is the fermi state at half filling .
the state ( [ ss0 ] ) has a unique spin per site because the charge degrees of freedom are frozen by the gutzwiller projection .
these authors then wrote ( [ ss0 ] ) in terms of hard - core boson variables .
mapping the spin up ( down ) into the empty ( occupied ) states of a hard - core boson , they found @xmath216 where @xmath217 is the hard - core boson creation operator at the @xmath218 site and @xmath219^{2}. \label{wave3}\end{aligned}\ ] ] using this equation , haldane and shastry showed that @xmath220 is the ground state of the spin hamiltonian @xmath221 where @xmath222 is the permutation operator of the spins at the positions @xmath148 and @xmath53 .
the permutation can be expressed in terms of spin operators as @xmath223 , so ( [ hs1 ] ) takes the form @xmath224 in @xcite , it was shown that the wave function ( [ wave3 ] ) corresponds to the state ( [ psi2 ] ) for the choice @xmath225 .
indeed , let us denote by @xmath226 the occupation number of a hard - core boson at the site @xmath148 .
the mapping between the spin variable @xmath227 and @xmath228 is given by @xmath229 and one can easily show that ( [ wave3 ] ) becomes @xmath230^ { s_n s_m/2 } , \qquad \sum_{i=1}^n s_i = 0 , \label{wave2}\end{aligned}\ ] ] where the constraint @xmath231 guarantees that the total third component of the spin vanishes .
this expression is proportional to ( [ psi2 ] ) . from these considerations one must expect the hs hamiltonian ( [ hs1 ] ) to be intimately related to the hamiltonian ( [ htot22 ] )
. to show this relation in detail , we need some simple identities . for generic values of @xmath142
, one can show that @xmath232 and @xmath233 the latter identity implies @xmath234 which plugged into ( [ htot22 ] ) yields @xmath235 t^a_i \ , t^a_j , \label{hbis}\end{aligned}\ ] ] where @xmath145 is the total spin operator defined in ( [ dec2 ] ) . in the uniform case , we have @xmath236 so @xmath237 where @xmath238 is the ground state energy of @xmath239 .
the positivity of @xmath240 implies a lower bound for the energies of the eigenstates of @xmath239 with total spin @xmath241 , i.e. , @xmath242 which is indeed satisfied by the lowest eigenenergy @xmath243 in the total spin sector @xmath244 , @xmath245 the latter formula can be derived from the general expression for the eigenvalues of ( [ hs1 ] ) .
the eigenvalues were found in @xcite and can be expressed in terms of a set @xmath246 of nonconsecutive integers @xmath247 , @xmath248 , called rapidities @xcite @xmath249 thus , for example , the ground state corresponds to the set @xmath250 ( @xmath4 even ) , and the lowest excited state of total spin 1 to @xmath251 . in the nonuniform case , where @xmath252 for some @xmath148 , we get a generalization @xmath253 of the hs model .
specifically , @xmath254 where @xmath255 \ , ( \mathcal{p}_{ij } -1 ) \label{hs3}\ ] ] and the ground state energy is @xmath256 - \frac{n}{4}.\ ] ] note that the coupling strength in ( [ hs3 ] ) is more complicated than just the inverse of the square of the distance between the spins measured along the chord . in @xcite , a different approach based on
the kz equation also led to the hamiltonian @xmath257 , and it was demonstrated that @xmath8 is an eigenstate of @xmath257 with eigenvalue @xmath258 .
we provide the details of this derivation in [ sec : alternative ] .
the spectrum of the uniform hs model exhibits an enhanced symmetry given by a yangian hopf algebra @xcite generated by the total spin operator @xmath145 and the operator @xcite @xmath259 the operators @xmath145 and @xmath260 all commute with the hamiltonian ( [ hs1 ] ) but they do not commute among themselves , and they are the key to compute the complete spectrum mentioned above . observe that @xmath261 can be constructed from the operator ( [ ci ] ) , @xmath262 it follows that @xmath8 is an eigenstate of ( [ newy ] ) ( with eigenvalue zero ) in both the uniform and the nonuniform cases . in the nonuniform case , however , the yangian symmetry is broken , meaning that ( [ newy ] ) does not commute with ( [ hs3 ] ) . in the last part of [ sec : alternative ] , we show that @xmath8 is also an eigenstate ( with eigenvalue zero ) of the operator @xmath263 which is a generalization of the inozemtsev invariant @xcite to the nonuniform case . in the uniform case ,
this operator commutes with the hamiltonian ( [ hs1 ] ) @xcite , but @xmath264 does not commute with @xmath265 in general . to illustrate the breaking of symmetries , we compute the spectrum for the three different choices of @xmath142 considered in @xcite : uniform , dimer and random .
the dimer model is defined by the following choice of parameters @xmath266/n } , \qquad 0 \leq \delta \leq \frac{1}{2 } , \qquad j = 1 , 2 \dots , n , \label{dim}\ ] ] where @xmath267 characterizes the dimerization of the ground state .
the value of @xmath268 defined in ( [ wid2 ] ) is @xmath269 which only vanishes in the uniform case @xmath270 .
the random model is defined by the choice @xmath271 where @xmath17 is a random variable uniformly distributed in the interval @xmath272 $ ] .
the parameter @xmath267 determines the degree of disorder in the ensemble .
the choice ( [ ran ] ) gives the random hs version of the spin 1/2 antiferromagnetic heisenberg model ( afh ) with random exchange couplings .
the latter model belongs to the universality class known as the random singlet phase .
the renyi entropies of this phase differ from that of the uniform afh model given by the @xmath14 model @xcite .
@llll @xmath273 & @xmath241 & @xmath274 & # + -19 & 0 & 3 & 1 + -16 & 0 & 0 & 1 + & 1 & 0 & 3 + -14 & 1 & -2 & 3 + & 1 & 2 & 3 + -13 & 0 & -1 & 1 + & 0 & 1 & 1 + & 1 & -1 & 3 + & 1 & 1 & 3 + -10 & 1 & 0 & 3 + -9 & 0 & 3 & 1 + & 1 & 3 & 3 + & 2 & 3 & 5 + -8 & 1 & -2 & 3 + & 1 & 2 & 3 + & 2 & -2 & 5 + & 2 & 2 & 5 + -5 & 2 & -1 & 5 + & 2 & 1 & 5 + 0 & 3 & 0 & 7 + @llll @xmath273 & @xmath241 & @xmath275 & # + -19.9502 & 0 & 0 & 1 + -16.6431 & 1 & 0 & 3 + -16.3598 & 0 & 0 & 1
+ -14.6851 & 1 & -2 & 3 + & 1 & 2 & 3 + -13.3167 & 0 & -2 & 1 + & 0 & 2 & 1 + -13.2717 & 1 & -2 & 3 + & 1 & 2 & 3 + -10.3401 & 1 & 0 & 3 + -9.3167 & 2 & 0 & 5 + -8.9669 & 1 & 0 & 3 + -8.9570 & 0 & 0 & 1 + -8.3167 & 2 & -2 & 5 + & 2 & 2 & 5 + -7.9934 & 1 & -2 & 3 + & 1 & 2 & 3 + -5.0000 & 2 & -2 & 5 + & 2 & 2 & 5 + -0.0000 & 3 & 0 & 7 + @lll @xmath273 & @xmath241 & # + -22.9229 & 0 & 1 + -19.8148 & 1 & 3 + -18.1316 & 0 & 1 + -16.8832 & 1 & 3 + -15.8047 & 1 & 3 + -15.2108 & 0 & 1 + -14.9450 & 1 & 3 + -14.0555 & 1 & 3 + -12.7601 & 0 & 1 + -11.4068 & 2 & 5 + -10.4704 & 1 & 3 + -10.0944 & 2 & 5 + -9.0416 & 1 & 3 + -8.8204 & 0 & 1 + -8.0155 & 1 & 3 + -7.7381 & 1 & 3 + -7.5353 & 2 & 5 + -5.0000 & 2 & 5 + -4.8865 & 2 & 5 + -0.0000 & 3 & 7 + the spectrum of the generalized hs hamiltonian ( [ hs3 ] ) for these three models is given in table [ compare ] for @xmath15 .
the hamiltonian @xmath253 , the dot product @xmath276 of the total spin operator with itself and the @xmath48-component @xmath277 of the total spin operator all commute with each other .
it is hence possible to diagonalize these operators simultaneously and label the resulting states by their energy @xmath273 , spin @xmath241 and z - component of the total spin @xmath278 .
we have omitted @xmath278 from the table , because @xmath279 also commute with the hamiltonian , and it follows that the @xmath280 states within each spin multiplet are degenerate .
instead we write only the number of states in each multiplet . in the uniform case
, the momentum operator also commutes with @xmath145 and the hamiltonian due to invariance under translation by one site .
the momentum operator can be defined as @xmath281 , where @xmath282 is the translation operator and @xmath283 is the operator that permutes spins @xmath148 and @xmath53 as defined just after ( [ hs1 ] ) .
the momentum eigenvalues in units of @xmath284 are given in the table .
the fact that the chain is invariant under the flip operation @xmath285 explains why states with momentum eigenvalue @xmath286 or @xmath287 always appear in degenerate pairs . in the dimer case , @xmath288 does not commute with the hamiltonian , but since the chain is invariant under translation by two sites , we can instead add the momentum operator @xmath289 to the set of mutually commuting operators . also in this case , states with @xmath290 eigenvalue @xmath287 always appear in degenerate pairs due to invariance under the flip operation ( [ flip ] ) . for the random case , there is no translational symmetry .
looking at the degeneracies between different spin multiplets , we observe that the number of degeneracies decreases in going from the uniform to the dimer case and in going from the dimer to the random case . in the dimer case ,
the yangian symmetry is broken and only degeneracies between pairs with opposite @xmath290 eigenvalues occur . in the random case , there are even less symmetries , and all spin multiplets have different energies .
this subsection shows that the generalization of the uniform hs model embraces several interesting models within a common framework .
we shall show below that more information about this model can be obtained exploiting the algebraic decoupling equations .
in 1987 , gebhard and vollhardt computed the two - point spin correlator in the gutzwiller state ( [ ss0 ] ) for an infinite chain , obtaining the exact formula @xcite @xmath291 where @xmath292 is the sine integral function @xmath293 the appearance of the term @xmath294 reflects the @xmath44 symmetry of the gutzwiller state .
it may be worthwhile to notice that this result triggered the haldane - shastry works @xcite . using the asymptotic expression of the sine integral function @xmath295 one obtains @xmath296 , \qquad n \gg 1 .
\label{si3}\ ] ] one can also derive ( [ si3 ] ) from the underlying cft that describes the long distance behaviour of the hs model that is given by the @xmath14 wzw model . the spin operators @xmath297 can be expressed in terms of the current operators @xmath200 ( [ jj ] ) and their antiholomorphic version @xmath298 as @xcite @xmath299 where @xmath300 and @xmath301 .
it is interesting to compare the spin correlator of the hs model with that of the afh model .
the latter correlator behaves as @xcite @xmath302 \label{corafh}\ ] ] where @xmath303 is a nonuniversal constant .
the logarithmic correction to the leading @xmath304 term is due to the marginal irrelevant operator @xmath305 , which added to the action of the wzw model gives the long distance behaviour of the afh model @xcite . in this respect ,
the hs model is at the fixed point of the renormalization group , so that the continuum hamiltonian does not contain the marginal operator @xmath306 , which explains why the logarithmic corrections in the correlator ( [ si3 ] ) are absent
. we shall next derive ( [ ss1 ] ) and several other expressions for spin correlators from ( [ ss20a ] ) . from ( [ ci ] ) , we find @xmath307 where it is assumed that @xmath211 , such that @xmath308 is imaginary .
note that @xmath309 and @xmath310 are one - body and two - body operators , respectively .
for the spin @xmath5 case , a product of spin operators acting on the same spin can always be written as a sum of terms containing at most one spin operator each , so it is sufficient to consider spin correlators for which all the spin operators act on different spins .
we hence assume that all the indices @xmath311 in @xmath312 are different and also different from the index @xmath148 .
@xmath312 is then an @xmath9-body operator .
if we choose @xmath9 odd in ( [ ss20a ] ) , we get @xmath313 linear equations involving only ( @xmath314)-point and ( @xmath315)-point spin correlators . starting from @xmath316 ,
this allows us to successively compute spin correlators with higher and higher numbers of operators as we shall see below .
finally , we mention the important result @xmath317 which follows from ( [ ss20a ] ) with @xmath316 and @xmath318 .
( [ ss4 ] ) is a system of linear equations for the two - point correlators @xmath319 with @xmath320 that can be solved numerically for generic values of the parameters @xmath321 , and analytically in special cases . in the following paragraphs
, we shall investigate some particular solutions .
let us choose @xmath322 then , in the limit @xmath323 ( [ ss4 ] ) becomes @xmath324 where the equality is reached in the limit @xmath325 .
in the uniform case , i.e. @xmath326 , translational invariance implies that @xmath327 only depends on the difference @xmath328 , hence it is sufficient to consider the case @xmath329 and @xmath330 in ( [ ss5 ] ) , @xmath331 where we have used that @xmath332 . to solve these equations
, we introduce the fourier transform of @xmath333 , @xmath334 using the identities @xmath335 ( [ ss6 ] ) turns into the differential equation @xmath336 the @xmath44 invariance of the ground state implies that @xmath337 , hence integrating ( [ ss9 ] ) yields @xmath338 finally , @xmath333 is given by the inverse fourier transform , @xmath339 since @xmath340 due to the @xmath44 symmetry , ( [ ss1 ] ) follows .
let us next show that ( [ ss4 ] ) can be solved analytically for a uniform spin chain with a finite number of sites @xmath4 , and not only in the large @xmath4 limit , as done above .
we write @xmath341 so that ( [ ss4 ] ) becomes @xmath342 if @xmath343 , the first term of this equation vanishes , which leads us to split this equation as @xmath344 to solve these equations , one introduces the finite fourier transform ( recall ( [ ss7 ] ) ) @xmath345 where @xmath346 .
now we multiply ( [ ss11a ] ) by @xmath347 and sum over @xmath348 to obtain @xmath349 while ( [ ss11b ] ) becomes @xmath350 the two sums appearing in ( [ ss13 ] ) are the discrete version of the integrals ( [ ss8 ] ) , and they are given by @xmath351 in terms of which ( [ ss13 ] ) and ( [ ss14 ] ) become @xmath352 where @xmath353 is a constant given by @xmath354 it is convenient to define the quantity @xmath355 which must satisfy @xmath356 taking the difference for two consecutive values @xmath357 and @xmath358 in ( [ ss17a ] ) yields @xmath359 which , when translated into the @xmath360 quantities and using ( [ ss15a ] ) for @xmath361 , becomes @xmath362 and hence @xmath363 one can check that ( [ ss17 ] ) is also satisfied and that the parameter @xmath353 is equal to @xmath364 . plugging this expression into ( [ ss12 ] ) , one finally finds the correlator @xmath365 } { m - \frac{1}{2 } } .
\label{ss19}\ ] ] in the limit @xmath366 , one recovers ( [ ss1 ] ) .
moreover , one can take the limit @xmath367 and @xmath368 with the ratio @xmath369 kept constant , obtaining @xmath370 . \label{ss20}\ ] ] for the uniform hs wave function for a chain with @xmath371 sites .
the results correspond to ( i ) the exact formula ( [ ss19 ] ) , ( ii ) the antiferromagnetic heisenberg model with nearest neighbour couplings , and ( iii ) the thermodynamic limit ( [ ss1]).,scaledwidth=60.0% ] equation ( [ ss20 ] ) can be obtained from the bosonization formulas ( [ si4 ] ) with the operators @xmath200 , @xmath298 and @xmath372 being defined on a cylinder of circumference @xmath373 , i.e. , @xmath374 .
the two - point correlators needed to reproduce ( [ ss20 ] ) can be deduced using the conformal transformation from this cylinder to the complex plane , @xmath375 , which yields @xmath376^{-1},\\ \fl\left\langle \cos\left(\frac { \varphi ( 2 \pi \rmi n_1 , - 2 \pi \rmi n_1)}{\sqrt{2}}\right ) \cos\left(\frac { \varphi ( 2 \pi \rmi n_2 , - 2 \pi \rmi n_2)}{\sqrt{2}}\right)\right\rangle_{\rm cyl } = \left [ 4
n \left|\sin \left ( \frac { \pi ( n_1 - n_2)}{n } \right)\right| \right]^{-1}.\end{aligned}\ ] ] as a summary of these results , we present a comparison between the correlators computed with the formulas ( [ ss1 ] ) and ( [ ss19 ] ) for the hs model and the correlation function for the afh model in figure [ comparison ] . in the dimer case ( [ dim ] ) , translational invariance by one site is broken , but there is still translational invariance for translation by an even number of sites . from the symmetries of the wave function , it follows that @xmath377 for @xmath54 odd and @xmath378 for @xmath54 even , so it is sufficient to compute @xmath379 for @xmath380 .
figure [ dimercor ] provides numerical results for this correlator for a chain with @xmath381 sites obtained from ( [ ss4 ] ) . for @xmath270 , we observe that the absolute value of the correlator decays as @xmath382 in the region @xmath383 as it should according to ( [ si3 ] ) . for @xmath384 ,
we find that the absolute value of the correlator decays as @xmath385^{-2}$ ] , when the spins are sufficiently far apart . as @xmath267 decreases towards zero
, @xmath369 needs to be closer and closer to one half for this approximation to hold .
it is expected that the decay is algebraic for @xmath383 rather than exponential even if the model is not critical because there are long - range interactions in the hamiltonian @xcite . ) for the dimer model with ( starting from above ) @xmath270 , @xmath386 , @xmath387 , @xmath388 , @xmath389 and @xmath381 ( only the region @xmath390 is shown ) .
the dotted lines are all proportional to @xmath385^{-2}$],scaledwidth=60.0% ] we shall now use ( [ ss20a ] ) to compute the four - point spin correlator @xmath391 .
as explained above , we can assume the indices @xmath392 to be different and consider the quantity @xmath393 choosing @xmath394 and @xmath395 ( @xmath396 all different ) in ( [ ss20a ] ) yields @xmath397 where @xmath398 is the two - point spin correlator found above . the expression can be simplified noting that , by rotational invariance of the ground state , one has the decomposition @xmath399 where @xmath400 are invariant tensors that can be obtained from @xmath401 as @xmath402 permuting the spin matrices in ( [ ss21 ] ) , one can easily relate these tensors , @xmath403 plugging ( [ ss22 ] ) into ( [ ss23 ] ) , one finds an expression , where the terms proportional to @xmath404 , etc , must cancel independently .
this leads to @xmath405 these equations can be solved numerically for the four - point correlators in terms of the two - point correlators which are known from the solutions of ( [ ss4 ] ) .
we can also exploit the relations ( [ ss22b ] ) to rewrite the above system in the compact form @xmath406 in the uniform model computed from the exact formula ( [ ss24 ] ) and the approximation ( [ ss26 ] ) , which is valid when all the spins included in the four - point correlator are separated by many sites.,scaledwidth=60.0% ] for the uniform model , we can compare these formulas to the cft prediction for the correlator of @xmath407 spins .
we shall only consider the most relevant term in the expression ( [ si4 ] ) for @xmath408 , namely @xmath409 where @xmath410 .
employing the aforementioned map between the cylinder and the plane one can derive @xmath411 \right|^ { s_i s_j}. \nonumber\end{aligned}\ ] ] a comparison of this formula for @xmath412 and the results obtained from the numerical solution of ( [ ss24 ] ) in the uniform case is given in figure [ fpcfig ] .
the @xmath413 wzw model has three primary fields @xmath414 associated to the spins @xmath415 , @xmath5 and @xmath416 .
the nontrivial parts of the fusion rules are given by @xmath417 this theory is equivalent to the product of the critical ising model and a free massless boson , so that their virasoro central charges add up to @xmath418 .
the current operators can be written in terms of the majorana field @xmath419 of the ising model and vertex operators of the chiral boson as @xmath420 in the following , we investigate the model obtained from a product of @xmath4 spin 1 fields , and we note that the primary fields with spin @xmath421 and conformal weight @xmath422 can be constructed as @xmath423 for brevity , we restrict the numerical analyses to the uniform case in this section .
the hamiltonian for the model obtained from a product of @xmath4 spin @xmath5 fields ( @xmath3 , @xmath424 ) is given in [ sec : k2spinhalf ] .
the above representation of the primary fields enables us to compute the chiral correlator of @xmath4 spin 1 fields with spin projections @xmath425 and parameters @xmath142 @xcite : @xmath426 the pfaffian @xmath427 comes from the correlator of the @xmath419 fields and is hence restricted to the positions where @xmath428 .
the sign factor @xmath429 ensures that @xmath8 is a singlet .
the wave function vanishes if @xmath4 is odd because an odd number of @xmath31 fields fuse to a single @xmath31 field .
we note that a different wave function containing some of the same elements was investigated in @xcite , where a three - body hamiltonian , whose ground state is close to the wave function , was found numerically .
to find the explicit expression for the hamiltonian , we rewrite ( [ cg16 ] ) into @xmath430 and consequently @xmath431.\ ] ] when we plug this into ( [ dec11 ] ) , assume @xmath211 , @xmath152 , and exploit the relations @xmath432=\rmi\vep_{abc}t_i^c , \qquad t_i^at_i^a=2 , \qquad \vep_{abc}t_i^bt_i^c=\rmi t_i^a,\ ] ] we find @xmath433 the hamiltonian for the spin 1 case is thus @xmath434 the spectrum of ( [ hk2j1 ] ) for the uniform case and @xmath15 is given in table [ spectrumspin1 ] . since @xmath435 commutes with @xmath145 , we again have that the @xmath280 states within each spin multiplet are degenerate .
furthermore , we can also classify the states by the eigenvalue of the momentum operator @xmath436 , where @xmath437 is the translation operator defined in ( [ trans ] ) with the permutation operator given in the spin 1 case by @xcite @xmath438 we observe that all degeneracies between multiplets can be explained from the symmetry under translation and under a flip of the spin chain .
this suggests that there is no yangian symmetry in this model .
we have tried to restore the yangian symmetry by creating accidental degeneracies in the spectrum by giving the terms in the hamiltonian ( [ dec14 ] ) different weights , but we did not find a solution .
@llll @xmath273 & @xmath241 & @xmath274 & # + 0.0000 & 0 & 0 & 1 + 1.3322 & 1 & 3 & 3 + 3.8562 & 2 & 0 & 5 + 9.7852 & 3 & 3 & 7 + 13.3457 & 1 & -1 & 3 + & 1 & 1 & 3 + 14.5984 & 2 & -2 & 5 + & 2 & 2 & 5 + @llll + 15.4088 & 0 & 3 & 1 + 15.5224 & 1 & -2 & 3 + & 1 & 2 & 3 + 17.1898 & 4 & 0 & 9 + 17.3510 & 2 & -1 & 5 + & 2 & 1 & 5 + 19.3269 & 3 & -2 & 7 + & 3 & 2 & 7 + @llll + 20.5618 & 1 & 0 & 3 + 23.3377 & 2 & 0 & 5 + 23.9121 & 2 & -2 & 5 + & 2 & 2 & 5 + 25.3391 & 1 & -2 & 3 + & 1 & 2 & 3 + 25.5448 & 2 & -1 & 5 + & 2 & 1 & 5 + in @xcite , it was suggested that the dimer version of the generalized hs model for suitably chosen @xmath267 could be close to the spontaneously dimerized phase of the @xmath439-@xmath440 heisenberg model because the hamiltonians of both models involve terms of the form @xmath441 and @xmath442 and because tuning @xmath267 can change the pattern of coupling strengths .
it turned out that there indeed is a significant overlap between the wave functions of the two models , and one may ask whether the @xmath3 and @xmath421 model resembles another model . in the uniform case , a natural choice for comparison is the bilinear - biquadratic spin 1 hamiltonian @xcite @xmath443\ ] ] with periodic boundary conditions @xmath444 .
we compare the state @xmath8 in ( [ psik2j1 ] ) to the ground state @xmath445 of ( [ hblbq ] ) for @xmath15 in figure [ bilinbiquad ] .
the phases of the bilinear - biquadratic spin 1 model for @xmath446 are also indicated in the figure . in the haldane phase
, there is a unique , singlet ground state with an energy gap to the first exited state . for @xmath15 ,
this is also what we have here , and hence one could expect that @xmath8 may be similar to @xmath445 in this parameter regime .
in fact , the overlap between the two states is seen to be very close to unity in an interval within the haldane phase .
the maximum is @xmath447 and occurs for @xmath448 .
a nonzero overlap is also seen in the dimer and trimer phases , where the condition @xmath449 is fulfilled . in the ferromagnetic phase ,
on the other hand , the overlap vanishes because the ground states of ( [ hblbq ] ) all have total spin @xmath4 .
the results presented below for the renyi entropy and the two - point spin correlator suggest that the spin 1 model at level @xmath3 is gapless for @xmath446 , and hence we do not expect the two models to be close to each other in that limit . between the ground state @xmath8 of ( [ hk2j1 ] ) and the ground state @xmath445 of ( [ hblbq ] ) as a function of @xmath450 for @xmath15 sites ( when the ground state of @xmath451 is degenerate , we choose the state in the ground state subspace , which maximizes @xmath452 ) .
the dotted vertical lines indicate the transitions between different phases of the bilinear - biquadratic spin 1 chain in the thermodynamic limit.,scaledwidth=60.0% ] it is interesting to ask , whether there is a simple connection between the properties of the conformal fields used to construct the wave function and the properties of the wave function .
this turns out to be the case for the uniform hs model , where it is known that the low energy states are described by the @xmath14 wzw model in the large @xmath4 limit @xcite . to investigate whether a similar result could be true for the spin 1 case
, we compute below the renyi entropy and the two - point correlator of the spin 1 chain numerically and compare the results to those expected for the @xmath413 wzw model .
we start with the renyi entropy @xmath453,\ ] ] where @xmath454 is the reduced density operator of the first @xmath455 spins obtained by tracing out the state of the last @xmath456 spins . rewriting this equation
, we can express the entropy in terms of an expectation value @xmath457^{-2 } , \nonumber\end{aligned}\ ] ] which we evaluate numerically by use of the metropolis monte carlo algorithm @xcite .
the renyi entropy of a conformal field theory with central charge @xmath22 is @xcite @xmath458+\textrm{constant},\ ] ] and @xmath459 for the @xmath413 wzw model ( recall ( [ su2 ] ) ) . in figure
[ s1entropy ] , we have plotted @xmath460 as a function of @xmath461/4 $ ] , and indeed the points fall almost on a straight line . considering @xmath22 as a fitting parameter , we find that @xmath462 is the best choice to describe the data , but @xmath459 also gives a fit that approximately follow the data points . ) for @xmath463 and @xmath464 obtained from monte carlo simulations of @xmath465 .
the data shown is the average of 40 trajectories with different initial conditions , and the variation of the mean of these trajectories is used to estimate the error bars .
the solid line is a fit of the form @xmath466 , where @xmath353 and @xmath22 are fitting parameters , and the dotted line is a fit of the same form but with @xmath22 fixed at @xmath467.,scaledwidth=60.0% ] and @xmath464 , @xmath468 and @xmath469 .
the points are obtained from monte carlo simulations , the solid lines are fits of the form @xmath470 $ ] , where @xmath353 and @xmath175 are fitting parameters , and the dotted lines are fits of the same form but with @xmath175 fixed at @xmath471 . as in the previous figure ,
the error bars are estimated from the variation of the mean of trajectories with different initial conditions.,scaledwidth=60.0% ] let us next consider the two - point spin correlation function . for the spin 1 model at level @xmath3 , we do not obtain a closed set of equations for the two - point spin correlators from ( [ ss20a ] ) and ( [ ss20b ] ) , so we use monte carlo simulations instead .
results for the uniform model and @xmath464 , @xmath468 and @xmath469 are shown in figure [ figcor ] .
affleck found that the spin correlation function for an infinite chain of spin 1 fields in the wzw model at level @xmath3 is proportional to @xmath472 for large @xmath9 @xcite , possibly with a multiplicative logarithmic correction of the form @xmath473 , where @xmath474 and @xmath475 are constants @xcite . referring back to ( [ ss20 ] ) and also to the scaling behaviour of the correlator in the dimer model discussed in section [ sec : dimer ] ,
it is very natural to expect that the most significant term of the correlation function for the finite spin 1 chain is proportional to @xmath476^b$ ] in the asymptotic region @xmath477 , where the exponent @xmath175 may or may not be equal to @xmath471 .
we therefore plot @xmath478 $ ] versus @xmath479 $ ] in figure [ figcor ] .
this transformation indeed gives data lying almost on straight lines despite the fact that we are considering chains of finite length . using a fit to extract the slope of the lines , we find that the exponent is around @xmath480 .
a similar exponent was found for the state considered in @xcite . as in @xcite
, we note that the deviation of @xmath175 from @xmath481 may be due to the presence of a logarithmic correction since a logarithmic correction of the type given above is equivalent to a power law to first order in the exponent : @xmath482 for @xmath483 small .
altogether , it is hence possible that the properties of the wave function are those of the @xmath413 wzw model , but it may also be that there are some deviations .
in conclusion , we have used a specific set of null vectors to derive quantum spin hamiltonians , whose ground states are the chiral correlators of the @xmath0 wzw model , and we have shown that the same null vectors lead to a set of algebraic , linear equations relating the spin correlators within each model . at level @xmath2 ,
these equations are sufficient to compute both two - point and higher - point spin correlators simply by solving a set of linear equations , which can even be done analytically in some cases .
the models involve hamiltonians with long - range interactions , and explicit analytical expressions for the corresponding ground state wave functions can be found in many cases .
the examples presented in the paper include both new spin models and new results for already known models .
it is interesting to note that our construction in some sense resembles that of the aklt @xcite bond in matrix product states ( mps ) and frustration free parent hamiltonians . as in mps
, the components of the wave functions are constructed from a product of @xmath4 operators ( see ( [ dec1 ] ) ) , where the @xmath53th operator depends only on the state of the @xmath53th spin , but the operators are chiral conformal fields instead of finite dimensional matrices .
similarly , the hamiltonian is a sum of @xmath4 hamiltonians ( see ( [ dec14 ] ) ) that annihilate the ground state individually , but the hamiltonians involve long - range interactions rather than only local interactions . these differences , in fact , circumvent one of the limitations of mps , namely how correlation functions may decay algebraically , and thus one can also use the states to describe critical behaviour . the uniform haldane - shastry model is closely related to the calogero - sutherland ( cs ) model , whose hamiltonian contains a kinetic term plus an inverse square coupling between particles moving on a circle @xcite ( see @xcite for a review ) . in turn ,
particular eigenfunctions of the cs model are in direct correspondence with the correlation functions in boundary conformal field theory .
in fact , the cs hamiltonian can be obtained from the null vector decoupling equations at level 2 of boundary fields and a bulk primary field at the centre of the disk @xcite .
these results suggest a possible relation of our work to those of @xcite , although the null vectors used there and those used in the present work are different .
the present paper raises the question , whether the properties of the wave functions are directly related to the properties of the fields used to construct the wave functions . the properties of the uniform model with @xmath2 and @xmath424 @xcite and the numerical results presented above for the two - point spin correlator and the renyi entropy for the uniform model with @xmath3 and @xmath421 suggest that the properties of the wave function may be similar to , but not necessarily identical to , the properties of the fields .
we are currently investigating this question in more detail .
there are several ways in which the explicit examples given in the paper can be generalized .
the construction in section [ sec : hamiltonians ] already includes the possibility to consider models at higher levels , models in which the fields used to construct the wave functions do not all have the same spin index , and models in which the individual terms @xmath484 are given different weights in the final hamiltonian .
furthermore , we have only considered one particular set of null vectors , but other hamiltonians can be derived by starting from other null vectors .
the present work shows that translationally invariant models of one - dimensional spin chains with long - range interactions can be achieved by choosing the @xmath321-coordinates to be uniformly distributed on the unit circle .
more generally , the @xmath321-coordinates are numbers in the complex plane , and various interaction patterns can be achieved by adjusting these coordinates .
this observation facilitates the construction of higher dimensional spin models .
in particular , it is natural to describe a two - dimensional spin lattice in the above framework by interpreting , e.g. , the real and imaginary parts of @xmath321 as the coordinates of the physical position of the @xmath53th spin in the lattice .
this construction will be presented elsewhere @xcite .
finally , the idea of finding hamiltonians from null vectors can also be applied to any lie group and to wzw models based on supergroups .
altogether , the present work thus opens up possibilities to find a huge class of spin models .
we conclude that null vectors are a powerful tool to gain insight into the behaviour of many - body systems , and we believe that more results will be obtained along these lines .
_ note added in proof . _
a spin 1 model with the same type of interactions as the model discussed in section 4 has been derived for the uniform case in @xcite , and an application of conformal field theory to propose wave functions for the fractional quantum hall effect is considered in @xcite .
the authors acknowledge discussions with d. haldane , d. doyon , g. mussardo , d. bernard , m. b. hastings , v. gurarie , c. nayak and h .- h .
this work has been supported by the carlsberg foundation , the eu project quevadis , and the grants fis2009 - 11654 and quitemad .
in @xcite , it was found by use of the knizhnik - zamolodchikov equation that the state ( [ psi2 ] ) is an eigenstate of the generalized hs hamiltonian ( [ hs3 ] ) .
we provide the details of this derivation in the following .
we start from the kz equation @xmath485 for the chiral correlator ( [ psi1 ] ) with components ( [ psi2 ] ) . using the conformal transformation @xmath486
, one can map the complex plane into the cylinder , where the chiral correlator becomes @xmath487 and @xmath488 is the conformal weight of the spin 1/2 primary field . expressing @xmath489 in terms of the @xmath142 variables and using ( [ psi2 ] ) for @xmath490
, one has @xmath491 the kz equation for this wave function is @xmath492 where @xmath493 as usually .
using the explicit expression for the wave function in ( [ psicyl ] ) and the relations @xmath494 one finds @xmath495 simplifying the above expression one gets an abelian version of the kz equation @xmath496 taking another derivative , one finds @xmath497 where we have used @xmath498 and @xmath499 summing over @xmath148 in the last expression , one gets @xmath500 the triple sum can be written as @xmath501 so that @xmath502 exchanging @xmath503 , this becomes @xmath504 using ( [ wid1 ] ) , ( [ wid2 ] ) and @xmath505 in ( [ dpsi ] ) , we find @xmath506 \\ = - \frac{1}{2 } \sum_{i \neq j } \left ( \frac { z_i z_j } { z_{ij}^2 } - \frac{1}{4 } w_{ij}^2 \right ) \ ;
s_i s_j + \frac{1}{16 } \sum_{i \neq j } w_{ij}^2 + \frac{n-2}{16 } \sum_{i \neq j } s_i s_j\nonumber\\ + \frac{1}{16 } \sum_{i \neq j } s_i s_j \ ; w_{ij } ( c_i - c_j ) = \frac{1}{8 } \sum_{i \neq j } s_i s_j + \frac{n-2}{16 } \sum_{i \neq j } s_i s_j + \frac{1}{16 } \sum_{i \neq j } w_{ij}^2 \nonumber\\ + \frac{1}{16 } \sum_{i \neq j } s_i s_j w_{ij } ( c_i - c_j ) = - \frac{n^2}{16 } + \frac{1}{16 } \sum_{i \neq j } w_{ij}^2 + \frac{1}{16 } \sum_{i \neq j } s_i s_j w_{ij } ( c_i - c_j ) .
\nonumber\end{aligned}\ ] ] from ( [ akz ] ) , we have @xmath507 so finally @xmath508 the operator on the left hand side of ( [ lhs1 ] ) can also be computed from the kz equation ( [ kz ] ) .
taking the derivative of ( [ kz ] ) and following the same steps as before , one finds @xmath509 the following identities are needed @xmath510 which finally lead to @xmath511\psi_{\rm cyl},\ ] ] where we have used @xmath512 because @xmath489 is a singlet .
it also follows from ( [ kz ] ) that @xmath513 and hence @xmath514 t_i^at_j^a + \frac{1}{48}\sum_{i \neq j } w_{ij}^2\\ - \frac{n ( n-2)}{48 } \bigg\ } \psi_{\rm cyl}.\nonumber\end{aligned}\ ] ] equating ( [ lhs1 ] ) and ( [ lhs2 ] ) , we arrive at @xmath515 t_i^at_j^a \ ; \psi_{\rm cyl } = \left [ \frac{1}{16 } \sum_{i \neq j } w_{ij}^2 - \frac{n ( n+1)}{16 } \right ] \psi_{\rm cyl}.\ ] ] since @xmath489 only differs from @xmath8 by an @xmath516-independent factor , we can replace @xmath489 by @xmath8 in ( [ hscs ] ) , and it follows that @xmath8 is an eigenstate of ( [ hs3 ] ) as claimed .
in @xcite , it was found numerically for small systems that @xmath8 is , in fact , the ground state of ( [ hs3 ] ) . with the results of the present paper , however
, it is clear from the construction that @xmath8 is the ground state .
we note also that the above derivation is not immediately generalizable to @xmath517 .
let us finally use the above approach to derive the generalization ( [ h3 ] ) to the nonuniform case of the inozemtsev invariant .
we start from the second derivative @xmath518 and take another derivative @xmath519 the terms involving 4 different spin variables vanish by virtue of the equation @xmath520 then , using the following identities @xmath521 we find after several manipulations @xmath522 hence , @xmath523 the left hand side can be computed using the kz equation . after a long computation one gets @xmath524 and hence @xmath525 , where @xmath264 is given by ( [ h3 ] ) .
the hamiltonian at level @xmath3 , which has the chiral correlator @xmath526 of @xmath4 spin @xmath5 fields as its ground state , follows from ( [ cg15 ] ) , ( [ dec10 ] ) , ( [ dec11 ] ) and ( [ dec14 ] ) . assuming @xmath211 @xmath152 , we find @xmath527 and @xmath528 t_{i_1}^at_{i_2}^at_{i_3}^bt_{i_4}^b.\nonumber\end{aligned}\ ] ] the spectrum of ( [ h212 ] ) for the uniform case and @xmath15 is given in table [ spectraonehalf ] . since two @xmath27 fields fuse to a @xmath26 and a @xmath31 field , and
since the fields should altogether fuse to @xmath26 to give a nonzero correlator , the ground state is @xmath529 times degenerate .
apart from this degeneracy , all degeneracies between spin multiplets can be explained from the invariance of the chain under translation by one site and under a flip of the chain .
this suggests that there is no yangian symmetry at level @xmath3 .
another difference compared to the uniform hs model is that not all energies are integers .
@llll @xmath273 & @xmath241 & @xmath274 & # + 0.0000 & 0 & -1 & 1 + & 0 & 0 & 1 + & 0 & 1 & 1 + & 0 & 3 & 1 + 0.2715 & 1 & 0 & 3 + 2.8686 & 1 & -2 & 3 + & 1 & 2 & 3 + @llll
+ 20.0000 & 1 & -1 & 3 + & 1 & 1 & 3 + 34.4000 & 2 & -2 & 5 + & 2 & 2 & 5 + 47.2000 & 2 & 3 & 5 + 52.0000 & 1 & 3 & 3 + 58.9285 & 1 & 0 & 3 + @llll + 107.2000 & 3 & 0 & 7 + 111.2000 & 2 & -1 & 5 + & 2 & 1 & 5 + 117.1314 & 1 & -2 & 3 + & 1 & 2 & 3 + 160.0000 & 0 & 3 & 1 + +
99 haldane f d m , ha z n c , talstra j c , bernard d and pasquier v , _ yangian symmetry of integrable quantum chains with long - range interactions and a new description of states in conformal field theory _ , 1992 2021 doyon b and cardy j , _ calogero - sutherland eigenfunctions with mixed boundary conditions and conformal field theory correlators _ ,
2007 _ j. phys . a : math .
_ * 40 * 2509 ( _ preprint _ arxiv : hep - th/0611054 ) greiter m , 2011 _ mapping of parent hamiltonians , from abelian and non - abelian quantum hall states to exact models of critical spin chains _ ( _ springer tracts in modern physics _ vol 244 ) ( berlin : springer ) | we propose to use null vectors in conformal field theories to derive model hamiltonians of quantum spin chains and corresponding ground state wave function(s ) .
the approach is quite general , and we illustrate it by constructing a family of hamiltonians whose ground states are the chiral correlators of the @xmath0 wzw model for integer values of the level @xmath1 .
the simplest example corresponds to @xmath2 and is essentially a nonuniform generalization of the haldane - shastry model with long - range exchange couplings . at level @xmath3
, we analyze the model for @xmath4 spin 1 fields .
we find that the renyi entropy and the two - point spin correlator show , respectively , logarithmic growth and algebraic decay .
furthermore , we use the null vectors to derive a set of algebraic , linear equations relating spin correlators within each model . at level @xmath2 ,
these equations allow us to compute the two - point spin correlators analytically for the finite chain uniform haldane - shastry model and to obtain numerical results for the nonuniform case and for higher - point spin correlators in a very simple way and without resorting to monte carlo techniques . _ keywords
_ : solvable lattice models , correlation functions , conformal field theory |
the promising and interesting physical properties of graphene@xcite have recently stimulated active research in other atomically thin materials , alternative and/or complementary to graphene.@xcite molybdenum disulfide ( mos@xmath0 ) , fabricated in its single - layer by means of mechanical exfoliation , exhibits a direct bandgap of 1.8 - 1.9 ev , contrary to the indirect band gap of its bulk counterpart.@xcite moreover , single - layer mos@xmath0 has also shown a mobility of 200 @xmath1vs which makes it appealing for the design of a new generation of more efficient transistors.@xcite in the field of spintronics , the absence of inversion symmetry in the crystal structure of single - layer mos@xmath0 allows valley polarization by optical pumping with circularly polarized light.@xcite this makes possible the design of devices based on spin and valley control.@xcite more recently , a remarkable seebeck coefficient has been measured in single - layer mos@xmath0 , opening a new field of application for those materials.@xcite concerning the optical properties , the photoluminescence in single - layers has shown higher efficiency than in multi - layers or bulk which is attributed to the direct/ indirect band - gaps , respectively .
the absorption spectra however are very similar in all the cases,@xcite an issue not yet explained .
the observed double - peak structure in the absorption spectra can be connected to the splitting of the valence band maximum around the high - symmetry point @xmath2 .
for single - layer mos@xmath0 , this splitting was explained as a consequence of spin - orbit coupling which is a result of the missing inversion symmetry.@xcite for bilayer and bulk , this splitting is mainly due to inter - layer interaction.@xcite in this context , reliable _ ab initio _ calculations of the absorption spectra are necessary for providing the correct interpretation of the reported experimental results .
however , the inherent complexity of the bethe - salpeter equation@xcite that is usually used to describe the excitonic effects in the optical spectra seems to be the reason for recent inadequately converged calculations . in ref .
the excitonic binding energy is strongly overestimated due to a low k - point sampling and in refs . and spin - orbit interaction was entirely neglected and the splitting of the excitonic peak is merely due to an unconverged k - point sampling .
the aim of our work is to provide well - converged optical spectra , in the framework of the bethe - salpeter equation _ including the effects of spin - orbit coupling_. this gives a reliable basis for the interpretation of previous experimental works on single- , double - layer and bulk mos@xmath0 .
we show that the the optical spectrum corresponds essentially to a step - function that is the result of the joint - density of states for parabolic dispersion in two - dimensional systems .
excitonic effects shift part of the oscillator strength into a discrete excitonic peak below the continuum onset .
the splitting of this excitonic peak can be directly related to the splitting of the valence band maximum around @xmath2 and is thus entirely due to spin - orbit coupling in the case of the single - layer and mostly ( but not entirely ) due to inter - layer interaction for double , and multi - layer mos@xmath0 .
starting point of the calculation of optical spectra are the kohn - sham wave functions and energies calculated with density - functional theory ( dft ) in the local - density approximation ( lda ) .
we use the code abinit@xcite where wave - functions are expanded in plane - waves and core electrons are simulated by norm - conserving relativistic pseudopotentials.@xcite the plane - wave energy cutoff is 30 a.u . for molybdenum , the 4s and 4p semi - core electrons
are explicitly calculated ( in addition to the 4d and 5s valence electrons ) .
this turns out to be crucial for the proper calculation of the exchange - contribution to the self - energy term in the gw calculations .
as noted earlier,@xcite spin - orbit interaction is important for mos@xmath0 and we calculate the spinor - wavefunctions as input for the following calculations on the level of many - body perturbation theory .
the inherent underestimation of the bandgap given by dft is corrected with the gw method.@xcite we use the non - self consistent version ( denoted as g@xmath3w@xmath3 ) without updating the dielectric function in the screened coulomb potential ( w ) or the wave - functions and energies in the green s function ( g ) .
these calculations are done with the yambo code.@xcite the dielectric function @xmath4 is calculated using the plasmon - pole approximation@xcite .
50 g - vectors are used ( for a vacuum distance of 40 a.u . between the periodic images of the single - layer / double - layer calculations ) .
200 unoccupied bands are used in the integration of the self - energy term , yielding converged band - gap corrections for single - layer , double - layer , and bulk .
the @xmath5-point sampling is @xmath6 for single and double - layer and @xmath7 for bulk .
the value of the gw - correction to the band - gap depends on the inter - layer distance in the periodic supercells approach .
it increases with increasing distance and converges roughly as @xmath8 ( where @xmath9 is the supercell dimension perpendicular to the layer ) .
this was shown for single layers of hexagonal boron nitride ( hbn)@xcite and for single - layers of mos@xmath0.@xcite at the same time , the excitonic binding energy was also shown to increase roughly as @xmath8 such that the two effects cancel and the resulting optical spectra hardly depend on the inter - layer distance . in our calculations , we use @xmath10 a.u . for the single layer calculations and of @xmath11 a.u . for the double - layer
starting from the kohm - sham wave - functions and the quasi - particle energies , the optical - spectra are calculated on the level of the bethe - salpeter equation ( bse):@xcite @xmath12 here , the electronic excitations are expressed in the basis of electron - hole pairs ( i.e. , vertical excitations at a given @xmath5-point from a state in the valence band with quasi - particle energy @xmath13 to a conduction - band state with energy @xmath14 .
the @xmath15 are the expansion coefficients of the excitons in the electron - hole basis and the @xmath16 are the eigenenergies ( corresponding to the possible excitation energies of the system ) .
if the interaction kernel @xmath17 is absent , eq .
( [ bethesalpeter ] ) simply yields @xmath18 , i.e. , the excitations of the system correspond to independent electron - hole pairs .
the interaction kernel @xmath17 describes the screened coulomb interaction between electrons and holes , and the exchange interaction , which includes the so called local fields effect .
@xmath17 `` mixes '' different single - particle excitations , from valence band states @xmath19 to conduction band states @xmath20 , giving rise to modified transition energies @xmath16 and ( possibly ) also to discrete excitonic states below the onset of the continuum . the optical absorption spectrum is given by the imaginary part of the dielectric function , @xmath21 , and can be calculated as @xmath22 where @xmath23 are the dipole matrix elements for electronic transitions from valence to conduction states .
we consider only light absorption with polarization , and thus the direction of the dipole operator @xmath24 , in the plane identified by the @xmath25 layer .
the out - of - plane polarization gives negligible contribution to absorption at low energies , because the local fields , which are strongly inhomogeneous in that direction , shift the oscillator strength to high energies .
excluding phonon - assisted transitions , the momentum @xmath26 is conserved in the absorption process .
we use an energy broadening @xmath27 ev in all the calculations to mimic the experimental resolution .
the bse calculations have been performed with the code yambo.@xcite since a much higher @xmath26-point sampling than for the gw - calculations is needed , we use lda wave - functions and energies , corrected by a `` scissor''-operator obtained from the gw - calculations .
while this approach does not take into account changes in the valence and conduction band dispersions , we have checked for the excitonic spectrum at a sampling of @xmath6 ( single - layer ) that the difference to a bse - calculation based on gw - energies is negligible .
first we study the electronic structure of single - layer , double - layer and bulk mos@xmath0 .
it is worth to mention that an accurate calculation of the exchange interaction in both the gw approximation and the bse requires the use of semi - core ( 4s and 4p ) orbitals for the molybdenum atoms .
the omission of semi - core states can lead to an erroneous wave - vector dependence of the gw - correction .
furthermore , the spin - orbit interaction has to be taken into account because it removes the degeneracy of the valence band maximum .
single - layer mos@xmath0 ( and , in general , odd number of layers ) belongs to the group of symmetry @xmath28,@xcite which lacks inversion symmetry .
this absence of symmetry , together with the strong spin - orbit coupling of molybdenum @xmath9 orbitals , splits the valence band edge at @xmath2.@xcite mos@xmath0 with even number of layers , such as the double - layer , and the bulk , instead , belong to the symmetry group @xmath29 , which does have inversion symmetry . nevertheless , also in this case a valence band splitting exists but it is caused ( predominantly ) by the interlayer interaction .
figure [ band ] shows the band structure , calculated with lda and gw , for single- , double - layer and bulk mos@xmath0 .
the only case where a direct bandgap is observed is the single - layer , either in lda or in gw , while for double - layer and bulk the bandgap is clearly indirect .
the origin of this difference is the interlayer interaction .
a second minimum in the conduction band lies on the high - symmetry line between @xmath30 and @xmath2 ( see vertical dashed lines ) . for the single - layer ,
this minimum is higher in energy than the one at @xmath2 but in the double - layer and bulk , the inter - layer interaction leads to a splitting which shifts the minimum to a lower energy than the minimum at @xmath2 .
moreover , in the double - layer and in the bulk , the valence band edge at @xmath30 rises higher than the valence band maximum at @xmath31 . as a consequence the system moves from a direct band - gap to an indirect gap semi - conductor .
the gw corrections are generally larger for single - layer mos@xmath0 then for the double - layer and the bulk , due to the smaller dielectric screening in the single - layer case .
our band - structures and their interpretation agree very well with the ones of the recent self - consistent gw calculations,@xcite even though details in the approximations are different : in our study , we do not use self - consistency in the gw - calculations which seems to add only a minor energy - shift to the g@xmath3w@xmath3 calculation . in turn
, we perform the gw calculations with the full spinor wave - functions while in ref .
spin - orbit coupling is introduced after the gw - calculations by rediagonalizing the hamiltonian matrices .
the values of the direct band gaps at @xmath2 are summarized in table [ band - table ] , combined with the valence band splitting . the values are in reasonable agreement with the ones of ref . .
small differences may be due to the differences in the approximations and from the different values of the cell dimensions .
we remark in this context the importance of the lattice constant .
small variations of the latter can shift the position of the conduction and valence band edges . commonly to other gw calculations , we thus use experimental lattice parameters ( in - plane lattice constant a=3.15 and c=12.3 for bulk@xcite ) in order to avoid artificial strain effects.@xcite a similar conclusion is drawn in a recent theoretical study about the influence of strain in the bandgap of mos@xmath0 , where strain tends to transform the single - layer mos@xmath0 into an indirect band - gap semiconductor.@xcite lccc + & 1-layer & 2-layer & bulk + @xmath32 ( lda ) & 1.69 & 1.68 & 1.67 + @xmath32 ( gw ) & 2.41 & 2.32 & 2.23 + @xmath33 ( lda ) & 134.3 & 173.8 & 220.1 + @xmath33 ( gw ) & 112.0 & 160.0 & 230.6 +
in the calculation of the optical absorption spectra of mos@xmath0 , the convergence with respect to @xmath26-sampling is of crucial importance in order to obtain reliable spectra ( see appendix ) .
we used a @xmath34 k - point in the case of the single - layer and double - layer and @xmath35 in the case of bulk .
local fields are included in all the calculations.@xcite we show in fig . [ bse ] , the calculated optical spectra for single- , double - layer and bulk mos@xmath0 .
the results of the bse are compared with the optical spectra without the excitonic effects , calculated in the random - phase approximation ( rpa , independent - particle picture ) using the gw - energies ( panels a - c ) .
the lda and gw band - gap energies are marked with vertical . finally ( panels d - f ) the bse results are compared with the experimental data@xcite . in panels
d - f the theoretical results are down - shifted by approximately 0.2 ev in order to compare with the experiments .
this discrepancy is in the margin error of common gw and bethe - salpeter calculations .
our results give a clear interpretation of the measured absorption spectra , in the low energy range , and provide reliable predictions within the given accuracy of @xmath36 0.2 ev , for the higher energy range , for which no experimental data are available .
the general features of the three optical spectra are very similar , with a double peak structure ( denoted as @xmath37 and @xmath38 ) at the energy threshold , accompanied by a plateau , and afterwards an abrupt increase of the optical absorption . the rpa spectrum below 2.8 ev resembles in all three cases the sum of two heaviside step functions .
the difference in the two step positions is given by the splitting of the valence band maximum at @xmath2 .
this step function profile is the fingerprint of the joint density of states of a 2d - system , with parabolic band structure .
`` switching on '' excitonic effects , preserves the plateau , but , in addition , a split excitonic peak below the onset of the continuum transition occurs .
the exciton binding energy is different in the three cases .
it is largest for the single - layer , where the electron - hole interaction is less screened .
the differences in quasi - particle band gap and excitonic binding energy almost cancel each other@xcite and the resulting excitonic peak positions are almost the same in the three cases . the peaks can be directly assigned to the peaks denoted @xmath37 and @xmath38 in the experiments.@xcite at higher energies , the relevant peak placed at 3.0 ev ( close to the blue color in the visible spectrum ) in the single - layer has larger relative intensity with respect to the peaks @xmath37 and @xmath38 .
such intense peak has not yet been reported , as the maximum detection energy in absorption experiments is around 2.4 ev.@xcite a further inspection of the excitonic eigenvectors allows to assign to each excitonic peak the contributing electron - hole transitions and their location at the brillouin zone . in all the cases , excitons @xmath37 and @xmath38
come from the energy transition at the @xmath2 point , even when the band gap is indirect .
this explains the similarity of the experimental optical absorption .
the other relevant peak , at 3.0 ev , comes from the part of the brillouin zone between @xmath2 and @xmath30 ( marked with a dashed line in fig . [ band ] ) , where we observe a high density of states due to the parallel conduction and valence bands.@xcite , obtained with the bs equation ( solid lines ) and rpa ( dashed lines ) .
lda and gw band gaps at @xmath2 are marked with vertical dashed lines , the absorption threshold is marked with a vertical solid line .
note that the wiggles above 2.5 ev in panel ( c ) are due to finite k - point sampling .
panels ( d)(f ) : symbols : experimental absorption spectra@xcite in comparison with the calculations ( solid lines , shifted by about -0.2 ev ) . ] to gain further insight of the optical spectra , we represent the excitonic wavefunctions related to the main peaks of the bethe - salpeter spectra .
the excitonic wave - function can be written as : @xmath39 where @xmath40 and @xmath41 are the real - space electron and hole coordinate and @xmath42 the lda wave functions .
the coefficients @xmath43 are obtained by diagonalizing the hamiltonian of the bethe - salpeter equation ( with energy @xmath16 ) . in order to represent
the six - coordinate function , we fix the hole position 1.0 above a molybdenum atom and we project onto the function @xmath44 onto the x - y plane . figure [ exciton](a ) shows the exciton wave function of the exciton @xmath37 of the mos@xmath0 single - layer ( for double - layer and bulk the wave function is essentially identical ) .
this exciton is largely spread , extended over 65 , in concordance with the small binding energy and with the small effect on the absorption threshold , as reported in the experiments.@xcite this also explains the large @xmath26-point grid needed to converge the results in the bethe - salpeter equation ( see appendix ) .
the exciton @xmath38 ( not shown here ) shows similar trends . on the opposite , in fig .
[ exciton](b ) we observe that the brightest exciton , at 3.0 ev , is remarkably localized , being confined to less than 30 , with a trigonal shape . among the properties of such exciton we point out the potentially high efficiency of recombination .
another interesting point ( for multi - layer and bulk ) is to explore if the excitons are confined in one layer or if their wavefunctions extend over several layers . with this aim
we show in fig .
[ exciton](d ) the exciton @xmath37 of the mos@xmath0 bulk .
the wave function spreads largely within the plane but it is undoubtedly constricted to one layer . the density in neighboring layers is negligible and the wave - function is very similar to the one of the @xmath37 exciton in the single layer ( fig .
[ exciton](c ) ) .
the large interlayer distance , due to the weak van der waals interaction , prevents the wave function from spreading to other layers , analogously to what happens in bulk boron nitride.@xcite
we have performed calculations of the quasi - particle band - structure and of the optical absorption spectra of single , double - layer , and bulk mos@xmath0 including excitonic effects and spin - orbit coupling at the same time . in agreement with previous calculations and with experimental evidence from photoluminescence
intensities,@xcite our gw - calculations demonstrate that only single - layer mos@xmath0 has a direct band gap at @xmath2 .
the inter - layer interaction makes the band gap indirect for multi - layer and bulk mo@xmath45 .
however , this does not particularly influence the optical absorption spectra in the visible range .
these display in all cases a strong excitonic peak ( composed of electron - hole pairs around @xmath2 ) below the onset of continuum transitions which has the step function shape , corresponds to the quasi-2d joint - density of states . the layered - structure of the material confines the first exciton mostly in a single - layer which causes the similarity of the optical spectrum in all the cases .
the position of the split excitonic peak is remarkably stable with respect to the number of layers ( and with respect to the inter - layer spacing in the calculation for single - layer mos@xmath0 ) .
this has been previously explained as a cancellation effect between the band - gap correction due to electron - electron interaction and the excitonic binding energy due to electron - hole interaction@xcite .
the splitting of the excitonic peak is directly related to the splitting of the valence band maximum around @xmath2 and is entirely due to spin - orbit coupling for the single - layer and ( mostly ) due to inter - layer interaction for the double - layer and bulk . at higher energy ( around 3 ev ) , the optical spectrum is dominated by electron - hole pairs from transitions around the center of the line @xmath46 where both valence and conduction bands have a minimum and the parallel shape of the bands causes a maximum in the joint density - of - states . for the single - layer a strongly - bound exciton causes an additional peak in the spectrum at 3.0 ev . for double - layer and bulk
, the shape of the spectrum changes and may allow for a spectroscopic distinction between layer numbers .
the authors acknowledges financial support from the anr ( french national research agency ) through project anr-09-blan-0421 - 01 .
calculations were done at the idris supercomputing center , orsay ( proj .
091827 ) , and at the tirant supercomputer of the university of valencia ( group vlc44 ) .
the convergence of the optical spectra is a mandatory issue to obtain reliable spectra , comparable with the experimental data .
we have performed calculations to check the convergence of the spectra , and in particular of the position of the first excitonic peak with respect to the @xmath26-point sampling .
a fine sampling can be quite costly in terms of computation time and in some bethe - salpeter calculations may reach unexpectedly large values .
figure [ convergence ] shows the bethe - salpeter spectra of mos@xmath0 single - layer for several @xmath26-grids ( for simplicity we have omitted the spin - orbit coupling ) .
additionally , we have represented the exciton binding energy , @xmath47 , as a function of the number of irreducible @xmath26-points , marking in each case the corresponding grid .
we observe the slow convergence of the first excitonic peak , not reached before a @xmath48 sampling .
moreover , for small grids some secondary and artificial peaks appear in the spectrum and the underlying heaviside function of the optical absorption is not reached unless dense grids are used ( @xmath49 ) .
this value is considerably larger than the one used in ref where only a @xmath50 grid was used , leading to isolated peaks in the absorption spectra where the density of states predicts a plateau following the excitonic peak .
our test calculations also explain why in refs . and
split excitonic peaks were observed even though spin - orbit coupling was not included .
the exciton binding energy @xmath47 also needs at least an @xmath51 grid to be converged within 0.1 ev .
it is worth to note that the second excitonic peak ( the most intense ) , located at 2.5 ev converges much faster , approximately for a @xmath52 grid .
the difference behavior of those excitons can be better understood by inspection of their wave functions , shown in fig .
[ exciton](a ) and ( b ) . in the case of the first peak ,
the wave function is very spread , more than 18 unit cells in each direction on the plane .
we remind that a @xmath53 grid in the reciprocal space allows only to map the exciton wave function in the real space @xmath53 the unit cell , therefore , inadequately small grids lead to an artificial confinement of the exciton which increases its binding energy .
this arguments also clarifies why the second exciton converge much faster , its wave function being confined within a few unit cells .
a. splendiani , l. sun , y. zhang , t. li , j. kim , c. y. chim , g. galli , and f. wang , nano lett .
* 10 * , 1271 ( 2010 ) .
ngel k. f. mak , c. lee , j. hone , j. shan , and t. f. heinz , phys .
* 105 * , 136805 ( 2010 ) . in order to speed up the calculation to obtain smooth spectra ,
we have also tested the double - grid method proposed in ref . .
in this case a basic grid of @xmath54 for single - layer and double - layer yields almost the same results as the @xmath34 grid . | we present converged ab - initio calculations of the optical absorption spectra of single - layer , bi - layer , and bulk mos@xmath0 .
both the quasiparticle - energy calculations ( on the level of the gw approximation ) and the calculation of the absorption spectra ( on the level of the bethe - salpeter equation ) explicitly include spin - orbit coupling , using the full spinorial kohn - sham wave - functions as input . without excitonic effects , the absorption spectra would have the form of a step - function , corresponding to the joint - density of states of a parabolic band - dispersion in 2d .
this profile is deformed by a pronounced bound excitonic peak below the continuum onset .
the peak is split by spin - orbit interaction in the case of single - layer and ( mostly ) by inter - layer interaction in the case of double - layer and bulk mos@xmath0 .
the resulting absorption spectra are thus very similar in the three cases but the interpretation of the spectra is different .
differences in the spectra can be seen around 3 ev where the spectra of single and double - layer are dominated by a strongly bound exciton . |
ion clustering has long been an essential ingredient of our physical understanding of electrolyte solutions at elevated concentrations .
@xcite to describe pairing of a counter - ion of type @xmath1 with an ion of type @xmath2 , we focus on the radial distribution of the _ closest _ @xmath1-ion to a distinguished @xmath2-ion .
we denote that normalized radial distribution by @xmath3 .
a famous discussion of fuoss @xcite arrived at the approximation @xmath4 for a primitive model of 1 - 1 electrolyte as in fig . 1 .
here @xmath5 is the magnitude of the formal ionic charges , @xmath6 is the distance of closest approach , @xmath7 is the solution dielectric constant , @xmath8 is number density of ions , and @xmath9 is the temperature .
we propose and test generalizations of eq . in the following .
several complications of the distributions of near - neighbor ion - pairs motivate the generalizations that we develop .
firstly , ion - clustering can be particularly sensitive to non - ionic interactions . comparison ( fig . 1 ) of atomically - detailed simulation results
@xcite with those of a corresponding primitive model @xcite straightforwardly exemplifies that point .
. only treats classic ionic interactions . secondly , even for primitive models
the fuoss approximation can be unsatisfactory ( fig . 2 ) .
thirdly , nearest - neighbor distributions generally depend on which ion of an ion - pair is regarded as the central ion ( fig . 3 ) .
the radial distribution of the anion nearest to a cation is different from the radial distribution of the cation nearest to an anion , @xmath10 . the approximation eq .
is symmetric @xmath11 we are lead then to generalizations by recalling that the probability that a ball of radius @xmath12 centered on an @xmath2-ion is _ empty _ of @xmath1-ions can be obtained from @xmath13 the assessment of the probability that the _ nearest _
@xmath1-ion is further away than @xmath12 .
the simple estimate @xmath14 with @xmath15 @xmath16 the density of @xmath1 ions , and @xmath17 the conventional radial distribution function , follows from the assumption of the poisson distribution for that probability . evaluating the derivative of eq . using eq .
gives @xmath18 for @xmath19 ( no correlations ) , this is the hertz distribution that is correct for that case.@xcite we recover the fuoss approximation with @xmath20 for @xmath21 , and zero ( 0 ) otherwise .
this derivation of the fuoss approximation eq . seems not to have been given before .
nevertheless , the suggested approximation eq . is a standard idea in the context of scaled - particle theories of the hard - sphere fluid.@xcite [ bf@xmath0/pc ) at @xmath22 = 300 k , @xmath23 = 1 atm , and the slightly supersaturated concentration of 1 mol / dm@xmath24 .
@xmath25 is the radial distribution of the nearest b - neighbor of an n - atom .
lower panel : results for a corresponding primitive model with dielectric constant and with ion charges and sizes matched to the [ tea][bf@xmath0/pc case above .
specifically the model dielectric constant is @xmath26 , and @xmath27 nm , @xmath28 nm , @xmath29 nm are distances of closest approach for the hard spherical ions .
the lower panel was produced by monte carlo simulation of a neutral system of 2@xmath30500 hard spherical ions in conventional cubical periodic boundary conditions at the same temperature and concentration as the results above , utilizing the towhee@xcite package adapted to the present system.[fig : teabf4_rdf],title="fig:",width=240 ] [ bf@xmath0/pc ) at @xmath22 = 300 k , @xmath23 = 1 atm , and the slightly supersaturated concentration of 1 mol / dm@xmath24 .
@xmath25 is the radial distribution of the nearest b - neighbor of an n - atom .
lower panel : results for a corresponding primitive model with dielectric constant and with ion charges and sizes matched to the [ tea][bf@xmath0/pc case above .
specifically the model dielectric constant is @xmath26 , and @xmath27 nm , @xmath28 nm , @xmath29 nm are distances of closest approach for the hard spherical ions .
the lower panel was produced by monte carlo simulation of a neutral system of 2@xmath30500 hard spherical ions in conventional cubical periodic boundary conditions at the same temperature and concentration as the results above , utilizing the towhee@xcite package adapted to the present system.[fig : teabf4_rdf],title="fig:",width=240 ] as discussed below , the poisson result eq
. follows from a maximum entropy development when the information supplied is the expected occupancy of the inner - shell.@xcite that information is sufficient if the occupancy @xmath31 is always low , _
i.e. _ , rarely larger than one .
thus , in contrast to the fuoss approximation , eq .
is correct for small @xmath12 because the expected coordination number tends to zero then .
for the same reason , the poisson approximation eq . is correct at low electrolyte concentration , and even when the solvent is treated at atomic resolution .
furthermore , it is natural to guess cation - anion chain or ring structures when ionic interactions drive well developed clustering .
1 shows a mean coordination number of less than two for counter - ion neighbors closer than about 0.5 nm , and supports the chain / ring picture of ion clusters formed .
it is plausible therefore that a choice of inner - shell radii leading to small coordination numbers should validly describe important features of well - developed ion - clustering . for computational analysis of reactive bi - molecular encounters in solution , identification of geometries of closest molecular pairs
is critical.@xcite because it is correct for low concentration and for small @xmath12 in any case , eq .
should be regarded as the general resolution of those questions .
when coordination numbers exceed one with reasonable probability , information on the expected number of pairs of counter - ions in the inner - shell should improve a maximum entropy model of these probabilities.@xcite a maximum entropy model involving pair information would predict the @xmath10 asymmetry . for a 1 - 1 electrolyte , the generalization eq .
is symmetrical in accord with the fuoss approximation .
the extent to which the observed asymmetry is significant gives an indication whether the poisson approximation is adequate .
to monte carlo results of fig . 1 .
[ fig : mc_fuoss_primitive],width=240 ] , width=240 ] in this work , the poisson approximation ( eq . ) is tested using three distinct simulation data sets .
two of these data sets have been noted already in considering fig .
those calculations treated solutions of tetraethylammonium tetrafluoroborate in propylene carbonate , one at atomic resolution ( [ tea][bf@xmath0/pc ) and the other on the basis of a primitive electrolyte solution model over a range of concentrations .
the third data set treated the ionic liquid 1-butyl-3-methylimidazolium tetrafluoroborate ( [ bmim][bf@xmath0 ) . to ensure the correct correspondence of the necessary simulation details with the results as they are discussed ,
those details are provided in the captions of the figures providing the simulation results .
for [ tea][bf@xmath0/pc , comparison ( fig .
4 ) of the numerical data with the approximation eq . shows agreement over a distance range wider than the sizes of the molecules as judged by the radial distributions ( fig . 1 ) .
these near - neighbor distributions show bi - modal probability densities with maxima at @xmath32 0.5 nm and 0.9 nm . these correspond , respectively , to a contact ion pair and to a solvent - separated near - neighbor ion - pair .
thus the poisson approximation eq .
in this case includes solvation structure in characterizing inter - ionic neighborship .
[ bf@xmath0/pc case of fig . 1 , comparison of the numerical data with the approximation eq . .
the local maximum at @xmath32 0.9 nm identifies solvent - separated nearest - neighbor ion - pairs . in this case ,
asymmetry of the two observed near - neighbor distributions is slight .
that suggests that more elaborate maximum entropy models are unnecessary , and indeed the poisson approximation is accurate .
the embedded molecular graphic shows one of the solvent - separated nearest - neighbor bf@xmath33 pc tea@xmath34 structures observed .
[ fig : g1poissontea],width=240 ] for the [ tea][bf@xmath0/pc case of fig . 1 .
upper panel : probabilities for occupancy by the b - atom of the bf@xmath33 anion of the inner - sphere of the n - atom of tetraethylammonium cation ( tea@xmath34 ) .
lower panel : probabilities for occupancy by the n - atom of the cation of the inner - sphere of the b - atom of the anion .
the curves lowest in each panel are similar , showing symmetry displayed also in fig .
[ fig : probability_teabf4],title="fig:",width=240 ] for the [ tea][bf@xmath0/pc case of fig . 1 .
upper panel : probabilities for occupancy by the b - atom of the bf@xmath33 anion of the inner - sphere of the n - atom of tetraethylammonium cation ( tea@xmath34 ) .
lower panel : probabilities for occupancy by the n - atom of the cation of the inner - sphere of the b - atom of the anion .
the curves lowest in each panel are similar , showing symmetry displayed also in fig .
[ fig : probability_teabf4],title="fig:",width=240 ] a plateau between @xmath32 0.5 nm and 0.9 nm in occupancy probabilities ( fig .
5 ) indicates saturation of counter - ion probability , and marks the inter - shell region . at the distance @xmath12 indicated by the vertical line
, the coordination numbers @xmath35 = 1 , 2 predominate , supporting the idea of the formation of cation - anion chain and ring structures .
the two sets of probabilities ( fig .
5 ) are qualitatively similar , reinforcing the symmetry of fig . 4 . , @xmath22 = 300k for each case .
for the highest concentration , the system size is 2@xmath30400 ions .
for all other cases , the system size is 2@xmath30200 ions . at the lowest concentration here
the distribution of the nearest neighbor @xmath36 is close to the full radial distribution function @xmath37 .
[ fig : mc_poisson_primitive],width=240 ] results ( fig .
6 ) for the primitive model of fig . 1
examine the sufficiency of the poisson approximation over a broader concentration range for such models .
the nearest - neighbor distributions are unimodal in this case .
correct at small @xmath12 where the probability densities are highest and properly normalized , the poisson approximation eq .
is accurate over the whole range shown . )
( top ) and anion tetrafluoroborate ( bf@xmath33 ) . on the cation molecule
, white , turquoise and dark blue balls stand for hydrogen , carbon , and nitrogen atoms , respectively . on the anion
, white and green balls stand for fluorine and boron atoms , respectively .
[ fig : bmimbf4_structure],title="fig:",width=288 ] ) ( top ) and anion tetrafluoroborate ( bf@xmath33 ) . on the cation molecule
, white , turquoise and dark blue balls stand for hydrogen , carbon , and nitrogen atoms , respectively . on the anion
, white and green balls stand for fluorine and boron atoms , respectively .
[ fig : bmimbf4_structure],title="fig:",width=67 ] cation and boron on bf@xmath33 anion obtained from md simulation . force field parameters and partial charge of the atoms were taken from andrade _
et al_.@xcite the initial unit cell , with a dimension of 40 @xmath30 40 @xmath30 40 is uniformly packed with 190 ion pairs using packmol.@xcite md simulation was performed using amber10 at constant pressure ( 1 bar ) and temperature ( 298.5k ) .
the system was first minimized , followed by 0.2ns equilibrium at a time step of 0.2 fs , then 1.3 ns equilibrium at a time step of 2 fs .
radial distribution functions were extracted from a production run of 3.0 ns .
[ fig : bmimbf4_rdf],width=240 ] , width=240 ] another example is the ionic liquid [ bmim][bf@xmath0 , with molecular structure shown in fig . 7 and radial distribution functions in fig . 8
. the poisson approximation ( eq . ) agrees with the observed @xmath38 at short range and displays a second maximum characterizing non - contact nearest neighbors , though in this case there is no additional solvent .
the near - neighbor b@xmath39c2 distribution ( fig .
9 ) , on the other hand , lacks a second maximum .
thus @xmath40 and @xmath41 for ionic liquid [ bmim][bf@xmath0 display the generally expected asymmetry .
this asymmetry is also reflected in occupancy probability profiles ( fig .
more general theoretical models are required for such cases , and we return to that theoretical discussion now . for [ bmim][bf@xmath0 ( fig .
upper panel : probabilities for occupancy by the b - atom of the bf@xmath33 anion of the inner - sphere of the mid - c - atom of 1-butyl-3-methylimidazolium ( bmim@xmath34 ) cation ( labeled as c2 in fig . 7 ) .
lower panel : probabilities for occupancy by the mid - c - atom of the cation of the inner - sphere of the b - atom of the anion .
these results illustrate the possibility of a structural motif of ion clusters as chains and rings , _
i.e. _ at the @xmath12 = 0.43 nm distance of the vertical dashed line probabilities of 1 and 2 neighbors predominate .
they also demonstrate asymmetry of the distributions of near - neighbor distances in their dependence on choice of the central ion , _
i.e. _ the distribution of nearest anions to a cation is different from the distribution of the nearest cations to an anion .
since @xmath42 is larger in the lower panel than in the upper panel , the bf@xmath33 anion is more likely to be a three - way junction in this analysis than is the bmim@xmath34 cation .
[ fig : probability_bmimbf4],title="fig:",width=240 ] for [ bmim][bf@xmath0 ( fig .
upper panel : probabilities for occupancy by the b - atom of the bf@xmath33 anion of the inner - sphere of the mid - c - atom of 1-butyl-3-methylimidazolium ( bmim@xmath34 ) cation ( labeled as c2 in fig . 7 ) .
lower panel : probabilities for occupancy by the mid - c - atom of the cation of the inner - sphere of the b - atom of the anion .
these results illustrate the possibility of a structural motif of ion clusters as chains and rings , _
i.e. _ at the @xmath12 = 0.43 nm distance of the vertical dashed line probabilities of 1 and 2 neighbors predominate .
they also demonstrate asymmetry of the distributions of near - neighbor distances in their dependence on choice of the central ion , _
i.e. _ the distribution of nearest anions to a cation is different from the distribution of the nearest cations to an anion .
since @xmath42 is larger in the lower panel than in the upper panel , the bf@xmath33 anion is more likely to be a three - way junction in this analysis than is the bmim@xmath34 cation .
[ fig : probability_bmimbf4],title="fig:",width=240 ]
the poisson distribution @xmath43 describes random occupancy consistent with the information @xmath44 . considering the relative entropy , @xmath45
the poisson distribution is a maximum entropy distribution satisfying the specific expected occupancy .
if we have more information , _ e.g. , _ the binomial moments@xcite @xmath46 we can seek the distribution which maximizes @xmath47 and satisfies the broader set of information . with the binomial moments ( eq . ) , the poisson distribution is seen to be correct if realized values of @xmath35 are rarely bigger than one ( 1 ) . if @xmath35 is never 2 or larger , binomial moments @xmath48 vanish . when @xmath48 binomial moments are small , and that is consistent with poisson prediction that they are zero .
this underlies our observation above the the poisson model , @xmath49 of eq . , is correct for small @xmath50 . beyond the mean occupancy ,
the next level of information is the pair - correlation information @xmath51 , the expected number of pairs of counter - ions in the indicated inner - shell .
carrying - out the maximization for the case that pair information is available induces the model @xmath52 , where @xmath53 , and @xmath54 are lagrange multipliers adjusted to reproduce the information @xmath55 and @xmath56 . explicitly addressing the normalization of these probabilities leads to @xmath57 and @xmath58
@xmath59 involves only the denominator of eq . , and can be considered a partition function sum over occupancy states with @xmath35-dependent interactions and interaction strengths adjusted to satisfy the available information .
the information required ( fig . 11 ) for this augmented maximum - entropy model is only subtly different for the two cases . nevertheless , the results ( fig .
12 ) agree nicely with the observed asymmetry . on a b atom . dashed curve : the number of b pairs occupying a sphere of radius @xmath12 on a c2 atom .
[ fig : pairinfo],width=240 ] probabilities of fig . 10 .
solid curves : direct numerical simulation as in fig .
10 ; dashed curve : poisson - based approximation , eq . ; dotted curves : augmented maximum entropy model utilizing the first two binomial moments , eq . .
the two - moment maximum entropy model is qualitatively reliable and therefore gives a satisfactory explanation of the observed asymmetry .
[ fig : maxent],width=240 ]
results for both the [ tea][bf@xmath0/pc ( fig . 1 ) and the ionic liquid [ bmim][bf@xmath0 ( fig .
8) identify a natural clustering radius where mean coordination numbers are near two .
this suggests arrangements of the closest neighbors leading to a structural motif of cation - anion chains and rings .
in contrast to the atomically detailed [ tea][bf@xmath0/pc results , a corresponding primitive model ( fig .
1 ) does not display those clustering signatures ( fig .
a generalization ( eq . ) of the fuoss ion - pairing model was obtained by recognizing that the poisson distribution is correct when the mean coordination numbers are low . on the basis of measurable molecular distribution functions , this generalization also establishes the distribution of molecular nearest neighbors for computational analysis of bi - molecular reactive processes in solution .
this poisson - based model is accurate for the [ tea][bf@xmath0/pc results , both for the primitive model and the atomically detailed case . for [ tea][bf@xmath0/pc ,
the atomically detailed numerical results and the statistical model identify _ solvent - separated _ nearest - neighbor ion - pairs .
distributions of nearest - neighbor distances typically depend on which ion of a pair is taken as the central ion , _
i.e. , _ the distribution of anions nearest to a cation is different from the distribution of the cations nearest to an anion .
the poisson - based model is _ not _ asymmetric in that way .
the numerical data for the ionic liquid [ bmim][bf@xmath0 prominently show the expected asymmetry .
that asymmetry can be treated by a maximum entropy model based on the expected number of _ pairs _ of counter - ions occupying the inner - shell of the central ion , information extracted from the simulations .
23ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ] + 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty @noop _
( , , ) @noop * * , ( ) @noop * * , ( ) @noop _ _ ( , ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop `` , '' ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) @noop * * , ( ) | an elementary statistical observation identifies generalizations of the fuoss approximation for the probability distribution function that describes ion clustering in electrolyte solutions .
the simplest generalization , equivalent to a poisson distribution model for inner - shell occupancy , exploits measurable inter - ionic correlation functions , and is correct at the closest pair distances whether primitive electrolyte solutions models or molecularly detailed models are considered , and for low electrolyte concentrations in all cases . with detailed models
these generalizations includes non - ionic interactions and solvation effects .
these generalizations are relevant for computational analysis of bi - molecular reactive processes in solution .
comparisons with direct numerical simulation results show that the simplest generalization is accurate for a slightly supersaturated solution of tetraethylammonium tetrafluoroborate in propylene carbonate ( [ tea][bf@xmath0/pc ) , and also for a primitive model associated with the [ tea][bf@xmath0/pc results . for [ tea][bf@xmath0/pc
, the atomically detailed results identify solvent - separated nearest - neighbor ion - pairs .
this generalization is examined also for the ionic liquid 1-butyl-3-methylimidazolium tetrafluoroborate ( [ bmim][bf@xmath0 ) where the simplest implementation is less accurate . in this more challenging situation
an augmented maximum entropy procedure is satisfactory , and explains the more varied near - neighbor distributions observed in that case . |
matrix variate distributions have proven to be useful for modelling three - way data , such as multivariate longitudinal data . however , in most cases , the underlying distribution has been elliptical such as the matrix variate normal and the matrix variate @xmath0 distributions .
however , there has been relatively little work done on matrix variate data that can account for skewness present in the data .
the work that has been carried out in the area of matrix variate skew distributions is mostly limited to the matrix variate skew - normal distribution .
herein , we derive a matrix variate skew-@xmath0 distribution .
the remainder of this paper is laid out as follows . in section 2 ,
some background is presented . in section 3 ,
the density of the matrix variate skew-@xmath0 distribution is derived and a parameter estimation procedure is given .
section 4 looks at some simulations , and we conclude with a summary and some future work ( section 5 ) .
one natural method to model three - way data is to use a matrix - variate distribution .
there are many examples in the literature of such distributions , the most well - known being the matrix - normal distribution . for notional clarity , we use @xmath1 to denote a realization of a random matrix @xmath2 .
an @xmath3 random matrix @xmath2 follows a matrix variate normal distribution with location parameter @xmath4 and scale matrices @xmath5 and @xmath6 of dimensions @xmath7 and @xmath8 , respectively .
we write @xmath9 to denote such a random matrix and the density of @xmath2 can be written @xmath10 one well known property of the matrix variate normal distribution @xcite is @xmath11 where @xmath12 is the multivariate normal density with dimension @xmath13 , @xmath14 is the vectorization of @xmath4 , and @xmath15 is the kronecker product .
although the matrix variate normal is arguably the most mathematically tractable , there are examples of non - normal cases .
one famous example is the wishart distribution @xcite arising as the distribution of the sample covariance matrix of a multivariate normal sample .
more recently , however , there has been some work done in the area of matrix skew distributions such as the matrix - variate skew normal distribution , e.g. , @xcite , @xcite , and @xcite .
more information on matrix variate distributions can be found in @xcite .
very recently , there has also been work done in the area of finite mixtures .
specifically , @xcite looked at clustering and classification of multivariate longitudinal data using a mixture of matrix variate normal distributions .
also , @xcite , looked at mixtures of matrix variate @xmath0 distributions .
various multivariate distributions such as the multivariate @xmath0 , and skew-@xmath0 , the shifted asymmetric laplace distribution , and the generalized hyperbolic distributions arise as special cases of a normal variance - mean mixture ( cf . * ? ? ?
* ch . 6 ) . in this formulation ,
the density of a @xmath16-dimensional random vector @xmath1 takes the form @xmath17 which is equivalent to the representation @xmath18 where @xmath19 and @xmath20 is a latent random variable with density @xmath21 .
the multivariate skew-@xmath0 distribution with @xmath22 degrees of freedom arises as a special case with @xmath23 , where @xmath24 denotes the inverse gamma distribution with density function @xmath25 a random variable @xmath26 has a generalized inverse gaussian ( gig ) distribution with parameters @xmath27 and @xmath28 if its density function can be written as @xmath29 where @xmath30 is the modified bessel function of the third kind with index @xmath28 . several functions of gig random variables have tractable expected values , e.g. , @xmath31
@xmath32 @xmath33 where @xmath34 , @xmath35 , and @xmath36 is the modified bessel function of the third kind with index @xmath28 .
these results will prove to be useful for parameter estimation for the matrix - variate skew-@xmath0 distribution .
we will say that an @xmath3 random matrix @xmath2 has a matrix variate skew-@xmath0 distribution , @xmath37 , if @xmath2 can be written @xmath38 where @xmath4 and @xmath39 are @xmath3 matrices , @xmath40 , and @xmath23 .
analogous to its multivariate counterpart , @xmath4 is a location matrix , @xmath39 is a skewness matrix , @xmath5 and @xmath6 are scale matrices , and @xmath22 is the degrees of freedom .
it then follows that @xmath41 and thus the joint density of @xmath2 and @xmath42 is @xmath43\right\ } , { \addtocounter{equation}{1}\tag{\theequation}}\label{eqn : joint}\end{aligned}\ ] ] where @xmath44 .
we note that the exponential term in can be written as @xmath45\right\},\ ] ] where @xmath46 therefore , the marginal density of @xmath2 is @xmath47\right\}dw.\end{aligned}\ ] ] making the change of variables given by @xmath48 we can write @xmath49\left[\delta({\mathbf{x}};{\mathbf{m}},{\mathbf\sigma},{\mathbf\psi})+\nu\right]}\right).\end{aligned}\ ] ] the density of @xmath2 , as derived here , is considered a matrix variate extension of the multivariate skew-@xmath0 density used by @xcite .
as discussed by @xcite and @xcite in the matrix variate normal case , the estimates of @xmath5 and @xmath6 are unique only up to a multiplicative constant .
indeed , if we let @xmath50 and @xmath51 , @xmath52 , the likelihood is unchanged .
this identifiability issue can be resolved , for example , by setting @xmath53 or @xmath54 .
note that @xmath55 , so the estimate of the kronecker product is unique . for the purposes of parameter estimation , note that the conditional density of @xmath42 is @xmath56^{\frac{\lambda}{2}}w^{\lambda-1}}{2k_{\lambda}(\sqrt{\rho({\mathbf{a}},{\mathbf\sigma},{\mathbf\psi})[\delta({\mathbf{x}};{\mathbf{m}},{\mathbf\sigma},{\mathbf\psi})+\nu}])}\exp\left\{-\frac{\rho({\mathbf{a}},{\mathbf\sigma},{\mathbf\psi})w+[{\delta({\mathbf{x}};{\mathbf{m}},{\mathbf\sigma},{\mathbf\psi})+\nu}]/{w}}{2}\right\}.\end{aligned}\ ] ] therefore , @xmath57 where @xmath58 .
finally , we note that @xmath59 where @xmath60 denotes the multivariate skew-@xmath0 distribution with location parameter @xmath14 , skewness parameter @xmath61 , scale matrix @xmath62 , and @xmath22 degrees of freedom .
this can be easily seen from the representation given in and the property of the matrix normal distribution given in .
note that the normal variance - mean mixture representation as well as the relationship with the multivariate skew-@xmath0 distribution present two convenient methods to generate random matrices from the matrix variate skew @xmath0 distribution .
the former is used in section 4 .
suppose we observe a sample of @xmath63 matrices @xmath64 from an @xmath3 matrix variate skew-@xmath0 distribution .
as with the multivariate skew-@xmath0 distribution , we proceed as if the observed data is incomplete , and introduce the latent variables @xmath65 .
the complete - data log - likelihood is then @xmath66-\frac{1}{2}\sum_{i=1}^nw_i{\,\mbox{tr}}({\mathbf\sigma}^{-1}{\mathbf{a}}{\mathbf\psi}^{-1}{\mathbf{a}}'),\end{aligned}\ ] ] where @xmath67 is constant with respect to the parameters .
we proceed by using an expectation - conditional maximization ( ecm ) algorithm @xcite described overleaf . *
1 ) initialization * : initialize the parameters @xmath68 .
set @xmath69 * 2 ) e step * : update @xmath70 , where @xmath71 where @xmath72[\delta({\mathbf{x}}_i;\hat{{\mathbf{m}}}^{(t)},\hat{{\mathbf\sigma}}^{(t)},\hat{{\mathbf\psi}}^{(t)})+\hat{\nu}^{(t)}]},\ ] ] and @xmath73 * 3 ) first cm step * : update the parameters @xmath74 .
@xmath75where @xmath76 and @xmath77 . the update for the degrees of freedom can not be obtained in closed form .
instead we solve for @xmath22 to obtain @xmath78 .
@xmath79 where @xmath80 is the digamma function . *
4 ) second cm step * : update @xmath5 @xmath81 \end{split}\ ] ] * 5 ) third cm step * : update @xmath6 @xmath82 \end{split}\ ] ] * 6 ) check convergence * : if not converged , set @xmath83 and return to step 2 .
note that there are several options for determining convergence of this ecm algorithm . in the simulations in section 4
, a criterion based on the aitken acceleration @xcite is used .
the aitken acceleration at iteration @xmath0 is @xmath84 where @xmath85 is the ( observed ) log - likelihood at iteration @xmath0 .
the quantity in can be used to derive an asymptotic estimate ( i.e. , an estimate of the value after very many iterations ) of the log - likelihood at iteration @xmath86 , i.e. , @xmath87 ( cf . * ? ? ?
* ; * ? ? ?
as in @xcite , we stop our em algorithms when @xmath88 , provided this difference is positive .
as discussed in @xcite and @xcite for parameter estimation in the matrix variate normal case , the estimates of @xmath5 and @xmath6 are unique only up to a multiplicative constant .
indeed , if we let @xmath50 and @xmath51 , @xmath52 , the likelihood is unchanged .
however , we notice that , @xmath55 , so the estimate of the kronecker product is unique .
we conduct two simulations to illustrate the estimation of the parameters . in both simulations ,
we take 50 different datasets of size 100 , from a @xmath89 matrix skew-@xmath0 distribution .
also , in both simulations , @xmath90 and @xmath91 . in simulation 1 , we took the location and skewness matrix to be @xmath92 and @xmath93 , respectively , and @xmath94 and @xmath95 in simulation 2 , where @xmath96 in figures [ fig : lineplots1 ] and [ fig : lineplots2 ] , we show line plots of the marginals for each column ( labelled v1 , v2 , v3 , v4 ) of a typical dataset from simulations 1 and 2 , respectively .
the dashed red lines denote the means . in figure [
fig : lineplots1 ] , the skewness in columns 1 , 2 , and 4 , for simulation 1 , is very prominent when visually compared to column 3 , which has zero skewness .
the skewness is also apparent in the lineplots for simulation 2 , however , because the values of the skewness are generally less than those for simulation 1 , it is not as prominent .
the component - wise means of the parameters as well as the component wise standard deviations are given in table [ tab : results ] .
we see that the estimates of the mean matrix and skewness matrix are very close to the true value for both simulations .
moreover , we see that the estimates of @xmath5 and @xmath6 also correspond approximately to the their true values , and thus so would the kronecker product , which is not shown .
[ tab : results ]
the density of a matrix variate skew-@xmath0 distribution was derived .
this distribution can be considered as a three - way extension of the multivariate skew-@xmath0 distribution .
parameter estimation was carried out using an ecm algorithm . because the formulation of multivariate skew-@xmath0 distribution this work is based on
is a special case of the generalized hyperbolic distribution , it is reasonable to postulate an extension to a broader class of matrix variate distributions .
ongoing work considers a finite mixture of matrix variate skew-@xmath0 distributions for clustering and classification of three - way data .
domnguez - molina , j. a. , g. gonzlez - faras , r. ramos - quiroga , and a. k. gupta ( 2007 ) .
a matrix variate closed skew - normal distribution with applications to stochastic frontier analysis . _
36_(9 ) , 16911703 .
lindsay , b. g. ( 1995 ) .
mixture models : theory , geometry and applications . in _ nsf - cbms regional conference series in probability and statistics _ ,
volume 5 .
california : institute of mathematical statistics : hayward . | although there is ample work in the literature dealing with skewness in the multivariate setting , there is a relative paucity of work in the matrix variate paradigm .
such work is , for example , useful for modelling three - way data .
a matrix variate skew-@xmath0 distribution is derived based on a mean - variance matrix normal mixture .
an expectation - conditional maximization algorithm is developed for parameter estimation .
simulated data are used for illustration . +
* keywords * : matrix variate distribution ; skew-@xmath0 distribution |
after the discovery of the higgs boson @xcite , we can finally analyze a particle that seems directly related to the electroweak symmetry breaking ( ewsb ) mechanism .
thus , almost fifty years since the standard model ( sm ) higgs boson was postulated @xcite , the study of the discovered particle properties can be used for the first time as a way to access the mechanism responsible for the ewsb . in particular ,
in the present note we focus on the analysis of the higgs interactions to the rest of sm particles .
a huge variety of data has already been collected , not only from higgs searches at both tevatron and the lhc , but also from the measurements of triple gauge boson vertices ( tgv ) s at the colliders and from the low energy electroweak precision measurements . in this context
we present a framework suitable to study the nature of the observed state via the analysis of its couplings , exploiting all the existing data sets . pursuing a model independent analysis of the higgs interactions ,
the effective lagrangian approach @xcite is arguably one of the most motivated frameworks from the theoretical point of view , given the lack of observation of any other new resonance .
hence , the effective lagrangian is useful as a model independent way to parametrize the low energy deviations in the higgs interactions caused by new physics ( np ) . following this motivation we present on the first part of this note the effective lagrangian expansion based on the assumption that the observed state is introduced as a doublet of @xmath1 , where then the @xmath2 symmetry is linearly realized in the effective theory which describes the indirect np effects at lhc energies @xcite . after presenting the results from a global analysis to the different ewsb related data sets @xcite , we focus on the very interesting complementarity between higgs analyses and tgv measurements at colliders @xcite . on the second part
we compare this phenomenology with the one that is derived from abandoning the assumption of the higgs as an @xmath1 doublet .
we study in this case the higgs interactions using the non linear or chiral effective lagrangian including a dynamical higgs boson @xcite .
we observe that the interesting higgs to gauge boson coupling correlation with the tgv interactions could be lost in this case , besides higher order differences , that both could lead to phenomenological observable consequences @xcite useful to disentangle the nature of the observed state .
we start assuming that , even if there is np associated with the ewsb sector , the observed particle is an elementary state which belongs to a light electroweak doublet scalar , and consequently , that the @xmath2 symmetry is linearly realized in the effective theory .
thus we think of the sm as an effective low energy theory but we still retain the gauge group , the particle spectrum and the pattern of spontaneous symmetry breaking as valid ingredients to describe nature at energies @xmath3 , where @xmath4 is the scale associated to np .
this fixes the complete set of higher dimensional operators that need to be considered at a given order . neglecting the effects of the dimension five total lepton number violating operator , the lowest order effective operators that can be built are of dimension six : @xmath5 where the operators @xmath6 have couplings @xmath7 .
if we restrict to @xmath8 and @xmath9even dimension six operators , 29 of them are relevant to study the higgs interactions @xcite barring flavor structure and hermitian conjugations .
eight of those modify the higgs interactions to the electroweak gauge bosons , one to gluons and one affects only higgs self interactions .
additionally there are three yukawa like fermionic operators , and the remaining ones modify both the fermionic couplings to the higgs boson as well as to the gauge bosons
. finally tgv interactions of on - shell @xmath10 s are modified by four of these operators , as well as , by one operator that only involves the electroweak gauge
boson self
couplings , @xmath11 , see @xcite .
previous to the analysis of the higgs data the equations of motion can be used to eliminate redundant operators from @xmath12 .
in addition , several operators are already strongly constrained by the use of electroweak precision data ( ewpd ) , to which they contribute at the tree level . for a detailed discussion on the reduction of the number of parameters in our effective lagrangian
see @xcite . after the choice of the basis presented there , the effective lagrangian relevant for the analysis of higgs and tgv data , and the subdominant effects to ewpd
is reduced to @xmath13 with @xmath14 where @xmath15 is the higgs doublet , @xmath16 and @xmath17 is the vacuum expectation value .
@xmath18 and @xmath19 with @xmath1 ( @xmath20 ) gauge coupling @xmath21 ( @xmath22 ) and pauli matrices @xmath23 . the dimension six operators in the final basis to be studied , eq .
( [ ourleff ] ) , contribute to the interactions of the higgs boson with the sm particles , in some cases adding new lorentz structures in the different vertices , as it was discussed in detail in @xcite .
in addition , two of them , @xmath24 and @xmath25 , contribute to the tgv interactions @xmath26 and @xmath27 , whose contributions in the commmonly used parametrization of @xcite are : @xmath28 with @xmath29 and @xmath30 , and where @xmath31(@xmath32 ) is the sine ( cosine ) of the weak mixing angle .
for completeness we note that @xmath11 contributes to the tgv parametrization in @xcite as @xmath33 , although it has no relevance for the present note
. finally , @xmath34 , @xmath35 , @xmath36 , @xmath37 and @xmath38 also give subdominant ( loop ) contributions to ewpd . in order to constrain these higher dimensional deviations with respect to the standard model expected behaviour
we build a @xmath39 function based on the signal strength measurements of the higgs production and decay at both the tevatron and lhc , together with the addition of the most precise determination of tgv s at the colliders in the present framework , as well as the inclusion of ewpd in a simplified way in terms of the oblique parameters @xmath40 and @xmath41 . the details on the data included on this global analysis , as well as the statistical details of the fit are described thoroughly in @xcite . in order to study the present bounds on the higgs interactions from higgs ,
tgv and ewpd sets we perform several global analyses including different data and dimension six operator sets .
the results are shown in fig .
[ fig:1dima ] . in the figure
one can see the @xmath42 dependence on the fit parameters when we consider all higgs collider data ( red solid lines ) , higgs collider plus tgv data ( dashed purple lines ) , and higgs collider plus tgv and ewpd ( dotted blue lines ) . in the columns we show the @xmath42 dependence with respect to the fit parameter written in the bottom of the column , with @xmath43 given in tev@xmath44 , and while the rest of undisplayed parameters have been marginalized .
in the upper row we use @xmath45 , @xmath46 , @xmath47 , @xmath48 , @xmath49 , and @xmath50 as fit parameters with @xmath51 , while in the bottom row the fitting parameters are @xmath45 , @xmath52 , @xmath48 , @xmath49 , @xmath50 , @xmath53 and @xmath54 . with the precision reachable from the considered data sets the main conclusion that we can extract
is that so far all the considered higgs and gauge interactions are completely compatible with the sm higgs hypothesis .
this causes all the @xmath42 panels to have the minima close to the 0 point ( sm point ) without any statistically significant deviation with respect to the sm pattern of interactions .
the degeneracies in @xmath45 , @xmath53 and @xmath54 are due to the interference with the sm corresponding vertices .
in addition , by comparing the results for the upper and lower operator sets considered , we can observe that the introduction of the fermionic operators to the analysis has a negligible effect on @xmath34 , @xmath35 , @xmath36 , @xmath37 and @xmath38 .
an exception is the observed weakening of the exclusion ranges on @xmath45 , which is strongly related to the inclusion of @xmath53 , as discussed in @xcite .
[ fig:1dima ] is useful to illustrate some of the interesting features and the potential of the data driven approach , as it allows us to identify which of the data sets impose the strongest constraints on a given operator .
for instance , by looking at either the @xmath48 or @xmath49 panels , either the upper or the lower row , we could already foresee comparing the results from the analysis of the higgs data to the results from the combined analysis of higgs and tgv data that the precision reachable from both data sets on @xmath48 or @xmath49 is at a similar level .
a conclusion that drives us directly to the following subsection .
as we have commented , two of the operators , @xmath24 and @xmath25 , contribute both to the higgs interactions with the gauge bosons , and to tgv interactions . in the results presented in fig .
[ fig:1dima ] we have used this double contribution on the direction of adding tgv data to the global analysis of the dimension six operators in order to further constrain the coefficients of the operators considered in eq .
( [ ourleff ] ) .
conversely , we can exploit this double contribution on qualitatively the opposite direction .
we present now the global analysis of only higgs data , but this time we rotate the exclusion bounds derived from the results to bounds on the tgv parameters by using eq .
( [ eq : wwv ] ) . as we have observed in the previous subsection that the fermionic operators do not affect the electroweak ones , for simplicity we show in this subsection the results of the space spanned by @xmath55 .
we present the results of the analysis to the higgs data only in fig .
[ fig : tgv ] .
there we plot in the red filled region the 95% c.l . allowed values in the plane @xmath56 after marginalizing over the other four parameters on the higgs analysis , @xmath57 and @xmath50 , _
i.e. _ we define @xmath58 and we impose the two - dimensional 95% c.l .
allowed region from the condition @xmath59 . from the global analysis of only higgs data from the lhc and tevatron
( red filled region ) , and relevant bounds from tgv measurements in different collider experiments as labeled in the panel .
we also show the estimated constraints obtainable by combining these bounds ( hatched region).,scaledwidth=45.0% ] in the figure we also show the relevant 95% c.l .
bounds from tgv measurements in different collider experiments as properly labeled .
these experimental measurements were all performed in the framework given by eq .
( [ eq : wwv ] ) , for further details see @xcite . as we can observe in the figure , the direct measurements of tgv s and the results derived from the global analysis of higgs data translated to the tgv space , are not only complementary because of the different correlation that they present , but actually the precision reachable from both types of determinations are at a comparable level . as a consequence
, this interesting complementarity between both analysis can be used as a further test of the sm and of the linearly realized ewsb . in the future ,
when the data sets are extended , the combination of both types of analysis has the potential to furnish the strongest possible bounds on this anomalous tgv space . indeed , after performing the current combination of results , as described in @xcite , the hatched region in fig .
[ fig : tgv ] sets the strongest 95% c.l .
exclusion region of the ones currently available .
we proceed now to compare the phenomenology of the effective lagrangian expasion assuming a linear realization of the ewsb with the phenomenology of the non linear or chiral lagrangian expansion , including a light higgs , but in this case generically as a singlet of @xmath1 .
while the first expansion is suitable for models with elementary higgs particles , the second one may be more adequated for dynamical " -composite- ones .
for the non linear case we consider the construction of the effective lagrangian presented in @xcite , up to four derivatives in the expansion and including the bosonic and yukawa like @xmath8 and @xmath9 even operators and @xmath9 violating operators to be included and studied @xcite . ] . for the comparison of the phenomenology
, we can start restricting to the subset of chiral operators that include the same gauge interactions that are already introduced by the dimension
six operators in the linear expansion , _
i.e. _ we consider the chiral structures weighted by the parameter @xmath60 in @xcite .
the lesser symmetry constraints in the non linear effective lagrangian means more possible invariant operators at a given order , which is translated into phenomenological decorrelations with respect to the linear case , differences that we present here .
for instance , in the unitary gauge , two of the non linear operators , @xmath61 and @xmath62 in the notation of @xcite , read : @xmath63 with their coefficients @xmath64 and @xmath65 taking arbitrary ( model dependent ) values , and where the model dependent functions @xmath66 encode the chiral interactions of the light @xmath67 . in contrast , the @xmath68 operator containing the same gauge interactions , @xmath25 , results in the combination : @xmath69 in consequence , in a general non linear analysis the @xmath70 interactions get decorrelated from the @xmath71 and @xmath72 couplings , all encoded in @xmath25 , and for the precise lorentz structures shown above .
these interactions construct in the linear case the correlation that leads to the interesting complementarity between higgs data analysis and the direct measurement of tgv s that we have highlighted in the previous section when exploiting the effective lagrangian analysis in the linear realization of the ewsb . thus in order to study whether the ewsb mechanism presents a correlated or a decorrelated behaviour , we can perform the global analysis of higgs , tgv and ewpd in the context of the non linear effective lagrangian , what is the first 8dimensional analysis in this basis @xcite . while the complete details of such study can be found in @xcite , here we only show a subset of the two dimensional results . but instead of showing the ranges of results as a function of the coefficients of the non linear operators included in the global analysis , we rotate part of the coefficients to a new set of variables , see @xcite .
this is illustrated in fig .
[ fig : nl1 ] , where the results of the combined analysis of the data sets considered have been projected into combinations of the non linear operators @xmath61 , @xmath73 , @xmath62 and @xmath74 : @xmath75 defined such that at order @xmath68 of the linear expansion @xmath76 , @xmath77 , while @xmath78 . in the expressions @xmath79 and @xmath80
stand for the relevant coefficients of the non linear operators considered for the comparison @xcite , where in the non linear case we have included the light higgs in the generic functions in eqs .
( [ p2unitary ] ) and ( [ p4unitary ] ) , assuming @xmath81 , where the dots stand for higher powers of @xmath82 that are irrelevant for the present note , and where to further simplify the notation @xmath83 , with @xmath79 being the global chiral operator coefficients . with these variables , the @xmath84 coordinate corresponds to the sm in the upper figure of fig .
[ fig : nl1 ] , while in the lower figure it corresponds to the linear regime ( at order @xmath68 ) . would future data point to a departure from @xmath84 in the variables of the first figure it would indicate bsm physics irrespective of the linear or non - linear character of the underlying dynamics ; while such a departure in the bottom one would be consistent with a non - linear realization of ewsb . and @xmath85 , which converge to @xmath49 and @xmath48 in the linear @xmath68 limit .
the dot at @xmath84 signals the sm expectation . *
lower * : a non - linear versus linear discriminator : constraints on the combinations @xmath86 and @xmath87 , which would take zero values in the linear ( order @xmath68 ) limit ( as well as in the sm ) , indicated by the dot at @xmath84 . for both figures
the lower left panels shows the 2-dimensional allowed regions at 68% , 90% , 95% , and 99% cl after marginalization with respect to the rest of parameters spanned in the analysis , see @xcite .
the star corresponds to the best fit point of the analysis .
the upper left and lower right panels in each figure give the corresponding 1-dimensional projections over each of the two combinations considered .
_ , title="fig:",scaledwidth=41.0% ] and @xmath85 , which converge to @xmath49 and @xmath48 in the linear @xmath68 limit .
the dot at @xmath84 signals the sm expectation . *
lower * : a non - linear versus linear discriminator : constraints on the combinations @xmath86 and @xmath87 , which would take zero values in the linear ( order @xmath68 ) limit ( as well as in the sm ) , indicated by the dot at @xmath84 . for both figures
the lower left panels shows the 2-dimensional allowed regions at 68% , 90% , 95% , and 99% cl after marginalization with respect to the rest of parameters spanned in the analysis , see @xcite .
the star corresponds to the best fit point of the analysis .
the upper left and lower right panels in each figure give the corresponding 1-dimensional projections over each of the two combinations considered .
_ , title="fig:",scaledwidth=41.0% ] additional phenomenological differences between both effective lagrangian expansions were further studied in @xcite .
they involve interactions that may be generated by leading operators in one of the expansions , while they can only be generated by subleading and then suppressed operators in the alternative approach .
several examples of this type of higher order differences , as well as a realistic analysis of the lhc capability to improve the precision on some of the interactions involved can be found in @xcite .
they require a more detailed analysis of anomalous tgv s .
in this note we have presented some of the interesting features of the model independent analysis of the higgs interactions by means of an effective lagrangian expansion . in the first part we have presented a global analysis of higgs , tgv and ewpd assuming that the higgs is part of an @xmath1 , and thus that the ewsb is linearly realized . from this analysis in terms of dimension
six operators we have concluded that currently the considered higgs couplings to the sm particles as well as the studied gauge self interactions are completely compatible with the sm higgs hypothesis within the reachable precision . with the data
driven approach that we have followed we have been able to indentify a very interesting complementarity between the direct measurement at the colliders of the tgv interactions and the results obtained from a global analysis of the higgs data .
the combination of both types of analysis has the potential to lead to the strongest sensitivity on deviations from the sm pattern of interactions in the tgv s , being useful as a consequence to further test the sm and the linearly realized ewsb .
interestengly we have illustrated that this correlated pattern of interactions may disappear when we consider an alternative expansion , the non linear effective lagrangian . in this case the lesser symmetry constraints could be translated in general scenarios in the decorrelation between higgs to gauge bosons and tgv interactions .
we have presented a possible set of variables to study how the collected data allow us to disentangle the two possible expansions , and as a consequence the nature of the observed state and ewsb mechanism . + the contents of this note are based on the published works @xcite , the author thanks his collaborators in these studies .
j.g - f acknowledges support from european union network fp7 itn invisibles ( marie curie actions , pitn - ga-2011 - 289442 ) , micinn fpa2010 - 20807 , consolider - ingenio 2010 program csd-2008 - 0037 and me fpu grant ap2009 - 2546 .
s. chatrchyan _ et al . _
[ cms collaboration ] , `` observation of a new boson at a mass of 125 gev with the cms experiment at the lhc , '' phys .
b * 716 * ( 2012 ) 30 [ arxiv:1207.7235 [ hep - ex ] ] .
g. aad _ et al . _ [ atlas collaboration ] , `` observation of a new particle in the search for the standard model higgs boson with the atlas detector at the lhc , '' phys .
b * 716 * ( 2012 ) 1 [ arxiv:1207.7214 [ hep - ex ] ] .
f. englert and r. brout , phys.rev.lett .
* 13 * , 321 ( 1964 ) .
p. w. higgs , phys.rev.lett .
* 13 * , 508 ( 1964 ) .
w. higgs , phys.lett .
* 12 * , 132 ( 1964 ) .
g. guralnik , c. hagen , and t. kibble , phys.rev.lett .
* 13 * , 585 ( 1964 ) .
p. w. higgs , phys.rev .
* 145 * , 1156 ( 1966 ) .
t. kibble , phys.rev .
* 155 * , 1554 ( 1967 ) .
s. weinberg , _ physica _ * 96a * , 327 ( 1979 ) ; w. buchmuller and d. wyler , nucl .
* b268 * , 621 ( 1986 ) ; c. n. leung , s. love , and s. rao , z. phys . * c31 * , 433 ( 1986 ) ; b. grzadkowski , m. iskrzynski , m. misiak and j. rosiek , jhep * 1010 * ,
085 ( 2010 ) [ arxiv:1008.4884 [ hep - ph ] ] . a. de rujula , m. gavela , p. hernandez , and e. masso , nucl .
* b384 * , 3 ( 1992 ) .
k. hagiwara , s. ishihara , r. szalapski , and d. zeppenfeld , phys .
rev . * d48 * , 2182 ( 1993 ) ; k. hagiwara , r. szalapski , and d. zeppenfeld , phys .
* b318 * , 155 ( 1993 ) [ arxiv : hep - ph/9308347 ] ; k. hagiwara , s. matsumoto , and r. szalapski , phys .
b357 * , 411 ( 1995 ) [ arxiv : hep - ph/9505322 ] .
k. hagiwara , t. hatsukano , s. ishihara and r. szalapski , nucl .
b * 496 * , 66 ( 1997 ) [ hep - ph/9612268 ] .
m. gonzalez - garcia , int .
j. mod .
* a14 * , 3121 ( 1999 ) , arxiv : hep - ph/9902321 ; f. de campos , m. c. gonzalez - garcia and s. f. novaes , phys .
lett . * 79 * , 5210 ( 1997 ) [ hep - ph/9707511 ] ; m. c. gonzalez - garcia , s. m. lietti and s. f. novaes , phys .
d * 57 * , 7045 ( 1998 ) [ hep - ph/9711446 ] ; o. j. p. eboli , m. c. gonzalez - garcia , s. m. lietti and s. f. novaes , phys .
b * 434 * , 340 ( 1998 ) [ hep - ph/9802408 ] ; m. c. gonzalez - garcia , s. m. lietti and s. f. novaes , phys .
d * 59 * , 075008 ( 1999 ) [ hep - ph/9811373 ] ; o. j. p. eboli , m. c. gonzalez - garcia , s. m .lietti and s. f. novaes , phys .
b * 478 * , 199 ( 2000 ) [ hep - ph/0001030 ] .
g. passarino , nucl .
b * 868 * , 416 ( 2013 ) [ arxiv:1209.5538 [ hep - ph ] ] .
t. corbett , o. j. p. eboli , j. gonzalez - fraile and m. c. gonzalez - garcia , phys .
d * 86 * , 075013 ( 2012 ) [ arxiv:1207.1344 [ hep - ph ] ] . | in a model independent framework , the effects of new physics at the electroweak scale can be parametrized in terms of an effective lagrangian expansion . assuming the @xmath0 gauge symmetry is linearly realized , the expansion at the lowest order span dimension
six operators built from the observed standard model ( sm ) particles , in addition to a light scalar doublet . after a proper choice of the operator basis we present a global fit to all the updated available data related to the electroweak symmetry breaking sector : triple gauge boson vertex ( tgv ) collider measurements , electroweak precision tests and higgs searches . in this
framework modifications of the interactions of the higgs field to the electroweak gauge bosons are related to anomalous tgv s , and given the current experimental precision , we show that the analysis of the latest higgs boson data at the lhc and tevatron gives rise to strong bounds on tgv s that are complementary to those from direct tgv measurements .
interestingly , we present how this correlated pattern of deviations from the sm predictions could be different for theories based on a non linear realization of the @xmath0 symmetry , characteristic of for instance composite higgs models .
furthermore , anomalous tgv signals expected at first order in the non linear realization may appear only at higher orders of the linear one , and viceversa . their study could lead to hints on the nature of the observed boson .
higgs , effective lagrangian , gauge couplings , lhc |
through x - ray observations of nearby galaxies , a class of ultraluminous x - ray ( ulx ) sources has emerged .
these are pointlike , non - nuclear sources with observed x - ray luminosities greater than @xmath3ergs@xmath1 @xcite .
of most interest are those sources with bolometric luminosities in excess of the eddington limit for a 20 m@xmath4 black hole , or l@xmath5ergs@xmath1 .
the true nature of these sources is unclear , and this class most likely includes several different types of objects . though some of these sources are located within a few parsecs of their host galaxy s dynamical center , they do not exhibit many of the characteristics of active galactic nuclei ( agn ) . because the ratio of x - ray to optical flux is a factor of 10 greater than that of agn @xcite , these objects are fairly easy to recognize in x - ray imaging data . assuming that the eddington limit is obeyed by black hole accretion , the existence of such luminous non - agn sources presents a puzzle .
several models have been proposed to account for the high luminosities of the ulxs . among
these are relativistic and non - relativistic beaming from stellar - mass black hole systems @xcite and accretion of matter into intermediate mass black holes ( imbhs ) . in several ulx systems
( ngc 1313 x-2 , m81 x-9 , etc . ) , detection of emission nebulae surrounding the ulx supports isotropic emission from the central source @xcite , which can not be described through beaming .
further , a number of ulx ( ngc1313 x-1 , etc . )
x - ray spectra are best fit with combined multi - component blackbody ( mcd ) and power law fits , similar to galactic black holes in their high - state .
recently , @xcite find that many spectral fits of ulxs require cool accretion disk temperatures of approximately 100ev .
the theoretical relationship between black hole mass and disk temperature ( t @xmath6 ) has been observed to hold true for stellar mass ( typically around 1kev ) and supermassive ( around 10 - 100ev ) black holes @xcite . using these scaling relations
, the cool accretion disk ulxs would correspond to a population of high - state imbhs with masses of @xmath7m@xmath4 . if ulxs do not obey the eddington limit
, they could be the result of an outburst ( such as can occur in low mass x - ray binaries within our own galaxy ) .
@xcite find evidence for approximately 5 galactic black hole x - ray binaries which exhibit luminosities in the ulx range during outbursts .
these sources would appear as transient ulxs .
the typical time scale for outburst of galactic x - ray transients is a few days to rise from quiescent level with a decline from peak brightness to quiescent value of 30 - 40 days @xcite .
another possible explanation is super - eddington emission from accretion disks surrounding stellar mass black holes @xcite .
sources of this type would be expected to have soft x - ray components well modeled by hot accretion disks ( @xmath8kev ) similar to superluminal x - ray sources in the galaxy ( e.g. belloni et al . 1997 ) .
likely , ulxs include a variety of different objects with both isotropic and non - isotropic emitters .
however , if some ulxs do indeed represent a class of high - state imbhs , similar to the high / soft ( thermal dominated ) state stellar mass black holes in our galaxy , we might also expect to see the low - state objects from this same population . in galactic black hole systems ,
the low - state is generally characterized by lower luminosity , with l@xmath9l@xmath10 @xcite , and a power law photon spectrum , typically with index @xmath11 @xcite .
indeed , the existence of some ulx sources ( ic 342 x-1 , ngc 5204 x-1 ) as possible low - hard ( pure power law ) state imbhs , well - fit by simple absorbed power laws , have been noted from chandra observations by @xcite . in this study
we seek to test a direct prediction of the imbh hypothesis ; namely , whether there is a class of sources with properties consistent with what we expect of low - state imbhs .
this requires two major assumptions : ( 1 ) that the emission from ulxs is isotropic and ( 2 ) that imbhs exhibit the same states ( whose classification was based on luminosity and spectral form ) as stellar mass black holes .
our goal is to find these `` low - state '' sources , if they exist , classify the properties of both high - state and low - state ulxs , and test whether these data are consistent or inconsistent with the predictions of the imbh hypothesis .
we present the results of a detailed analysis of ulxs in nearby galaxies observed with the european space agency s _ xmm - newton _ observatory . only _ xmm - newton _ provides the count rates and bandpass necessary to distinguish different spectral models for most ulxs , accurately determine both the temperature of the thermal component expected for high - state objects , and determine whether this component is required in the spectral modeling of these objects .
since the xmm - newton x - ray spectra of ulxs are similar in quality to spectra for galactic x - ray binaries obtained in the 1980s , our spectral classification in this paper will remain purely schematic .
thus , our classifications as low and high state objects are a first approximation , based on the quality of the spectra available . in section 2 ,
we detail the observations examined from the _ xmm - newton _ archives and explain the data analysis for the individual point sources . in section 3
, we discuss the spectral fitting technique as well as simulations we conducted to determine their validity .
we discuss the implications of our results in section 4 .
the data used in this investigation were drawn from the _ xmm - newton _ public data archive . assuming that low - state ulxs exist in the luminosity range of @xmath12ergs@xmath1 , we conducted simulations to determine the optimum criteria for observations capable of resolving point sources of this luminosity .
our simulations provided a guide for choosing which of the vast number of archival xmm datasets we should examine .
this luminosity range was chosen on the assumption that an approximately 100m@xmath4 black hole would radiate at @xmath13 of the l@xmath10 in the low - state @xcite . within the luminosity range of interest ( l@xmath0ergs@xmath1 ) ,
there are a number of known objects that could be confused with ulx sources .
one type of source is supernova remnants ( snrs ) .
these sources are often easy to distinguish based on their characteristic spectrum : with poor signal to noise we expect a steep power law and as the signal to noise increases , emission lines become clearly visible .
super - eddington accreting neutron stars have been observed to have luminosities within this range for a short period of time .
neutron star x - ray binaries often have spectra well fit by a hot multi - color disk blackbody model , or with low signal to noise , by a bremsstrahlung model .
both models have similar curvature and a 0.7 - 2.0kev blackbody model is indistinguishable from the bremsstrahlung model .
we chose to use the bremsstrahlung model because it is the simpler model and gives an adequate qualitative description to the data .
we expect that for low temperature bremsstrahlung sources , the spectrum should be easily distinguishable from a power law with @xmath14 ( as is expected for a low - state object ) .
if , however , the neutron star spectrum has kt @xmath15kev , as observed for some ns x - ray binaries , our simulations show that we can not distinguish between the power law and bremsstrahlung models .
the most common sources we expected to find in this luminosity range were the analogs to galactic black hole x - ray binaries in their high - soft ( thermal dominated ) state .
these sources typically have spectra well fit by a blackbody with temperature of @xmath16kev combined with a power law with index @xmath17 .
our simulations sought to determine the number of photons required to distinguish between spectral fits corresponding to a power law model with @xmath14 and a combined blackbody and power law model .
these models qualitatively correspond to those of a low - state ( pure power law spectrum ) and high - state ( thermal dominated spectrum ) x - ray binary .
since we do not know the proper normalization between the blackbody and power law components for high - state objects ( it varies from source to source ) , we tested whether each of the components separately , e.g. blackbody or a steep power law , could be distinguished from the simulated `` low - state '' spectrum .
we chose to simulate spectra in _ xspec _ using the command _
fakeit none_. we used generic response and ancillary response matrices . simulating a power law model with a @xmath18 , we found that for 200 , 400 , and 1000 counts , these models were distinguishable at @xmath19% confidence from a blackbody source ( with kt constrained to the range of 0.6 to 1.3kev , similar to that of galactic black holes ) .
we found that for a lower number of counts the distribution in @xmath2 values increases to include a larger range of @xmath2 values ( i.e. @xmath2 = 1.3 - 2.0 compared to @xmath2 = 1.5 - 1.7 ) . simulating a power law with @xmath20
, we find the same trend .
we determined that at roughly 400 counts the distributions of @xmath2 from a @xmath18 and @xmath20 power law become entirely separable at @xmath19% confidence . in order to distinguish between the different spectral fits for objects with l@xmath21ergs@xmath1 , we select all galaxies that were observed for at least 10 ks ( with the exception of the bright ulx in ngc 5408 , which had enough photons for analysis despite the low exposure time ) with _ xmm - newton _ and that are closer than 8 mpc .
we estimated that these criteria would give us a minimum of 400 counts for objects with l@xmath22ergs@xmath1 .
we emphasize that the criteria quoted , based on the simulations , were used as a guide in choosing the sample of galaxies examined in this study .
these simulations are not used as the statistical basis for our object - by - object analysis ( discussed in section 3 ) .
our sample of galaxies is selective in that it represents objects of interest in the x - ray band .
we include details on these host galaxies in table [ tbl-1 ] .
_ xmm - newton _ spectral information of individual x - ray sources had previously been published for approximately 60% of the host galaxies .
we include references in the alternate i d column and footnotes of table [ tbl-2 ] .
we do not compare our results with these previous studies on a source by source basis .
we found that abstracts describing the proposals for _ xmm - newton _ observations were available for only 13 of the 32 galaxies examined .
of these 13 , only one observation cited the motive as a study of ulxs ( ngc 1313 ) .
however , 7 of the remaining 19 galaxies contained sources classified as ixos , intermediate x - ray objects , by @xcite .
if the remaining galaxies were not studied due to their ulx population , the effects of bias are small with roughly 25% of the sources studied explicitly due to their connection with ulx sources .
our host galaxies include only spirals and irregulars .
figure [ fig1 ] displays the distribution of galaxy type .
we reduced the data using the _ xmm - newton _ science analysis system ( sas ) version 6.0.0 . since the processed pipeline products ( pps )
were created with earlier versions of sas , the observation data files ( odf ) were used to produce calibrated photon event files for the epic - mos and pn cameras using the commands _ emchain _ and _ epchain_. following this , the events tables were filtered using the standard criteria outlined in the _ xmm abc guide_. for the mos data ( both mos1 and mos2 cameras ) , good events constitute those with a pulse height in the range of 0.2 to 12 kev and event patterns that are characterized as 0 - 12 ( single , double , triple , and quadruple pixel events ) . for the pn camera ,
only patterns of 0 - 4 ( single and double pixel events ) are kept , with the energy range for the pulse height set between 0.2 and 15 kev .
bad pixels and events too close to the edges of the ccd chips were rejected using the stringent selection expression `` flag = = 0 '' .
time filtering was applied as needed by editing the light curve produced in _
xmmselect _ for the entire observation .
flare events ( distinguished by their high count rate ) detected in all three cameras , were cut using the _ tabgtigen _ task as outlined in the _
abc guide_. typical count rate parameters for filtering were rate @xmath23 for mos detectors and rate @xmath24 for the pn detector .
such filtering was only done as needed .
pre - filtered exposure times are listed in table [ tbl-1 ] .
the number of counts from the filtered net exposure times for the individual sources are listed in table [ tbl-2 ] .
we note that the filtered data are not always sufficiently clean that a high signal - to - noise is maintained up to 10kev .
sources with a high background flux level , relative to the source spectrum , show poorer signal - to - noise in the spectrum above 1kev . before extracting spectra of the brightest sources , contour maps of the x - ray observation
were overlaid on digital sky survey ( dss ) images .
this ensured that bright foreground stars and background agn were easily distinguished , and thereby not included in our sample .
also , we checked the _ xmm - newton _ source positions with ned and simbad to determine if they coincide with any known background galaxies or qsos . a list of these bright fore - ground or background sources is included in table [ tbl-7 ] .
spectra for the bright point sources were extracted based on their apparent brightness in the ccd images .
no explicit source detection algorithm was necessary .
we used the sas task _
especget_. with this task we created spectra ( for both the source and background ) , response matrices , and ancillary response files for all three epic cameras , when possible .
the typical extraction radius was 20 arcseconds , but depending on both the size and proximity of a source to another source , the extraction radius ranged from 9 - 87 arcseconds .
background spectra were extracted either in an annulus centered on the source , or in a circle of appropriate size away from the source , depending on the proximity of the candidate source to other x - ray sources .
annuli were used for sources that were not located within a few arcseconds of another source , thus annular background extraction radii were not used for sources with small extraction radii . for sources in crowded regions , we used circular extraction radii close to the source .
we extracted backgrounds close to the source in order to correct for emission local to the ulx . once the spectra were obtained , they were rebinned to require at least 20 counts per bin , using the command _ grppha _ in lheasoft .
the list of sources , with position and count information , is included in table [ tbl-2 ] .
we only included sources for our spectral studies that had 400 or greater pn counts ( or mos for sources in ngc 253 and m81 and ngc 4945 xmm3 and ngc 2403 xmm1 , for which pn spectra were not available ) .. the extracted spectra were fit with standard models in xspec v11.3.1 .
for each source , we fit the pn and mos spectra simultaneously in the 0.3 - 10 kev range .
we allowed a free normalization constant to account for the differences in flux calibration between the three cameras ( similar to @xcite ) .
each source was first fit with an absorbed single component model . in all cases
we used the standard absorption model _
wabs _ , leaving the column density as a free parameter . for those sources where the hydrogen column density was unconstrained , we fixed the value to the galactic foreground value listed in table [ tbl-1 ] .
results of the single - component fits are seen in table [ tbl-3 ] .
we include in this table only the best - fit parameters for those sources best described by a single - component model .
the addition of a blackbody component to these single - component fits changes the @xmath25 value by a negligible amount and therefore is not statistically significant .
more specifically , the addition of a blackbody model to the power law fit corresponds to a @xmath26 , which is the @xmath27% confidence level using the f - test for two degrees of freedom .
the flux values quoted represent the unabsorbed flux in the pn spectra , in the 0.3 - 10 kev band .
all errors quoted , here and subsequently , correspond to the 90% confidence level for one degree of freedom ( @xmath28 ) . the luminosities were calculated from the unabsorbed flux using the distances quoted in table [ tbl-1 ] .
both flux and luminosity correspond to those of the best fit model ( power law or bremsstrahlung ) .
it should be noted that since our selection criteria was based on a count rate cutoff , due to the variety of spectral forms , the inferred luminosity cutoff will not be uniform . in table
[ tbl-3 ] we denote the single component model we choose as the better fit in bold .
this notation also indicates the model ( power law or bremsstrahlung ) used to compute the quoted flux . for @xmath29% of the power law sources ,
the @xmath25 difference ( @xmath30 ) between the power law and bremsstrahlung models is only marginally different . of these sources ,
the average kt value for the bremsstrahlung fit is 5.54kev . from our simulations
we find that at high temperatures the bremsstrahlung fit becomes indistinguishable from a simple power law .
thus , given the high temperatures of the bremsstrahlung fits for these sources , they are equally well described by the power law model .
typical kt bremsstrahlung values for accreting neutron stars are from 3.0 to 7.0kev @xcite . from these single - component sources ,
we classify 16 sources as low - state ulx sources .
this classification is based on ( 1 ) the shape of the spectrum , well - fit by an absorbed power law , ( 2 ) the luminosity of the sources ( they needed to be luminous enough to be included in our sample , l@xmath0ergs@xmath1 ) , and ( 3 ) the x - ray location of the object within the optical galaxy ( based on digital sky survey ( dss ) images ) .
the third criterion was important in limiting the effects of contamination from fore - ground and background sources within our sample of ulx sources .
thus , we overlaid x - ray contours from the xmm image on the dss images , determining the location of the x - ray source as within or outside of the optical extent of the galaxy .
we note that for two of the sources classified as low - state ulxs ( holmberg i xmm2 and ngc 2403 xmm4 ) , there is uncertainty of whether the x - ray source is in fact within the optical galaxy due to the quality of the optical dss images .
the images used to determine the third criterion are available online ( http://www.astro.umd.edu/$\sim$lwinter/second3.html ) .
we further discuss these criteria in section 4 .
we state that the first criterion for classification as a low / hard state ulx is a spectrum well - fit by a power law .
of the 30 sources in table 2 , three sources were clearly not well - fit by either the power law or bremsstrahlung model .
seven of the remaining 27 sources were clearly not within the optical extent of the host galaxy .
of the remaining sources excluded from classification as a low - hard state object ic 2574 xmm1 was better fit with a bremsstrahlung model ( with @xmath31 ) .
in an additional observation of the source ngc 4258 xmm2 , a simple power law model is not an adequate fit to the data ( while the luminosity of the source is not within the ulx " classification range mentioned in the introduction ) . for the remaining sources ,
ngc 247 xmm2 and ngc 253 xmm2 , there was sufficient doubt on the spectral form where the bremsstrahlung and power law model as well as the addition of a thermal component all yielded adequate fits to the spectra .
therefore , we excluded these sources from a classification , noting the ambiguity of the model fits for these sources . for those sources we classify as ulxs , we include computed bolometric luminosities in table [ tbl-11 ] .
to compute the bolometric luminosities for these ulx sources , we used the exponentially cutoff power law spectrum of @xcite , model _ cutoffpl _ in _ xspec _ , with a cutoff energy of 10kev . from observational studies of galactic x - ray binaries , it has been observed that low - state objects have spectra that cut - off at high energies ( @xmath32kev ) @xcite .
thus we chose the exponential model _
cutoffpl _ over a simple power law .
this also minimizes the total luminosity for flat power law sources .
we computed an unabsorbed bolometric flux in the @xmath33kev range through use of the _ dummyresp _ command ( which extends the model beyond the observation s energy range ) .
the luminosity was then computed using the distances listed in table [ tbl-1 ] .
we quote these values as l@xmath34 ( the luminosity obtained from extrapolating the power law portion of the spectrum as an exponentially cut - off power law ) in table [ tbl-11 ] .
we note that these values represent an upper limit on the bolometric luminosity for steep power law ( @xmath35 ) objects , since we would expect the power law component to cutoff at some low energy .
however , for shallow spectrum ( @xmath36 ) sources l@xmath34 is a lower limit .
this is because , schematically , a steep power law diverges at low energies while a shallow power law diverges at high energies . for a number of sources
, we found that an improvement in reduced @xmath25 was achieved through fitting their spectra with an absorbed two - component blackbody and power law model .
we chose a simple blackbody model over the multi - component disk model , _ diskbb _
, for purely schematic reasons .
namely , observations of galactic x - ray binary systems were fit with blackbody models in the 1980s , when the signal - to - noise of these objects was comparable to that for our _ xmm _ data for ulx sources .
we also note that the _ diskbb _ model does not give an entirely accurate physical description of the data as it neglects the effects of general relativity . as a schematic model , the blackbody model is simpler than _ diskbb _ , with the same number of degrees of freedom .
in addition , for low temperatures both models yield virtually identical temperatures . for this study
, we chose the simpler model .
we defer to a further paper a discussion of the different models for the thermal component . in table
[ tbl-4 ] we present the results for the sources which are fit significantly better by the two - component model , these are sources where the improvement in @xmath25 is greater than 8 ( determined from our simulations in appendix a ) .
we include in table [ tbl-4 ] the improvement in @xmath25 of the two - component fit over the simple power law .
we include the power law best fits to these sources in the appendix for comparison with other analyses . in order to determine whether the blackbody component is statistically significant for the sources fit with a two - component model
, we simulated spectra based on accurate modeling of some of the brightest sources : ngc 247 xmm1 , ngc 5408 xmm1 , and holmberg ii xmm1 .
these sources span the observed range of the ratios of the blackbody to power law component and thus represent those from our sample with a weak blackbody relative to the power law component , intermediate case , and a strong blackbody , respectively .
our simulations are described in full in appendix a. we found that , using a @xmath37 criterion , which corresponds to the 99% significance level as according to the f - test for the addition of two extra parameters , we can readily detect the strong and intermediate thermal components in all spectra with more than 400 counts .
the weak thermal emission can not be detected in 400 count spectra , but is readily detected in 2000 count spectra .
this gives us confidence that our results are statistically meaningful .
of the sources in table [ tbl-4 ] , we classified high - state ( or thermal dominated ) ulx sources based on three criteria : ( 1 ) spectra characterized by an absorbed power law and blackbody model , ( 2 ) luminosity , and ( 3 ) x - ray source within the optical extent of the host galaxy .
the luminosity criteria required that these sources have unabsorbed luminosities l@xmath38ergs@xmath1 ( we used l@xmath39ergs@xmath1 as our hard cutoff ) .
if the sources are radiating at the eddington luminosity , this cutoff luminosity corresponds to objects with masses greater than 20m@xmath4 . from table [ tbl-4 ] ,
27 observations are recorded with l@xmath40ergs@xmath1 .
the addition of a thermal component to these sources is statistically significant over a pure power law model .
of the 27 observations , 3 correspond to multiple observations of a single source . from an analysis of the dss images ,
all 24 of these sources are within the optical extent of their host galaxies .
however , m51 xmm5 appears to be coincident with the center of its host , a dwarf companion galaxy to m51 .
the location , coupled with the high luminosity ( l@xmath41ergs@xmath1 ) leads us to classify this source as an agn .
we also excluded two sources ( ngc 1313 xmm2 and m81 xmm2 ) from our sample of high / soft state ulxs due to their previous identification as supernovae .
of the remaining 21 sources , ngc 253 xmm4 had a luminosity of @xmath42ergs@xmath1 in one observation and @xmath43ergs@xmath1 in a second .
this significant change in luminosity , with one observation below our luminosity cutoff and another a factor of @xmath44 higher than the other , led us to exclude this source as a high / soft state ulx .
it is likely that this source is a stellar mass x - ray binary within its host galaxy , where one of the observations captured the source in an outburst . in table
[ tbl-13 ] , we list sources that have @xmath45 values less than 8 for a single observation .
most of these sources have weak blackbody normalizations compared to the power law normalization .
we classify these sources as being well - fit by a two - component model while acknowledging the uncertainty in the fit as determined by the simulations .
the addition of the thermal component is not significant enough for these sources to be classified with certainty in either table 2 or 3 .
the simple power law fits for these sources are included with those for sources in table [ tbl-4 ] in the appendix .
we note that due to their high luminosity we included six of these sources ( ngc 4490 xmm2 , ngc 4490 xmm3 , ngc 4736 xmm1 , m51 xmm2 , m51 xmm6 , and m101 xmm3 ) with uncertain fit parameters as ulx high - state sources .
two of these sources ( ngc 4490 xmm3 and m51
xmm6 ) had unabsorbed luminosities @xmath46 times the @xmath47ergs@xmath1 cutoff used for high - state ulx classification .
the other four sources had luminosities above the threshold , as well as weak blackbody components compared to the power law ( see the appendix for simulations ) .
we used these points to justify including these sources with the table [ tbl-4 ] sources in the following discussions with the proviso that their spectral fits do not indicate absolutely the necessity of the additional thermal component .
for this reason , we denote these sources with a special symbol ( a circle ) in subsequent figures while including them as `` high - state '' ulx objects . for our ulx sources modeled by a combined blackbody and exponentially cutoff power law
, we computed bolometric luminosities using two methods .
the first method is recorded as l@xmath34 in table [ tbl-11 ] .
we computed the flux from 0.1 - 100kev using an unabsorbed blackbody and exponentially cutoff power law model using the xspec command _
dummyresp_. for the second method , recorded as l@xmath48 in table [ tbl-11 ] , we estimate a more accurate bolometric luminosity calculated from the flux in the range of @xmath49kt-100kev where kt is the blackbody temperature obtained from the model . in galactic x - ray binary systems ,
the power law component of the x - ray spectrum is believed to be from comptonization in a corona .
the photons supplying this energy originate from the blackbody continuum emanating from the accretion disk .
thus , a natural cutoff for this power law component occurs at the peak emission of the blackbody ( which is approximately @xmath50kt ) .
the estimated values ( obtained from cutting off the combined unabsorbed blackbody and cutoff power law model at the value @xmath49kt ) differ with regard to the full estimate ( flux from the fully integrated blackbody added to the separate flux from the cutoff power law from @xmath50kt to 100kev ) depending on the normalization factors used ( for both the blackbody temperature and the spectral index @xmath2 ) . choosing three sources displaying a range of blackbody to power law strength ( holmberg ii xmm1 , ngc 253 xmm1 , and ic 0342 xmm3 ) we found that the estimated values were within 88.3 , 95.1 , and 96.8% of the more complete estimation .
given their close proximity ( within approximately 90% ) we quote these estimated values as a good approxmiation to the bolometric luminosity .
we note that our bolometric luminosities for all of the classified ulx sources , on average , are a factor of 1.08 greater than the x - ray luminosities in the @xmath51kev band for the objects best fit by a combined blackbody and power law .
thus , to good approximation , the x - ray luminosity is the bolometric luminosity .
however , for the objects best fit by a simple power law ( low - state ulx sources ) , the average bolometric luminosity is roughly a factor of 7 greater than the x - ray luminosity in our band .
this average is dominated by the steep power law objects , in particular holmberg ii xmm1 ( @xmath52 ) . excluding this object , we get an average bolometric luminosity that is 2.8 times the x - ray flux and more indicative of the general properties of these power law - fit objects . in addition
, in this large sample of point sources , we came across a number of objects whose spectra were not well fit by the models we employed .
these sources have luminosities exceeding l@xmath53ergs@xmath1 , if they are associated with the host galaxy , and are placed in tables 2 and 3 as well as appendix b. these sources include two supersoft sources , one possible agn , and three sources well fit with additional absorption models ( including a partial covering model and a model of hot gas ) .
we briefly describe these sources in the appendix .
we have determined best - fit spectral parameters of the bright x - ray sources in 32 nearby galaxies . in choosing three `` standard '' models for our study
, we hoped to accurately separate high and low state ulxs from other types of luminous x - ray sources .
we specifically chose to fit the data with the bremsstrahlung model in order to identify neutron star x - ray binaries within our sample .
the models we used are purely schematic , and they do not physically explain the phenomena occurring , but are standard and qualitatively simple models often used to fit the spectra of galactic x - ray binaries .
we cross - referenced the x - ray positions of our sources with both ned and simbad in order to identify known supernovae , galaxies , and stars . in addition , we examined the dss optical images to place the position of our sources within their respective galaxies .
such analysis aimed to minimize contamination of our sample of ulxs with bright background and foreground sources .
further , we examined _ xmm - newton s _ optical monitor data in the visual bands ( u , b , v ) .
the _ xmm _
pps contain point source detection files for the om data .
we overlaid these point source detections with x - ray contour maps in order to determine the * brightest possible * optical count rates for the x - ray sources , which were then converted into fluxes using the om calibration documentation . in figure
[ fig7 ] , we plot the distribution of the logarithm of the x - ray to optical flux for the brightest possible optical counterpart inside the _ xmm - newton _ error circle
. only 13 of the 32 host galaxies had visible band om data during the observations .
of these 13 galaxies , 40 of the x - ray sources were in the range of the om data and only 14 were coincident with an optical point source .
therefore , the majority of our sources have x - ray / optical flux ratios that are * larger * than those displayed .
figure [ fig7 ] illustrates the lowest possible x - ray / optical flux ratios and also , by the sparsity of sources included in the diagram , it illustrates the fact that a majority of the sources have no obvious optical counterpart and thus have very large x - ray / optical flux ratios .
we estimate the point source detection limit of the om u filter as approximately @xmath54erg@xmath55s@xmath1 . for an unabsorbed x - ray flux of @xmath56erg@xmath55s@xmath1 , typical of objects with l@xmath57ergs@xmath1 located at a distance of 8mpc ,
this corresponds to @xmath58 .
therefore , the average value for our sources should fall around 2 or greater . the average distribution for qsos and agn centers around 0 and 0.8 for bl lacs @xcite .
our objects have ratios of l@xmath59l@xmath60 at least 10 times higher than those of agn and 100 times greater than stars .
recently , @xcite identify six ulxs from the catalog of @xcite as qsos .
they hypothesize that a large number of ulxs may in fact be quasars at higher redshift than their supposed host galaxy .
however , unlike the objects studied in @xcite , our ulx sources are all spatially coincident with the optical host galaxy .
in addition , a majority of our ulxs are not in the proximity of a noticeable optical point source
. the x - ray / optical flux ratios of our sources are much larger , on average , than might be expected for a qso .
it is also worth noting that while some cataloged ulxs may be qsos , optical identifications have been made associating other ulxs with a type b supergiant companion @xcite .
the spectral fits indicate that to high statistical probability ( see appendix a ) we can distinguish a class of low - state ulxs from the high - state objects .
this is assuming , as indicated in the introduction , that ulxs are isotropic emitters with luminosity and spectral form similar to galactic stellar - mass x - ray binaries . in section 3 , we stated that our ulx classification depends upon three criteria : ( 1 ) spectral form , ( 2 ) luminosity , and ( 3 ) location of the x - ray source within the optical host galaxy ( as determined from the dss images ) .
we have chosen simple , parametric ` non - physical ' models for the spectra because the signal to noise of most of the observations does not allow anything else to be constrained .
of the sources in table [ tbl-3 ] , 16 are classified as `` low - state '' objects or low / hard state ulxs , having unabsorbed luminosities @xmath61ergs@xmath1 and spectra that are well fit by power law models . throughout this paper
, we use the term low - state ulxs to include `` low - state imbh candidates '' ( sources with l@xmath62ergs@xmath1 and spectra well - fit by a simple absorbed power law ) and low - state sources with luminosities that clearly classify them as ulx sources ( l@xmath63ergs@xmath1 ) .
these low - state ulx sources are listed in table [ tbl-11 ] . in the spectral fitting section
, we noted that a power law and high temperature bremsstrahlung model are indistinguishable .
therefore , it is important to consider the luminosity of these sources in the claim that they are not neutron star x - ray binaries accreting at the eddington luminosity . of the low - state ulx sources ,
only two of the 16 sources have bolometric luminosities below the eddington luminosity of a 3m@xmath4 object ( @xmath64ergs@xmath1 ) , corresponding to the maximum mass of a neutron star .
all of the sources have values exceeding the eddington limit for a 2m@xmath4 neutron star .
further , 26 sources have unabsorbed l@xmath38ergs@xmath1 , corresponding to l @xmath65 l@xmath10 at m @xmath66 m@xmath4 as expected for `` high - state '' imbhs , and spectra that are well fit by combined blackbody and power law models . these are `` high - state '' objects .
the spectral fits for these sources are listed in tables 3 and 4 . in a statistical sense , we find that the greater the number of counts in the observation the greater our confidence in the thermal component contributing to a better fitting model
. we explain our confidence levels obtained from spectral simulations in the appendix .
in addition to these high and low state ulxs , we find a large number of sources best fit by a combined blackbody and power law model but below our threshold of l@xmath67ergs@xmath1 for a high - state ulx ( listed in table 3 ) .
many of these sources may be accreting stellar mass black holes with m@xmath24m@xmath4 .
some of these non-``ulx '' sources were found away from the optical extent of the targeted galaxy ( from our analysis of the dss images ) , and therefore may be background agn . for galactic black hole x - ray binaries , spectral indices of low - state ( or power law dominated ) objects
are typically lower than those of high - state objects , with @xmath68 and @xmath69 @xcite . in figure
[ fig2 ] , we plot the distribution of the spectral index for both high - state and low - state objects .
the spectral index for the high - state objects is the value of @xmath2 from the two - component fit . as in the galactic sources , it is clear that the spectral indices of the high - state objects are indeed larger . of further interest
, the distribution of spectral index for low - state objects looks remarkably similar to the distribution of spectral index for moderate luminosity quasars , many of which are thought to be the analogs of low - state black holes @xcite .
this supports the classification of these objects as accreting black holes . for the high - state objects ,
we find mean values of @xmath70 , with a root mean square ( rms ) deviation of s@xmath71 , and l@xmath72ergs@xmath1 , @xmath73 .
this calculation excludes the 3 objects with spectral indices greater than 3.5 . for the low - state objects ,
we find mean values of @xmath74 , with a rms deviation of s@xmath75 , and l@xmath76ergs@xmath1 , @xmath77 .
this value of @xmath78 is softer than the typical hard - state value of @xmath79 , but within the @xmath80 range used to classify this state for galactic x - ray binaries @xcite .
computing a kolmogorov - smirnov two - sample test , separating the sources into the category of low - state or high - state , we find a likelihood of approximately 0.03 that the spectral indices belong to the same distribution .
the low hard x - ray state of x - ray binaries is associated with a low accretion rate from the companion object with l@xmath81l@xmath10 @xcite . therefore , on average , we expect the luminosities of the low - state objects to be lower than the high - state objects .
figure [ fig3 ] displays the luminosity of the objects as a function of the spectral index . on average ,
the highest luminosity low - state objects have luminosities lower than those of the high - state objects .
the lower l@xmath82 values of the low - state objects imply that they may indeed be accreting at a lower rate than the high - state objects .
this can further be seen in the bolometric luminosities listed in table [ tbl-11 ] .
if these objects are accreting at a rate similar to galactic low / hard state black holes ( @xmath83 ) @xcite , we can estimate their masses as @xmath84 with @xmath85 as the eddington luminosity for a 1m@xmath4 object ( @xmath86ergs@xmath1 ) .
our mass estimations , based upon our limits to the bolometric luminosities , yield masses of @xmath87 ( see table [ tbl-11 ] ) , precisely what we might expect for a population of imbhs . if the high - state ( thermal dominated ) ulxs represent a class of intermediate mass black hole systems , their x - ray spectra should be well described by a combined blackbody and power law model . scaling for the mass of the black hole
, we would expect a relationship of t@xmath88m@xmath89 between black hole mass and blackbody temperature @xcite .
this would indicate a thermal component of @xmath90 100ev for masses of @xmath91m@xmath4 .
a few objects have been reported to display this property @xcite . in figure [ fig4 ] , we graph the distribution of the thermal component for our classified high - state objects .
we find that there are two peaks in the distribution among the thermal component , one at approximately 100 ev and another centered close to 1 kev .
this could indicate two different classes among the high - state objects .
it is possible that those objects with blackbody components near 100ev are indeed high - state intermediate mass black holes .
we note that the soft excess in pg quasars has also been modeled as a blackbody with kt@xmath92ev , but it has been suggested that this could be the result of a process not directly related to black hole accretion ( such as the presence of a warm absorber : @xcite ) .
another possible explanation is that the soft component is the result of ionized reflection from the disk @xcite . while the possibility exists that the `` thermal '' component of these 100ev sources is not directly related to black hole accretion or is related in a `` non - thermal '' ( i.e. ionized reflection ) sense , as may be the case with the soft excess in pg quasars , we assume that the soft component for the objects we classify as high - state ulxs originates from a thermal disk .
we use this assumption to test the imbh hypothesis , thus speculation on the nature of the soft component is beyond the scope of this paper .
the second peak , centered around 1 kev , has a temperature reminiscent of the galactic black hole systems in our own galaxy .
these systems may thus be stellar - mass black holes accreting matter near the eddington limit .
if this were the case , we would expect the luminosities of the sources exhibiting a higher blackbody temperature to be lower than those with cooler blackbody components . in the second graph of figure [ fig4 ]
, we plot the relationship between blackbody temperature and l@xmath93 in the 0.3 - 10 kev band .
once again , two groups are seen in the distribution of high - state ulxs .
the most luminous objects are those with low blackbody temperatures . on average ,
the less luminous sources exhibit higher blackbody temperatures .
for the sources with l@xmath94ergs@xmath1 , the mean blackbody temperature is 0.31 while the sources below this luminosity threshold have a mean blackbody temperature of 0.61 .
the second , low - luminosity , group in the distribution of high - state ulxs is clearly distinguishable in both plots of figure [ fig4 ] .
we found that , with the exceptions of ngc 253 xmm1 , m81 xmm1 , and ngc 5204 xmm1 , the spectra of these objects could also be well - described by an absorbed comptonization ( _ compst _ ) model @xcite used to fit galactic black holes in the `` very high '' state when they are radiating at the eddington limit .
this model simulates compton scattering of cool photons on the hot electrons of a completely ionized plasma .
we present the best - fit parameters for the comptonization model in table [ tbl-9 ] .
this `` very high '' state has been observed @xcite in a few galactic black holes .
yet another rubric for the very high state emerged in @xcite and @xcite , where they identify this as the `` anomalous '' state , a state whose spectrum can be well fit by a comptonized scattering model .
regardless of the name , our best - fit comptonization sources likely fit into this category .
the luminosities of these sources suggest that they are stellar mass black hole systems in this anomalous / very high state . as with the low - state ,
we include mass estimates for our high - state objects in table [ tbl-11 ] .
we assume that the high - state objects are radiating at l@xmath10 resulting in a minimum mass if there is no beaming .
we find masses of @xmath95m@xmath4 , consistent with `` normal '' stellar mass bhs , for the sources well fit by the comptonization model .
the other high - state ulxs masses range from @xmath96m@xmath4 based on eddington rates , analogous to the low - state ulx masses computed .
it is important to note that the initial simulations ( appendix ) and discussions in the spectral simulations section need to be considered in relation to the impact they pose to our classification scheme and the results presented in these sections .
while it is indeed possible that some of the objects with a weak blackbody component and a relatively small number of counts would be mis - categorized as a pure power law spectrum , one can ask what such a possible situation would do to the correlations that we have seen .
these putative objects , by assumption would have lower luminosities , however their temperatures are unknown and it is entirely unclear if they would destroy the kt / l(x ) correlation . as we have shown in our simulations it is unlikely that the fitted power law index would change and thus the presence of a low state as indicated by the spectral index would not change .
this would create a new type of object , one with a flat power law and a black body component , which is not seen in the milky way , nor among the high signal to noise objects .
in addition to the existence of ulxs with low blackbody temperatures , the temperature distribution of the high - state ( thermal dominated ) ulxs ( figure [ fig4 ] , left panel ) displays a `` gap '' which is of particular interest there is a complete absence of objects with temperatures in the range 0.26kev to 0.50kev .
it is tempting to take this as evidence for a gap in the mass distribution of these accreting black holes .
since , for a given luminosity , we expect the temperature to vary as @xmath97 , this factor of two gap in the temperature distribution translates into a factor of four gap in the black hole mass distribution .
if this result is borne out by further study , it provides an important clue to the origin and evolution of intermediate mass black holes .
one popular idea is that intermediate mass black holes formed from the collapse of massive population iii stars @xcite .
models suggest that pop iii stars with zero age main sequence ( zams ) masses in the range 25140@xmath98 and above 260@xmath98 collapse to produce black holes @xcite whereas in the range of zams masses 140260@xmath98 , pair - instability supernovae lead to the complete disruption of the stars ( i.e. , no remnant black hole remains ) .
hence , this model for imbh formation predicts a gap in the imbh initial mass function in the range of approximately 60200@xmath98 ( although this is uncertain on the low end due to the effect of the pulsational pair - instability on the pre - collapse core ) .
one possibility is that the gap in our observed temperature distribution ( and hence the inferred gap in the mass function ) is due to this effect of the pair instability supernovae in pop iii stars .
this would require that the current imbh mass function is approximately the same as the initial imbh mass function .
in other words , it requires that most imbhs ( especially those just below the gap ) have not grown significantly due to accretion since their formation and , hence , that the ulx phase represents a short fraction of the life - time of an imbh ( @xmath99 , where @xmath100 is the e - folding timescale for eddington limited black hole growth with radiative efficiency @xmath101 ) .
an alternative interpretation of the inferred mass gap is to suppose that two fundamentally different modes of formation lead to a strong bi - modality in the final black hole mass function .
black hole masses below the gap can be readily understood through normal stellar processes .
a separate and distinct population of significantly more massive black holes may result from dynamical processes in the core of dense globular clusters @xcite .
supposing that the galaxy s bright x - ray population is representative of low - redshift galaxies , we expected to find a number of sources similar to galactic x - ray binaries in our sample . in our sample
, we find approximately 24 sources with luminosities below our high - state ulx cutoff ( @xmath102ergs@xmath1 ) , x - ray positions within the optical extent of their host galaxy , and no obvious optical counterpart .
the unabsorbed luminosities for these sources range from @xmath103ergs@xmath1 ( @xmath51kev band ) .
two of these sources were transients in the _ xmm _ data .
of the four host galaxies with multiple observations examined , two of these galaxies contained solely ulx sources in our luminosity regime ( holmberg ii and ngc 5204 ) .
each of the remaining two ( ngc 253 and ngc 4258 ) had a transient source best fit by a combined blackbody and a power law .
this suggests an interesting diagnostic in terms of distinguishing our ulx sources from a normal hmxb population . in our own galaxy ,
most hmxbs vary on timescales of days or less and most of the black holes in the milky way are transients , though some hmxbs are indeed persistent . the figures in @xcite , determined through detailed mass - transfer calculations , indicate that transient behavior should not be expected from a population of imbhs .
thus , on average , our ulx sources should remain x - ray bright in multiple observations . through a literature search , we found that 37/42 of our ulx sources are well detected in rosat observations and thus are luminous for greater than 10 years and therefore are not transients .
examination of the long term light curves show that most of these sources vary by less than a factor of 3 over the timescale from rosat to xmm .
the sources that have been above the eddington limit in the milky way and the magellenic clouds do so transiently , for a small fraction of the time .
as best as we can tell , from the light curves from einstein , rosat , asca , chandra and xmm the ulxs are , rarely , transients , and are almost always ` on ' , unlike galactic `` ultra - luminous '' objects . as a possible further diagnostic , we constructed a color - color diagram for our ulx sources .
we adopted the colors of @xcite in order to compare our sample with their sample of galactic x - ray sources .
thus , our colors were constructed from unabsorbed model fluxes in four energy bands : 3 - 4 , 4 - 6.4 , 6.4 - 9.7 , and 9.7 - 16 kev .
the _ xspec _ command _ dummyresp _ was used to calculate a flux based on the model for the 10 - 16 kev range .
we plot colors for a pure unabsorbed power law ( from @xmath104 ) and an unabsorbed mcd model ( _ diskbb _ in _ xspec _ with kt@xmath105ev ) for comparison . comparing our figure [ fig5 ] with figure 8 of @xcite
, we find that our ulx sources largely lie along the same regions as their black hole sources .
a few ulx sources , however , lie in the region occupied by atoll and z - sources in the plot of @xcite .
these sources were those best fit by a comptonization model . in this section ,
we examine the environment in which ulx sources reside .
we investigate the claim that the ulx population is proportional to the host galaxy s star formation rate ( sfr ) ( ranalli et al .
2002 ; grimm et al .
2003 ) . towards this end
, we use the far - infrared luminosity of the host galaxy as an indicator of the sfr . in order to compare the ulx population of a galaxy with the sfr we followed a similar approach to @xcite .
we calculate the fir flux from observations taken by the _ infrared astronomical satellite_. as in @xcite , the flux between 42.4 and 122.5 @xmath106 m is approximated as 1.26@xmath107s@xmath108s@xmath109erg@xmath55s@xmath1 .
the values of the flux at 60@xmath106 m ( s@xmath110 ) and 100@xmath106 m ( s@xmath111 ) were obtained from either @xcite or ned .
luminosities were calculated using the distances quoted in table [ tbl-1 ] .
we list these values in addition to the number of ulxs observed in individual galaxies in table [ tbl-10 ] .
the number of ulxs includes both the objects we classify as high and low state ulx as well as those sources resolved by chandra . in figure [ fig6 ]
, we show two plots relating the number of ulxs to l@xmath112 .
it has been suggested by @xcite that the luminosity function in the x - ray regime from hmxbs is related to sfr . in our first plot
, we find that the galaxies with the highest l@xmath112 seem to have fewer ulxs than may be expected from the luminosity functions of @xcite , who present a relationship showing a scaling of the number of hmxbs with luminosities over a set threshold with the host galaxy s sfr ( see equation 7 ; grimm et al .
2002 ) . using this relationship
, we would expect that a galaxy with a sfr approximately equal to that of m51 ( @xmath113m@xmath4yr@xmath1 from their table 1 ) to have @xmath114 objects with luminosities greater than @xmath3ergs@xmath1 .
we find 5 objects with this luminosity in m51 , consistent with their result .
however , for ngc 4945 , a galaxy with approximately the same l@xmath112 and therefore sfr , we find only one source with a luminosity in this range .
however , we note that ngc 4945 is a seyfert , implying that the l@xmath112 may primarily be caused by the agn and not a direct indication of sfr .
in addition to high l@xmath112 sources with few ulx we find a number of sources with very small sfr but which contain a ulx . for sources with sfr @xmath115m@xmath4yr@xmath1 , which corresponds roughly to sources with l@xmath112 less than that of ngc 4736 , we would expect @xmath116 sources with luminosities above @xmath3ergs@xmath1 . however , there are a number of bright ulxs in galaxies with very low sfrs ( for instance holmberg ii , holmberg ix , ngc 5204 , and ngc 5408 ) .
thus , in a direct comparison , our results do not agree with the predictions of @xcite .
the second plot displays the average number of ulxs / galaxy , binned according to luminosity .
this plot is extremely similar to fig .
15 of @xcite for spiral galaxies .
thus , once again , it seems that the connection between sfr and the ulx population in spirals is supported . for irregular galaxies
, however , there seems to be more of a spread in the distribution .
this could be the result of poor sampling most of the bins contain only one galaxy .
another possibility is that there is no direct correlation in irregular galaxies or that the overall star formation in these galaxies is less ordered or clumpier .
if the latter is the case , the overall sfr of the galaxy is only an average over a wide range of values .
we shall address this issue again in the next paper in this series ( l.m .
winter et al . , submitted ) where we discuss the local environments of the ulxs in our sample . in figure [ fig8 ]
we plot the distribution of column densities among the ulxs .
we subtracted the galactic column density towards the galaxy ( obtained from the nh ftool and listed in table [ tbl-1 ] ) from the values obtained through spectral fits .
we note that , on average , the ulxs have large column densities .
the typical galactic column density along a line of sight is @xmath117@xmath55 .
if the ulx is located on the opposite side of its host galaxy , we might expect maximum column densities of @xmath118@xmath55 .
however , most of our sources have column densities well above this value .
this is in agreement with the analysis of 5 ulxs by @xcite and may imply , as they suggest , that the local environment of the ulxs contains an extra source of absorption .
we are investigating this further , comparing the x - ray absorption column densities with hi data ( l.m .
winter et al . , submitted ) . in order to better understand the relationship between sfr and the ulx population , it is necessary to extend ulx studies to other wavelengths . in particular , it becomes important to analyze uv and ir images close to the ulx .
we present the results of an _ xmm _ survey of the ulx population in nearby galaxies . in this study
, we assumed that ulxs are isotropic emitters .
for our selected ulx sources ( which excluded transient sources and supernovae ) , this assumption was supported by the finding that 37/42 of our ulxs were found to be ` on ' in rosat observations .
this implies that theses sources exhibited high luminosities for time scales of at least 10years , a property that is not seen in galactic eddington - limit exceeding sources ( such as black hole x - ray binaries undergoing an outburst ) .
we also assumed that if some ulx sources represent a class of imbh x - ray binaries , they would exhibit spectral states analogous to galactic stellar mass black hole x - ray binaries .
this is the hypothesis we set out to test , classifying a source as a ulx based on ( 1 ) spectral form , ( 2 ) luminosity , and ( 3 ) coincidence of the x - ray source within the optical host galaxy . due to the quality of spectra available for these distant x - ray sources ,
our classification of spectral form is really a first approximation describing the basic curvature of the spectrum . through this study
, we have found that there exists a population of objects whose x - ray spectral properties closely match the low / hard state spectra of galactic black holes , but whose luminosities lie in the range of l@xmath119ergs@xmath1 . in the milky way , black holes with these spectral properties
radiate at only @xmath120 of the eddington limit .
if this is also true for this population , it indirectly implies that these objects have a mass greater than @xmath121m@xmath4 ranging up to 1500m@xmath4 and thus should be imbhs .
the existence of such objects was `` predicted '' on the basis that the ulxs previously studied shared the x - ray spectral characteristics of high - state galactic black holes ; namely , an x - ray spectrum best fit by a combined blackbody and a power law @xcite , but with much higher luminosities .
if these objects are high - state imbhs , the corresponding low - state objects should also exist .
our survey has also uncovered a large population of objects whose x - ray spectra are well modeled by the canonical description of galactic black holes in the high - state ( thermal dominated ) , a black hole with a steep power law , but whose bolometric luminosities exceed @xmath122ergs@xmath1 , ranging up to @xmath123ergs@xmath1 and whose blackbody temperatures are less than 0.3kev . if these objects are radiating at @xmath124 the eddington limit like their milky way counterparts their implied masses are from @xmath125m@xmath4 , a range very similar to that implied by the low - state objects . using the m@xmath89 scaling of mass to temperature ,
the observed spectral temperatures give masses of @xmath126m@xmath4 a considerably larger value .
in general agreement with the expectations of the imbh hypothesis , the objects with high - state spectra are more luminous than those with low - state spectra .
we note that these results have required the high signal to noise of _ xmm _ in order to discern the spectrum of these objects .
many of these objects have also been observed by chandra and their spectra have been well - fitted by simple power laws .
in addition to classification of the sources , we investigated some of the properties of the ulx sources .
we found a gap in the temperature distribution of high / soft state ulxs .
this gap may indicate a gap in mass distribution , which may provide clues to the nature of ulxs .
we also found that our ulxs are persistent sources ( not transients ) which occupy regions on the color - color diagram of @xcite also occupied by galactic black hole sources .
lastly , the existence of a substantial population of ulxs in nearby dwarf and other low star formation rate galaxies argues that ( in agreement with @xcite ) there is more than one source term for the origin of ulxs , with at least some of them not being associated with recent star formation , at least statistically .
we conclude , from an x - ray spectral and luminosity point of view , that our data are consistent with many of these objects having the properties expected of an imbh population .
however , we also find two other populations of objects , those whose blackbody temperature and luminosity correspond to that of stellar mass black holes with kt @xmath127kev and @xmath128 less than @xmath122ergs@xmath1 and a small population of objects whose x - ray spectra and luminosities are consistent with that of stellar mass black holes in the very high state .
thus , ulx selected purely on the basis of @xmath51kev x - ray luminosities are a composite class with @xmath129 being `` normal '' stellar mass black holes and the rest being consistent with a population of imbhs . in a follow - up paper
we will discuss the environments of these objects as revealed by _ xmm _
om uv imaging and the implications this has for the origin of ulxs .
l.w . gratefully acknowledges kip kuntz and m. coleman miller for helpful discussions .
we would also like to acknowledge r. narayan for asking the question `` are there any low - state ulxs ? '' at the kyoto black hole meeting .
anderson , s. et al .
2003 , , 126 , 2209 bauer , f.e . ,
brandt , w.n .
, & lehmer , b. 2003 , , 126 , 2797 begelman , m. c. 2002 , , 568 , l97 belloni , t. , et al .
1997 , , 488 , l109 brandt , w.n . , iwasawa , k. , & reynolds , c.s .
1996 , , 281 , l41 chen w. , schrader c. r. , livio m. , 1997 , , 491 , 312 colbert , e.j.m . &
ptak , a.f .
2002 , , 143 , 25 done , c. & gierlinski , m. 2003 , , 342 , 1041 ebisawa , k. , et al .
2003 , , 597 , 780 eracleous , m. , shields , j.c . , chartas , g. , & moran , e.c .
2002 , , 565 , 108 freedman , w. l. , et al .
1994 , , 427 , 628 freeman , k. c. , karlsson , b. , lynga , g. , burrell , j. f. , van woerden , h. , & goss , w. m. 1977 , a&a , 55 , 445 gierliski , m. & done , c. 2004 , , 349 , l7 grimm , h .- j . ,
gilfanov , m. , & sunyaev , r. 2003 , , 339 , 793 guainazzi , m. , et al .
2000 , a&a , 356 , 463 gltekin k. , miller m.c . , hamilton d.p . , 2004 ,
apj , 616 , 221 gutierrez , c.m . &
lopez - corredoira , m. 2005 , preprint ( astroph/0502290 ) hartwell , j. , et al .
2004 , , 348 , 406 heger a. , woosley s.e . , 2002 ,
apj , 567 , 532 ho , l.c .
, filippenko , a.v . , & sargent , w.l.w .
1997 , , 112 , 315 humphrey , p.j . ,
fabbiano , g. , elvis , m. , church , m.j .
, & balucinska - church , m. 2003 , , 344 , 134 jenkins , l.p .
, roberts , t.p . , warwick , r.s . ,
kilgard , r.e . ,
& ward , m.j .
2004 , , 349 , 404 jones , c. 1977 , , 214 , 856 jonker , p.g . & nelemans , g. 2004 , , 354 , 355 kaaret , p. , corebel , s. , prestwich , a.h .
, & zezas , a. 2003 , science , 299 , 365 kalogera , v. , henninger , m. , ivanova , n. , & king , a.r . 2004 , , 603 , l41 karachentsev , i.d . ,
sharina , m.e . , dolphin , a.e . , et al . 2002 , a&a , 385 , 21 kelson , d. d. 1996 , , 463 , 26 king , a.r .
2001 , , 552 , l109 kong , a.k.h .
2003 , , 346 , 265 kong , a.k.h . &
di stefano , r. 2003 , , 590 , 13 kording , e. , falcke , h. , & markoff , s. 2002 , a&a , 382 , l13 kubota , a. & makishima , k. 2004 , , 601 , 428 kubota , a. et al .
2001 , , 560 , l147 kuntz , k. et al .
2005 , , 620 , l31 lehmann , i. et al .
2005 , a&a , 431 , 847 lira , p. , lawrence , a. , & johnson , r.a .
2000 , , 319 , 17 liu , j. f. , bregman , j. n. , & seitzer , p. 2004
, , 602 , 249 madau p. , rees m.j .
, 2001 , apj , 551 , l27 magdziarz , p. & zdziarski , a.a .
1995 , , 273 , 837 makishima , k. , et al .
2000 , , 535 , 632 martin , c.l .
, kobulnicky , h.a . , & heckman , t.m . 2002 , , 574 , 663 mcclintock , j.e . & remillard , r.a .
2004 , preprint ( astroph/0306213 ) miller , j. m. , fabbiano , g. , miller , m. c. , & fabian , a. c. 2003 , , 585 , l37 miller , j.m . ,
fabian , a.c .
, & miller , m.c .
2004 , , 614 , l117 miller , m. c. & colbert , e. j. m. 2004 , ijmpd , 13 , 1 miller m.c . &
hamilton d.p . , 2002 , mnras , 330 , 232 miyamoto , s. et al . 1991 , , 383 , 784 pietsch , w. et al .
2001 , a&a , 365 , 174 pietsch , w. , haberl , f. , & vogler , a. 2003 , a&a , 402 , 457 ptak , a. & colbert , e. 2004 , , 606 , 291 pakull , m.w .
& mirioni , l. 2003 , rmxaa , 15 , 197 porquet , d. , reeves , j.n . ,
obrien , p. , & brinkmann , w. 2004 , a&a , 422 , 85 radecke , h.d .
1997 , a&a , 319 , 18 reynolds , c.s . , et al .
1997 , , 286 , 349 roberts , t.p . &
warwick , r.s .
2000 , , 315 , 98 roberts , t.p . ,
warwick , r.s . ,
ward , m.j .
, & goad , m.r .
2004 , , 349 , 1193 roberts , t.p .
2005 , , 357 , 1363 ross , r.r . &
fabian , a.c .
2005 , , 358 , 211 sakai , s. , madore , b.f .
, & freedman , w.l .
1996 , , 461 , 713 sakai , s. & madore , b.f .
1999 , , 526,599 schlegel , e.m . ,
petre , r. , colbert , e.j.m . , & miller , s. 2000 , , 120 , 2373 schlegel , e.m .
2004 , , 603 , 644 shapley , a. , fabbiano , g. , & eskridge , p.b .
2001 , , 137,139 schlegel , e.m . ,
petre , r. , colbert , e.j.m . , & miller , s. 2000 , aj , 120 , 2373 smith , d.a . & wilson , a.s .
2001 , , 557 , 180 stocke , j.t .
1983 , , 273 , 458 summers , l.k . , stevens , i.r . , strickland , d.k . , & heckman , t.m .
2003 , , 342 , 690 summers , l.k . ,
stevens , i.r . , strickland , d.k .
, & heckman , t.m .
2004 , , 351 , 1 sunyaev , r.a .
& titarchuk , l.g .
1980 , a&a , 86 , 121 swartz , d.a .
, ghosh , k.k . , tennant , a.f . , & wu , k. 2004 , , 154 , 519 terashima , y. & wilson , a.s .
2004 , , 601 , 735 tolstoy , e. , saha , a. , hoessel , j.g . , & mcquade , k. 1995 , , 110 , 1640 tosi , m. , bellazzini , m. , aloisi , a. , greggio , l. , leitherer , c. , & montegriffo , p. 2001
, , 122 , 1271 tully , r.b .
1988 , nearby galaxies catalog .
cambridge univ . press ,
cambridge voges , w. et al . 1999 ,
a&a , 349 , 389 vogler , a. & pietsch , w. 1996 , a&a , 311 , 35 wang , d. et al .
2004 , , 609 , 113 zdziarski , a. a. & gierliski , m. 2004 , progress of theoretical physics supplement , 155 , 99 zimmermann , h.u . & aschenbach , b. 2003 , a&a , 406 , 969 llllllll ngc247 & sab(s)d & 1.54 & 3.09 & & 0110990301 & 14536 & - + ngc253 & sab(s)c;hii & 1.40 & 3.73 & & 0110900101 , 0152020101 & 30711 , 110591 & starburst + ngc300 & sa(s)d & 3.11 & 2.56 & & 0112800101 & 43967 & - + ngc625 & sb(s)m ?
sp ; hii & 2.15 & 2.62 & & 0085100101 & 26288 & - + ngc1313 & sb(s)d ; hii & 4.0 & 4.17 & & 0106860101 & 41310 & - + ic0342 & sab(rs)cd ; hii & 30.3 & 3.9 & 1 & 0093640901 & 11217 & - + ngc1569 & ibm & 21.7 & 1.6 & 1 & 0112290801 & 15582 & starburst + ngc1705 & sa0- pec ; hii & 3.9 & 5.1 & 2 & 0148650101 & 58926 & starburst + mrk 71 & bcd ; hii & 3.9 & 3.4 & 3 & 0141150201 & 45919 & galaxy pair + ngc2403 & sab(s)cd ; hii & 4.15 & 3.56 & & 0150651201 & 11415 & - + holmberg ii & i m & 3.42 & 2.70 & & 0112520701 , 0112520901 & 13528 , 6860 & - + holmberg i & iab(s)m & 3.49 & 3.6 & 4 & 0026340101 & 26280 & - + m81 & sa(s)ab;liner & 4.12 & 3.6 & 4 & 0111800101 & 127913 & hol ix also in field of view + m82 & i0 ; hii & 4.14 & 3.9 & 5 & 0112290201 & 29387 & starburst + holmberg ix & i m & 4.0 & 3.6 & 4 & 0112521001 & 10350 & m81 also in field of view + sextans a & ibm & 3.85 & 1.4 & 6 & 0026340201 & 21618 & - + ic 2574 & sab(s)m & 2.29 & 3.6 & 7 & 0026340301 & 24263 & bursting star - formation + ngc 4214 & iab(s)m ; hii & 1.49 & 2.7 & & 0035940201 & 14744 & - + ngc 4258 & sab(s)bc;liner & 1.2 & 7.2 & & 0059140901 , 0110920101 & 16146 , 21895 & - + ngc4395 & sa(s)m;liner & 1.33 & 4 & & 0112521901 & 15842 & - + ngc4449 & ibm ; hii & 1.39 & 3.08 & & 0112521701 & 15522 & - + ngc4490 & sb(s)d & 1.78 & 7.8 & 1 & 0112280201 & 17754 & interacting with ngc4485 + ngc4631 & sb(s)d & 1.28 & 7.5 & 1 & 0110900201 & 53850 & - + ngc4736 & ( r)sa(r)ab;liner & 1.43 & 4.3 & 1 & 0094360601 & 23461 & - + ngc4945 & sb(s)cd ; sy2 & 15.9 & 3.1 & & 0112310301 & 23062 & - + ngc 5204 & sa(s)m ; hii & 1.42 & 4.8 & 1 & 0142770101 , 0142770301 & 19205 , 16387 & - + m51 & sc ; sy2 & 1.55 & 7.2 & 1 & 0112840201 & 20924 & galaxy pair + m83 & sab(s)c;hii & 3.94 & 6.2 & & 0110910201 & 30627 & starburst + ngc5253 & i m pec;hii & 3.77 & 3.2 & 1 & 0035940301 & 47216 & starburst + m101 & sab(rs)cd & 1.17 & 7.4 & 8 & 0104260101 & 43019 & - + ngc5408 & ib(s)m ; hii & 5.73 & 4.8 & 9 & 0112290601 & 7757 & - + circinus & sa(s)b ; sy2 & 57.8 & 4 & 10 & 0111240101 & 110496 & - + lllllllll ngc247 xmm2 & @xmath130 & @xmath131 & * 47.7/54 * & @xmath132 & @xmath133 & 48.8/54 & 0.33 & 0.38 + ngc 253 xmm2 ( obs 1 ) & @xmath134 & @xmath135 & * 69.1/74 * & @xmath136 & @xmath137 & 74.7/74 & 0.52 & 0.87 + ngc300 xmm4 & @xmath138 & @xmath139 & 90.6/45 & @xmath140 & @xmath141 & 117.6/45 & - & - + ngc1313 xmm4 & @xmath142 & @xmath143 & * 141.7/149 * & @xmath144 & @xmath145 & 140.1/149 & 0.33 & 0.69 + ic0342 xmm1 & @xmath146 & @xmath147 & * 159.5/185 * & @xmath148 & @xmath149 & 160/185 & 3.5 & 6.37 + ic0342 xmm2 & @xmath150 & @xmath151 & * 77.5/85 * & @xmath152 & @xmath153 & 74.9/85 & 4.64 & 8.44 + ic0342 xmm4 & @xmath154 & @xmath155 & 64/58 & @xmath156 & @xmath157 & * 56.9/58 * & 0.69 & 1.26 + mrk71 xmm1 & @xmath158 & @xmath159 & * 55.3/54 * & @xmath160 & @xmath161 & 59/56 & 0.19 & 0.27 + ngc2403 xmm4 & @xmath162 & @xmath163 & * 62.3/71 * & @xmath164 & @xmath165 & 62.3/71 & 0.31 & 0.48 + holmii xmm1 ( obs 2 ) & @xmath166 & @xmath167 & * 266.7/252 * & @xmath168 & @xmath169 & 309.4/252 & 3.5 & 3.1 + holm
i xmm2 & 0.35 & @xmath170 & * 39.2/45 * & @xmath171 & @xmath172 & 68.8/45 & 0.10 & 0.16 + holm i xmm3 & 0.35 & @xmath173 & * 34.4/32 * & 0.35 & @xmath174 & 42.1/32 & 0.12 & 0.19 + ic2574 xmm1 & @xmath175 & @xmath176 & 120.9/103 & @xmath177 & @xmath178 & * 107.5/103 * & 0.35 & 0.47 + ic2574 xmm2 & @xmath179 & @xmath180 & * 45.7/51 * & @xmath181 & @xmath182 & 65.4/52 & 0.22 & 0.34 + ic2574 xmm3 & @xmath183 & @xmath184 & * 40.3/49 * & 0.229 & @xmath185 & 76.5/49 & 0.22 & 0.34 + ngc4214 xmm1 & @xmath186 & @xmath187 & * 41.9/38 * & @xmath188 & @xmath189 & 44.5/38 & 0.25 & 0.22 + ngc4258 xmm2 ( obs 2 ) & @xmath190 & @xmath191 & * 83.6/57 * & @xmath192 & @xmath193 & 85.5/57 & 0.30 & 1.9 + ngc4258 xmm3 & @xmath194 & @xmath195 & * 38.9/37 * & @xmath196 & @xmath197 & 41.3/37 & 0.20 & 1.2 + & 3.8 & 1.82 & * 4/11 * & 2.7 & 7.14 & 5/11 & 0.077 & 0.48 + ngc4258 xmm4 & @xmath198 & @xmath199 & * 41.1/48 * & @xmath200 & @xmath201 & 45.2/48 & 0.39 & 2.4 + & @xmath202 & @xmath203 & * 77.03/77 * & @xmath204 & @xmath205 & 77.8/77 & 0.33 & 2.0 + ngc4395 xmm2 & @xmath206 & @xmath207 & * 38.6/36 * & @xmath208 & @xmath209 & 52/37 & 0.15 & 0.28 + ngc4395 xmm4 & @xmath210 & @xmath211 & * 16/25 * & @xmath208 & @xmath212 & 21.6/26 & 0.15 & 0.28 + ngc4449 xmm2 & @xmath213 & @xmath214 & * 103.5/112 * & @xmath215 & @xmath216 & 112.1/112 & 0.29 & 0.33 + ngc4490 xmm4 & @xmath217 & @xmath218 & _ 51.6/50 _ & @xmath219 & @xmath220 & 50.3/50 & 0.84 & 6.1 + ngc4490 xmm5 & @xmath221 & @xmath222 & * 60.1/65 * & @xmath223 & @xmath224 & 61.6/65 & 0.41 & 2.98 + ngc4631 xmm4 & @xmath225 & @xmath226 & 261.5/74 & @xmath227 & @xmath228 & 207.8/74 & - & - + ngc4631 xmm5 & @xmath229 & @xmath230 & 641.8/153 & @xmath229 & @xmath231 & 659/153 & - & - + ngc4945 xmm3 & @xmath232 & @xmath233 & * 30.1/30 * & @xmath234 & @xmath235 & 30.3/30 & 0.38 & 0.43 + ngc5204 xmm2 & @xmath236 & @xmath237 & _ 42.37/42 _ & @xmath238 & @xmath239 & 42.2/42 & 0.15 & 0.41 + & @xmath240 & @xmath241 & * 41.4/47 * & @xmath242 & @xmath243 & 39.4/47 & 0.25 & 0.69 + m51 xmm3 & @xmath244 & @xmath245 & * 63.2/72 * & @xmath246 & @xmath247 & 69.2/72 & 0.18 & 1.1 + m51 xmm4 & @xmath248 & @xmath249 & * 34.8/37 * & @xmath250 & @xmath251 & 34.8/37 & 0.16 & 0.99 + llllllll ngc247 xmm1 & @xmath252 & @xmath253 & @xmath254 & 86.5/93 & 25.7 & 6.2 & 7.1 + ngc253 xmm1 & @xmath255 & @xmath256 & @xmath257 & 225.9/230 & 36.7 & 2.7 & 4.5 + & @xmath258 & @xmath259 & @xmath260 & 567/580 & 44.6 & 3.4 & 5.7 + ngc253 xmm2 ( obs 2 ) & @xmath261 & @xmath262 & @xmath263 & 460.3/498 & 47.1 & 1.6 & 2.7 + ngc253 xmm3 & @xmath264 & @xmath265 & @xmath266 & 68.5/82 & 23.4 & 0.60 & 1.0 + & @xmath267 & @xmath268 & @xmath269 & 347.4/407 & 34.4 & 0.80 & 1.3 + ngc253 xmm4 & @xmath270 & @xmath271 & @xmath272 & 66.7/57 & 6.9 & 15 & 25 + & @xmath273 & @xmath274 & @xmath275 & 309.3/291 & 12.1 & 1.4 & 2.2 + ngc253 xmm5 & @xmath276 & @xmath277 & @xmath278 & 26.5/23 & 5.3 & 0.26 & 0.43 + & @xmath279 & @xmath280 & @xmath281 & 223.7/296 & 60.1 & 1.4 & 2.2 + ngc253 xmm6 & @xmath282 & @xmath283 & @xmath284 & 417.9/407 & 17.1 & 1.9 & 3.1 + ngc253 xmm7 & @xmath285 & @xmath286 & @xmath287 & 335.8/339 & 21.2 & 1.1 & 1.8 + ngc300 xmm1 & @xmath288 & @xmath289 & @xmath290 & 443.7/420 & 26.1 & 1.3 & 1.0 + ngc300 xmm2 & @xmath291 & @xmath292 & @xmath293 & 102.6/97 & 31.34 & 1.1 & 0.86 + ngc300 xmm3 & @xmath294 & @xmath295 & @xmath296 & 87.7/79 & 14.2 & 1.2 & 0.93 + ngc300 xmm6 & @xmath297 & @xmath298 & @xmath299 & 34.6/35 & 13 & 0.27 & 0.20 + ngc1313 xmm1 & @xmath300 & @xmath301 & @xmath302 & 194.1/201 & 35.4 & 0.64 & 1.3 + ngc1313 xmm2 & @xmath303 & @xmath304 & @xmath305 & 425.2/419 & 38.9 & 2.0 & 4.2 + ngc1313 xmm3 & @xmath306 & @xmath307 & @xmath308 & 441.7/424 & 336.6 & 10 & 22 + ic0342 xmm3 & @xmath309 & @xmath274 & @xmath310 & 129.5/107 & 56.3 & 31 & 56.4 + ngc1705 xmm1 & @xmath311 & @xmath312 & @xmath313 & 53/85 & 8.9 & 0.10 & 0.44 + ngc1705 xmm3 & @xmath314 & @xmath315 & @xmath316 & 69.8/65 & 11.1 & 0.15 & 0.48 + ngc2403 xmm1 & @xmath317 & @xmath318 & @xmath319 & 81.4/79 & 10.8 & 1.99 & 3.1 + ngc2403 xmm2 & @xmath320 & @xmath321 & @xmath322 & 163.1/151 & 16.4 & 1.0 & 1.6 + ngc2403 xmm3 & @xmath323 & @xmath324 & @xmath325 & 84.2/105 & 8.4 & 0.64 & 1.1 + holm ii xmm1(obs
1 ) & @xmath326 & @xmath327 & @xmath328 & 997.5/976 & 136.7 & 12 & 10 + m81 xmm1 & @xmath329 & @xmath330 & @xmath331 & 1273.7/1243 & 533.1 & 4.5 & 7.0 + & @xmath332 & @xmath333 & @xmath334 & 203.5/204 & 21.4 & 4.8 & 7.4 + m81 xmm2 & @xmath335 & @xmath336 & @xmath337 & 833.9/616 & 524.3 & 13 & 22 + m81 xmm4 & @xmath338 & @xmath339 & @xmath340 & 48.9/50 & 28.2 & 0.43 & 0.70 + m81 xmm5 & @xmath341 & @xmath342 & @xmath343 & 89/80 & 8.5 & 0.38 & 0.59 + holm ix xmm1 & @xmath344 & @xmath345 & @xmath346 & 866.6/878 & 134.3 & 10 & 16 + ngc4258 xmm1 & @xmath347 & @xmath348 & @xmath349 & 91.1/76 & 10.3 & 0.34 & 2.1 + ngc4258 xmm2 ( obs 1 ) & @xmath350 & @xmath351 & @xmath352 & 73.4/61 & 24.1 & 0.31 & 1.9 + ngc4395 xmm1 & @xmath353 & @xmath354 & @xmath355 & 168.2/154 & 26.9 & 1.4 & 2.7 + ngc4449 xmm1 & @xmath356 & @xmath357 & @xmath358 & 111.2/116 & 4.3 & 1.56 & 1.8 + ngc4449 xmm3 & @xmath359 & @xmath360 & @xmath361 & 119.9/87 & 34.1 & 1.1 & 1.3 + ngc4490 xmm1 & @xmath362 & @xmath363 & @xmath364 & 66.5/63 & 35 & 0.88 & 6.4 + ngc4631 xmm1 & @xmath365 & @xmath253 & @xmath366 & 371.3/345 & 12.1 & 0.96 & 6.5 + ngc4631 xmm2 & @xmath367 & @xmath368 & @xmath369 & 107.4/97 & 12.1 & 0.25 & 1.7 + ngc4631 xmm3 & @xmath370 & @xmath371 & @xmath372 & 127.1/96 & 18.9 & 0.15 & 1.0 + ngc4945 xmm1 & @xmath373 & @xmath374 & @xmath375 & 96.1/120 & 20 & 0.59 & 0.68 + ngc4945 xmm2 & @xmath376 & @xmath377 & @xmath378 & 105.8/113 & 8.7 & 0.66 & 0.76 + ngc4945 xmm4 & @xmath379 & @xmath380 & @xmath381 & 58.4/60 & 17.1 & 0.38 & 0.44 + ngc5204 xmm1 & @xmath382 & @xmath280 & @xmath383 & 543.0/559 & 49.1 & 1.98 & 5.6 + & @xmath384 & @xmath385 & @xmath386 & 461.4/496 & 71.6 & 2.92 & 8.0 + m51 xmm1 & @xmath387 & @xmath388 & @xmath389 & 97/80 & 13.4 & 0.31 & 1.9 + m51 xmm5 & @xmath390 & @xmath391 & @xmath392 & 59.8/70 & 196.2 & 220 & 1900 + m83 xmm4 & @xmath393 & @xmath394 & @xmath395 & 84.8/89 & 6.6 & 0.2 & 0.92 + m101 xmm1 & @xmath396 & @xmath397 & @xmath398 & 249.9/231 & 53.1 & 0.45 & 2.9 + m101 xmm2 & @xmath399 & @xmath400 & @xmath401 & 251.6/261 & 37.2 & 0.7 & 4.6 + ngc5408 xmm1 & @xmath402 & @xmath403 & @xmath404 & 316.4/337 & 80.4 & 3.97 & 10.9 + circinus xmm1 & @xmath405 & @xmath406 & @xmath407 & 749.4/861 & 13.5 & 12 & 23 + circinus xmm2 & @xmath408 & @xmath409 & @xmath410 & 438.5/430 & 79.4 & 5.6 & 10.7 + circinus xmm3 & @xmath411 & @xmath412 & @xmath413 & 269.3/260 & 15.9 & 7.6 & 14.5 + llllllll ngc300 xmm5 & @xmath414 & @xmath415&@xmath416 & 46.6/53 & 7.6 & 0.17 & 0.13 + ngc1705 xmm2 & @xmath417 & @xmath418 & @xmath419 & 85.5/74 & 6.5 & 0.09 & 0.27 + holm i xmm1 & @xmath420 & @xmath421 & @xmath422 & 97.4/93 & 5.4 & 0.6 & 0.93 + m81 xmm3 & @xmath423 & @xmath424 & @xmath425 & 77.1/78 & 4.25 & 0.53 & 0.82 + sextans a xmm1 & @xmath426 & @xmath427 & @xmath428 & 269.1/271 & 2.3 & 0.60 & 0.14 + ngc4214 xmm2 & @xmath429 & @xmath430 & @xmath431 & 46.4/44 & 4.5 & 0.4 & 0.35 + ngc4395 xmm3 & @xmath432 & @xmath433 & @xmath434 & 52/56 & 3.9 & 0.29 & 0.56 + ngc4490 xmm2 & @xmath435 & @xmath436 & @xmath437 & 42.4/54 & 7.1 & 0.65 & 4.7 + ngc4490 xmm3 & @xmath438 & @xmath439 & @xmath440 & 72.1/78 & 4.6 & 12 & 87.4 + ngc4736 xmm1 & @xmath441 & @xmath442 & @xmath443 & 54.9/51 & 7.9 & 8.1 & 17.9 + m51 xmm2 & @xmath444 & @xmath445 & @xmath446 & 70.7/68 & 4.5 & 0.36 & 3.0 + m51 xmm6 & @xmath447 & @xmath448 & @xmath449 & 36.9/41 & 4.07 & 5.6 & 35 + m51 xmm7 & @xmath450 & @xmath451 & @xmath452 & 31.7/29 & 6.1 & 0.26 & 1.6 + m83 xmm1 & @xmath453 & @xmath454 & @xmath455 & 177.7/209 & 4.7 & 0.63 & 2.5 + m101 xmm3 & @xmath456 & @xmath457 & @xmath458 & 145.5/131 & 3.4 & 0.56 & 3.7 + m101 xmm4 & @xmath459 & @xmath460 & @xmath461 & 158.2/138 & 7.5 & 0.34 & 2.2 + m101 xmm5 & @xmath462 & @xmath368 & @xmath463 & 45.1/44 & 2.8 & 0.13 & 0.85 + lllll ngc247 xmm1 & 13.4258 & 7.07734 & 54 & hs ulx + ngc253 xmm1 & 9.31469 & 2.44574 & 19 & hs ulx + ngc253 xmm2 & 4.3701 & 2.15292 & 17 & hs ulx + ngc253 xmm6 & 5.05828 & 3.92514 & 30 & hs ulx + ngc1313 xmm3 & 37.0364 & 27.9692 & 215 & hs ulx + ngc1313 xmm4 & 1.50345 & & 116 & ls imbh cand .
+ ic0342 xmm1 & 14.1215 & & 1086 & ls ulx + ic0342 xmm2 & 19.8129 & & 1524 & ls ulx + ic0342 xmm3 & 114.015 & 95.4068 & 734 & hs ulx + mrk71 xmm1 & 0.2993 & & 23 & ls imbh cand .
+ ngc2403 xmm1 & 4.1497 & 2.14873 & 17 & hs ulx + ngc2403 xmm4 & 0.57068 & & 44 & ls imbh cand .
+ holmberg ii xmm1 & 0.88906 & & 68 & ls imbh cand .
+ & 16.8335 & 11.4543 & 88 & hs ulx + holmberg i xmm2 & 10.5158 & & 809 & ls ulx + m81 xmm1 & 15.7004 & 3.17932 & 24 & hs ulx + holmberg ix xmm1 & 31.0582 & 28.1445 & 216 & hs ulx + ngc4214 xmm1 & 0.26699 & & 21 & ls imbh cand .
+ ngc4258 xmm3 & 0.46503 & & 36 & ls imbh cand .
+ ngc4395 xmm1 & 9.04609 & 2.94683 & 23 & hs ulx + ngc4449 xmm2 & 2.48586 & & 191 & ls imbh cand .
+ ngc4490 xmm1 & 16.8513 & 3.21972 & 25 & hs ulx + ngc4490 xmm2 & 7.36612 & 4.51554 & 35 & hs ulx + ngc4490 xmm3 & 240.653 & 176.04 & 1354 & hs ulx + ngc4490 xmm4 & 1.66829 & & 128 & ls imbh cand .
+ ngc4490 xmm5 & 12.7136 & & 978 & ls ulx + ngc4631 xmm1 & 10.6527 & 8.59661 & 66 & hs ulx + ngc4736 xmm1 & 31.6561 & 27.3664 & 211 & hs ulx + ngc4945 xmm3 & 0.44985 & & 35 & ls imbh cand .
+ ngc5204 xmm1 & 22.4756 & 2.20492 & 17 & hs ulx + ngc5204 xmm2 & 5.57769 & & 429 & ls ulx + m51 xmm2 & 4.25898 & 3.57502 & 28 & hs ulx + m51 xmm3 & 2.10133 & & 162 & ls imbh cand .
+ m51 xmm4 & 2.56064 & & 197 & ls imbh cand .
+ m51 xmm6 & 46.7642 & 39.5189 & 304 & hs ulx + m101 xmm1 & 8.04916 & 7.68224 & 59 & hs ulx + m101 xmm2 & 7.54268 & 4.96709 & 38 & hs ulx + m101 xmm3 & 10.9792 & 1.03659 & 8 & hs ulx + ngc5408 xmm1 & 20.9211 & 11.5369 & 89 & hs ulx + circinus xmm1 & 70.3579 & 56.1033 & 432 & hs ulx + circinus xmm2 & 208.746 & 0.69929 & 5 & hs ulx + circinus xmm3 & 771.157 & 0.212957 & 2 & hs ulx +
in order to determine the number of counts required to distinguish whether a blackbody component is statistically significant for the sources fit with a two - component model , we simulated spectra based on that of some of the brightest sources .
we chose to simulate spectra of bright two - component spectra exhibiting three different cases : ( 1 ) the flux from the blackbody dominates over the power law component at 2kt , ( 2 ) an intermediary case , and ( 3 ) the flux from the power law dominates over the blackbody component at 2kt .
such simulations would allow us to determine the uncertainty in our claims of a combined fit being a better descriptor of the data .
this is necessary because there is no a priori model which predicts the relative intensities of the two components and , as we know from studies of galactic black holes , these components show a wide variety of relative intensities . to this end
, we simulated spectra using the best - fit absorbed blackbody and power law model with the _ fakeit _ command in xspec .
we chose ( 1 ) ngc 247 xmm1 , ( 2 ) ngc 5408 xmm1 , and ( 3 ) holmberg ii xmm1 as our seed observations .
these objects all have very high signal - to - noise and thus the fits are robust
. the respective ratios of powerlaw flux to blackbody flux contributions at 2kt are : ( 1 ) @xmath464 , ( 2 ) 1.77 , and ( 3 ) 3.52 .
all of these sources have comparable blackbody temperatures indicative of our high - state ulx candidates ( roughly kt @xmath465 ) .
we simulated 500 spectra each , using the two - component model , for each of 2000counts , 1000counts , 400counts , and 200counts for the pn .
each simulated spectrum , based on the best - fit blackbody and power law model , was fit with an absorbed blackbody and powerlaw model as well as an absorbed pure - powerlaw model .
we placed the constraint that the blackbody temperature must lie within the range of @xmath466kev ( the range at which it would be detectable in the _ xmm - newton _ bandpass ) .
we allowed the power law index to vary over the range @xmath467 for the power law component of the combined blackbody and power law model .
however , we placed a constraint that the power law component must lie within the range @xmath468 for the simple power law model to be consistent with our fits to the sources we claim are best fit by simple power laws .
this constraint ensures that the spectral index would exhibit that of our classified `` low - state '' objects .
when analysing and classifying our real spectra , we declared a detection of the thermal disk component if the addition of this component ( to a baseline powerlaw model ) led to an improvement of the goodness of fit parameter by at least @xmath469 .
we can use the above simulations to address the detectability of a thermal disk component using this @xmath470 threshold as a function of the relative strength of the thermal component and the number of counts in the spectrum . for each simulation , we fit the spectrum with both a single absorbed power - law and a 2-component powerlaw and thermal disk model and compute the quantity @xmath471 . in fig .
a1 , we plot the distribution of @xmath470 from our 500 simulations for the weak and strong blackbody component for spectra with 400counts and 2000counts .
it is clear that we can not detect a weak thermal component in a 400 count spectrum the vast majority of the simulations ( @xmath472% ) result in @xmath473 .
however , even a weak thermal component is easily detected in a 2000 count spectrum ( not a single simulation gave @xmath473 ) .
the strong blackbody case is detectable with high significance even in a 400 count spectrum ( @xmath474% of the simulations resulted in @xmath37 ) . when we increase the upper limit of the range of the spectral index in _ xspec _ for the power law model to @xmath475 , our confidence levels decrease .
for a weak thermal component with 400counts , all of the simulations result in @xmath473 . at 2000counts ,
only @xmath121% of the simulations for a weak thermal component have @xmath473 . for a strong thermal component ,
20% of the simulations yield @xmath473 for 400counts while , as was the case for @xmath476 , none of the simulations gave @xmath473 for 2000counts . thus allowing the @xmath2 to `` float freely '' or remain unconstrained further decreases the confidence levels . when the @xmath2 parameter is allowed to float the spectra are fit with higher @xmath2 values in order to compensate for the missing blackbody component .
when the upper limit was instituted at 2.0 or 3.0 , we found that all 500 simulated spectra were fit with a @xmath476 or 3.0 , respectively , for the simple power law model .
this same affect is not seen in the @xmath2 of the two - component model , where the value ranges between 1.4 and 4.0 with a peak in the distribution near that of the original model used to simulate the spectra .
thus , higher power law indices ( @xmath477 ) can indicate the necessity of an added blackbody component .
since there are no galactic black holes whose broad band spectra are well fit by steep power laws it seems that restricting the allowed power law indices is more consistent with the nature of galactic black holes . in fitting our sources with the three `` standard '' models
, we allowed the @xmath2 to float , thus the problem of a missing blackbody component being compensated for by a steep power law should not have factored into our classification criteria .
our next set of simulations sought to determine our confidence in the simple power law fit being an adequate descriptor of the spectra .
binning provides a problem in distinguishing between a powerlaw and a curvature in the spectrum at the low energy range , since the binning procedure can wash out a low kt blackbody from the spectrum . for this reason
, we chose to simulate unbinned pn spectra for a source we categorized as a low - state object , ic 0342 xmm1 .
ic 0342 xmm1 represents characteristics typical of our low - state candidates , namely it is within the proper luminosity range and it has a power law index of @xmath14 ( the median of the distribution for low - state objects is 2.03 ) and a hydrogen column density near the median of all the fitted values ( where the median value is @xmath478@xmath55 and the value of ic 0342 x-1 s column density is @xmath479@xmath55 ) .
we chose this source for these reasons and the high number of counts in its pn spectrum .
instead of using the @xmath25 statistic ( used for binned data ) , we chose to use the maximum - likelihood statistic , _ c - stat _ , in _ xspec _ ( which uses unbinned data ) .
we simulated 500 spectra using the best - fit parameters for the simple power law fit using the _ fakeit _ command .
we fit the simulated spectra with two models : the simple powerlaw and a combined blackbody and power law model and computed the change in the goodness of fit , @xmath470 .
for the two - component model , once again we placed the constraint that the blackbody temperature remain in the range that it would be detectable by _ xmm - newton _
, 0.07 - 4kev .
we followed this procedure for 1000 , 2000 , and 4000 counts . at both the 1000 and 2000 count level ,
the addition of a second component has no effect on the c - statistic ( the distributions in c - space for both the power law fit and the two - component fit are indistinguishable ) . at the 4000 count level , the c - statistic distributions for the two models separate such that there is an 26% confidence that the two - component model is a better fit to the data .
we find , when we examine the model parameters , that the power law index ( @xmath2 ) for the two - component model ranges between @xmath480 with the mean value @xmath481 .
the mean value for all three count levels used clustered around this value , though the range in @xmath2 increased as the counts decreased .
the mean blackbody temperature for simulations with 4000 counts was 1.17kev ( with the range varying between the amount previously noted ) with a median of 1.06kev . for simulations with lower counts
, the blackbody temperature becomes higher ( 1.24kev for 1000 counts ) with a higher median ( 1.83kev for 1000 counts ) .
this tells us that the fitting procedure tends to approximate a pure power law spectrum as a two - component spectrum with @xmath2 equal to that of the true spectral index but with a blackbody temperature higher than those observed in our study ( 1.1kev or higher ) which in the _ xmm _ band can be approximated as a power law .
if we found spectra in our sample that were best fit with a low spectral index and a high blackbody temperature , we might suspect that the spectrum s true nature is a power law .
we also note that if the hydrogen column density is large , much @xmath482@xmath55 , a low temperature ( kt ) blackbody component can be much more difficult to detect .
the following sources were not best fit by the standard models employed in this study : this source was classified as a super - soft x - ray source by @xcite .
we find that the standard single - component absorbed blackbody model is a much better model for this spectrum .
in fact , the power law , bremsstrahlung , and combined models do not fit the data within the 90% confidence range .
fitting an absorbed blackbody , we find the best fit corresponds to the following parameters : @xmath483@xmath55 , @xmath484kev , and @xmath485 .
this fit yields an unabsorbed flux of @xmath486erg@xmath55s@xmath1 .
the spectrum of this source clearly identifies it as a super - soft x - ray source . as with ngc300 xmm4
, the standard models employed in this study did not adequately match the data .
the best fitting model corresponds again to an absorbed blackbody .
the corresponding parameters are as follows : @xmath487@xmath55 , @xmath488kev , and @xmath489 .
this fit yields an unabsorbed flux of @xmath490 erg@xmath55s@xmath1 .
the position of this source shows it to be coincident with a globular cluster associated with that galaxy .
this source was identified as a bulge x - ray source , possibly powered by accretion , in a rosat study of ngc4631 @xcite .
the spectrum of this source was best fit with an absorbed power law + an absorbed _ vapec _ model .
this indicates the prescence of hot gas , indicating a possible thermal x - ray source .
the spectrum of this source was not adequately fit with any of the standard models used in this investigation .
the spectrum exhibits a prominent fe k line in the pn spectrum that is well fit by a gaussian ( _ zgauss _ ) at 6.4 kev .
we find that the entire spectrum is best fit with a partial covering fraction absorption model ( _ pcfabs _ ) in combination with the normal absorption , a power law , and a gaussian .
the best fit parameters yield : absorption column density , @xmath491@xmath55 , partial covering absorption , @xmath492@xmath55 , partial covering fraction @xmath493 , @xmath494 , and @xmath495 .
the source is clearly located within the optical galaxy , and is thus unlikely to be a background agn .
the spectrum and luminosity ( l@xmath496ergs@xmath1 ) of this source suggests that it is an agn .
the location of the source , from the digital sky survey , places it within the dwarf companion of m51 making a value of the optical flux hard to constrain .
the best fit to this source was an absorbed blackbody + power law and the spectral parameters are listed in table 4 .
like ngc4945 xmm5 , this source was best fit by a partial absorption model .
however , this source showed no evidence of an fe k line .
we fit this source s spectra using a partial covering fraction absorption model in combination with the normal absorption model and a power law .
the best fit parameters yield : absorption column density , @xmath497@xmath55 , partial covering absorption , @xmath498@xmath55 , partial covering fraction @xmath499 , @xmath500 , and @xmath501 . the unabsorbed flux in the range of 0.3 - 10 kev equals @xmath502erg@xmath55s@xmath1 .
table c4 includes the parameters for the `` ulx '' sources best fit by the _ compst _ model .
a discussion of these sources and interpretation of the data is included in section 4.3 .
llllllll ngc247 xmm1 & 00 47 03.8 & -20 47 46.2 & 3458 , 1389 , 1379 & 20.33 , 5.8 , 6.4 & 1rxs j004704.8 - 204743 & sa & - + ngc247 xmm2 & 00 47 03.1 & -20 37 02.5 & - , 597 , 600 & - , 1.9 , 1.4 & - & sa & - + ngc253 xmm1 & 00 47 32.8 & -25 17 52.6 & - , 3156 , 2985 & - , 11.38 , 9.9 & ngc253 psx-2 & near center & 1 + & & & - , 12654 , 12812 & - , 8.7 , 9.1 & & - & + ngc253 xmm2 & 00 47 22.4 & -25 20 55.2 & - , 825 , 942 & - , 2.8 , 2.97 & ngc253 psx-5 & sa & 1 + & & & - , 10347 , 10304 & - , 8.2 , 8.5 & & - & + ngc253 xmm3 & 00 47 35.2 & -25 15 13.8 & - , 870 , 1065 & - , 3 , 3.41 & ngc253 psx-7 & sa & 1 + & & & - , 5988 , 6131 & - , 4.2 , 4.6 & & - & + ngc253 xmm4 & 00 47 23.3 & -25 19 06.5 & - , 649 , 703 & - , 1.4 , 1.3 & - & sa & - + & & & - , 3823,3738 & - , 1.7 , 1.95 & - & - & + ngc253 xmm5 & 00 47 17.6 & -25 18 12.1 & - , 295 , 313 & - , 1.08 , 1.04 & ngc253 psx-4 & sa & 2 + & & & - , 4199 , 4303 & - , 3.6 , 3.8 & & - & + ngc253 xmm6 & 00 47 42.8 & -25 15 05.5 & - , 6081 , 6407 & - , 3.8 , 4.4 & ngc 0253 [ vp99 ] x40 & sa & 1 + ngc253 xmm7 & 00 47 09.2 & -25 21 21.7 & - , 4300 , 4454 & - , 2.6 , 3.0 & - & sa & - + ngc300 xmm1 & 00 55 09.9 & -37 42 13.9 & 6778 , 2248 , 2453 & 18.6 , 4.9 , 5.4 & - & sa & - + ngc300 xmm2 & 00 55 10.6 & -37 48 36.7 & 1364 , 456 , 463 & 3.1 , 0.9 , 0.9 & - & edge sa ?
& - + ngc300 xmm3 & 00 54 49.7 & -37 38 53.8 & 915 , 442 , 435 & 2.2 , 0.9 , 0.9 & - & sa & - + ngc300 xmm4 & 00 55 10.9 & -37 38 53.8 & 745 , 224 , 233 & 1.9 , 0.3 , 0.4 & xmmu j005511 -3749 ; sss & sa & 3 + ngc300 xmm5 & 00 55 21.1 & -37 29 19.5 & 750 , 247 , 250 & 1.7 , 0.4 , 0.5 & - & out & - + ngc300 xmm6 & 00 54 44.2 & -37 51 04.5 & 517 , 187 , 165 & 1.1 , 0.3 , 0.3 & - & out & - + ngc625 xmm1 & 01 35 06.8 & -41 26 17.1 & 5832 , 577 , 2119 & 3 , 0.6 , 1.4 & - & sa & - + ngc1313 xmm1 & 03 18 19.9 & -66 29 10.7 & 2876 , 900 , 810 & 8.6 , 3.2 , 2.8 & ngc1313 [ cps95 ] x-1 ; ixo 07 & sa & 4,5 + ngc1313 xmm2 & 03 17 38.8 & -66 33 05.3 & 7568 , 2357 , 2108 & 25.4 , 8.8 , 7.8 & ngc1313 [ cps95 ] x-3 ; sn 1978
k & edge sa & 6 + ngc1313 xmm3 & 03 18 22.5 & -66 36 06.2 & 6960 , 2179 , 1793 & 23.3 , 8.1 , 6.6 & ngc1313 [ cps95 ] x-2 ; ixo 08 & edge sa ? &
4,5 + ngc1313 xmm4 & 03 18 18.5 & -66 30 05 & 2075 , 659 , 567 & 5.1 , 1.8 , 1.5 & ngc1313 [ spc2000 ] x-8 & near center & - + ic0342 xmm1 & 03 45 55.8 & + 68 04 54.5 & 1802 , 1216 , 1105 & 33.6 , 12.4 , 11.1 & ic0342 [ rw2000 ] x-1 ; ixo 22 & sa & 5 , 7 , 8 + ic0342 xmm2 & 03 46 15.0 & + 68 11 11.2 & 1147 , 541 , 184 & 21.1 , 5.5 , 1.7 & ic0342 [ rw2000 ] x-3 & sa & 7 , 8 + ic0342 xmm3 & 03 46 48.6 & + 68 05 43.2 & 1186 , 670 , 606 & 21.7 , 6.8 , 6.0 & ic0342 [ llj2000 ] x-2 & near center & 7 , 8 + ic0342 xmm4 & 03 46 57.2 & + 68 06 20.2 & 551 , 338 , 377 & 9.5 , 3.9 , 3.6 & ic0342 [ rw2000 ] x-6 & near center & 7 , 8 + ngc1569 & & & & & & & + ngc1705 xmm1 & 04 54 57.6 & -53 24 23.5 & 1174 , 400 , 371 & 2.4 , 0.6 , 0.6 & rx j0454.9 - 5324 & out ? & - + ngc1705 xmm2 & 04 54 19.6 & -53 20 41.9 & 933 , 375 , 397 & 1.9 , 0.6 , 0.6 & & out ? & - + ngc1705 xmm3 & 04 54 38.1 & -53 18 16.2 & 698 , 372 , 418 & 1.4 , 0.6 , 0.6 & wga j0454.7 - 5318 & out ? & - + mrk71 xmm1 & 07 28 51.8 & + 69 07 27 & 832 , 225 , 207 & 3.9 , 0.97 , 0.9 & - & out & - + ngc2403 xmm1 & 07 36 25.6 & + 65 35 40 & - , 1199 , 672 & - , 1.0 , 0.60 & ngc2403 [ rw2000 ] x-1 & edge sa & - + ngc2403 xmm2 & 07 36 50.2 & + 65 36 02.1 & 1964 , 729 , 672 & 1.99 , 0.63 , 0.60 & - & near center & - + ngc2403 xmm3 & 07 36 55.4 & + 65 35 40.3 & 1497 , 489 , 378 & 1.4 , 0.40 , 0.33 & - & near center & - + ngc2403 xmm4 & 07 37 02.5 & + 65 39 35.2 & 1004 , 274 , 288 & 0.52 , 0.15 , 0.21 & ngc2403 [ rw2000 ] x-4 & edge sa ? & - + hol ii xmm1 & 08 19 28.8 & + 70 42 20.3 & 31052 , 1257 , 10807 & 272.2 , 75.5 , 72.7 & holm ii x-1 ; ixo 31 & near center ?
& 9 + & & & 3853 , 1361 , 1452 & 78.3 , 18.9 , 20.5 & & - & + hol i xmm1 & 09
41 30 & + 71 12 34 & 687 , 768 , 754 & 2.9 , 2.7 , 2.6 & - & out ?
& - + hol i xmm2 & 09 39 59.7 & + 71 06 40.2 & 575 , 203 , 224 & 2.5 , 0.7 , 0.7 & 1wga j0940.0 + 7106 & out ?
& - + hol i xmm3 & 09 42 06.7 & + 71 04 45.3 & 452 , 141 , 141 & 1.7 , 0.4 , 0.4 & - & out ?
& - + m81 xmm1 & 09 55 32.9 & + 69 00 34.8 & 50788 , - , 18988 & 51.1 , - , 21.7 & ngc3031 [ rw2000 ] x-11 & sa & - + & & & - , 1227 , - & - , 12 , 13.1 & & - & - + m81 xmm2 & 09 55 24.8 & + 69 01 11.7 & 17871 , - , 4121 & 13.2 , - , 4.4 & sn 1993j & sa & 10 + m81 xmm3 & 09 55 10.6 & + 69 05 02.2 & - , - , 1970 & - , - , 1.5 & ngc3031 [ rw2000 ] x-05 & sa & - + m81 xmm4 & 09 55 24.3 & + 69 10 00.2 & - , - , 1197 & - , - , 1.0 & ngc3031 [ rw2000 ] x-08 & edge sa & - + m81 xmm5 & 09 55 49.2 & + 69 05 30.5 & - , - , 2077 & - , - , 2.3 & - & sa & - + m82 & & & & & & & + hol ix xmm1 & 09 57 53.3 & + 69 03 48.7 & 14976 , 6546 , 6586 & 207.3 , 64.3 , 65 & hol ix x-1 ; ixo 34 & ? & 4 , 5 + sextans a xmm1 & 10 11 24.6 & -04 42 17.2 & 3963 , 1323 , 1242 & 17.2 , 5 , 4.5 & - & out & - + ic2574 xmm1 & 10 28 42.4 & + 68 28 17.8 & 1047 , 673 , 623 & 8.3 , 2.6 , 2.5 & - & sa & - + ic2574 xmm2 & 10 26 33.5 & + 68 29 32.1 & 533 , 335 , 300 & 4.3 , 1.3 , 1.2 & - & out & - + ic2574 xmm3 & 10 27 22.2 & + 68 18 47.6 & 538 , 293 , 301 & 4.2 , 1.1 , 1.2 & - & out & - + ngc4214 xmm1 & 12 15 37.0 & + 36 19 29.4 & 434 , 230 , 225 & 2.9 , 1.6 , 1.5 & ngc4214 [ hss2004 ] 11 & sa & - + ngc4214 xmm2 & 12 15 58.2 & + 36 22 38.5 & 626 , 160 , 285 & 3.4 , 0.6 , 1.5 & - & out & - + ngc4258 xmm1 & 12 18 47.8 & + 47 20 51.7 & 828 , 470 , 444 & 7.0 , 2.8 , 2.8 & - & sa & - + & & & & & & - & - + ngc4258 xmm2 & 12 18 57.8 & + 47 16 06.8 & 732 , 337 , 337 & 5.8 , 2 , 2.1 & ngc4258 [ rw2000 ] x-7 & sa & - + & & & 716 , 268 , 290 & 1.8 , 1 , 1.0 & & - & + ngc4258 xmm3 & 12 18 56.5 & + 47 21 24.3 & 489 , 182 , 184 & 3.8 , 0.9 , 1.1 & ngc4258 [ rw2000 ] x-5 & sa & - + & & & - , 160 , 142 & - , 0.3 , 0.3 & & - & + ngc4258 xmm4 & 12 19 23.2 & + 47 09 37.2 & 644 , 277 , 224 & 4.9 , 1.3 , 1.3 & hellas 288 & edge sa ? & - + & & & 964 , 410 , 382 & 0.3 , 0.2 , 0.1 & & - & + ngc4395 xmm1 & 12 26 01.5 & + 33 31 29 & 2162 , 862 , 921 & 15.5 , 5.1 , 5.6 & ngc4395 [ rw2000 ] x-1 ; ixo 53 & sa & - + ngc4395 xmm2 & 12 25 25.3 & + 33 36 46.4 & 392 , 252 , 208 & 2.6 , 1.5 , 1.2 & - & sa ? & - + ngc4395 xmm3 & 12 25 32.6 & + 33 25 27.9 & 763 , 277 , 278 & 5.2 , 1.5 , 1.6 & - & out ? & - + ngc4395 xmm4 & 12 25 42.7 & + 33 40 00.1 & 516 , 60 , 40 & 3.1 , 0.3 , 0.2 & - & out & - + ngc4449 xmm1 & 12 28 18 & + 44 06 30.9 & 1409 , 608 , 593 & 10.6 , 3.6 , 3.5 & xrb ? ; ngc4449 [ rw2000 ] x-7 ; source 27 & sa & - + ngc4449 xmm2 & 12 28 09.3 & + 44 05 03.9 & 1503 , 527 , 586 & 11.3 , 3.1 , 3.5 & snr ?
; ngc4449 [ rw2000 ] x-1 ; source 10 & sa & - + ngc4449 xmm3 & 12 28 11.1 & + 44 06 43.9 & 1094 , 549 , 404 & 8.3 , 3.2 , 2.3 & snr ; source 15 & sa & - + ngc4490 xmm1 & 12 30 32.4 & + 41 39 14.6 & 746 , 323 , 393 & 6.0 , 1.9 , 2.2 & ngc4490 [ rw2000 ] x-1 & sa & - + ngc4490 xmm2 & 12 30 36.5 & + 41 38 33.3 & 656 , 299 , 310 & 5.3 , 1.7 , 1.7 & ngc4490 [ rw2000 ] x-2 & near center & - + ngc4490 xmm3 & 12 30 43.3 & + 41 38 11.5 & 832 , 501 , 461 & 5.8 , 2.6 , 2.2 & ngc4490 [ rw2000 ] x-4 & sa & - + ngc4490 xmm4 & 12 30 31.1 & + 41 39 08.1 & 546 , 291 , 286 & 4.4 , 1.7 , 1.6 & cxou j123030.8 + 413911 & sa & - + ngc4490 xmm5 & 12 30 30.3 & + 41 41 40.3 & 413 , 587 , 482 & 2.7 , 3.1 , 2.3 & ngc4485 [ rw2000 ] x-1 & sa & - + ngc4631 xmm1 & 12 41 55.8 & + 32 32 14 & 5093 , 1969 , 1762 & 13 , 4.2 , 3.8 & ngc4631 [ rw2000 ] x-1 ; ixo 68 & sa & - + ngc4631 xmm2 & 12 41 57.5 & + 32 32 01 & 1273 , 531 , 400 & 3.3 , 1.2 , 0.9 & ngc4631 [ rw2000 ] x-2 & sa & - + ngc4631 xmm3 & 12 41 58.2 & + 32 28 49.6 & 1271 , 443 , 457 & 2.8 , 0.9 , 0.9 & cxousexsi j124158.0 + 322851 & out & - + ngc4631 xmm4 & 12 42 16.1 & + 32 32 48.8 & 957 , 400 , 429 & 2.3 , 0.8 , 0.8 & [ vp96 ] h13 & sa & - + ngc4631 xmm5 & 12 42 11.2 & + 32 32 33.6 & 1626 , 894 , 866 & 3.9 , 1.9 , 1.8 & [ vp96 ] h12 ; [ hfe2003 ] psx-01 & sa & - + ngc4736 xmm1 & 12 50 50.2 & + 41 07 12 & 713 , 273 , 227 & 6.7 , 2.1 , 1.7 & ngc4736 x-4 & near center & - + ngc4945 xmm1 & 13 05 33.3 & -49 27 36.3 & 1456 , 646 , 595 & 7.9 , 2.9 , 2.6 & ngc4945 [ gmb2000 ] x-2 & sa & - + ngc4945 xmm2 & 13 05 38.4 & -49 25 45.3 & 1393 , 600 , 523 & 7.6 , 2.7 , 2.3 & ngc4945 [ r97 ] x-3 & sa & - + ngc4945 xmm3 & 13 05 18.8 & -49 28 24 & - , 357 , 362 & - , 1.6 , 1.6 & & sa & - + ngc4945 xmm4 & 13 05 22.2 & -49 28 26.3 & 731 , 331 , 332 & 4.0 , 1.5 , 1.5 & ngc4945 [ bir96 ] x-1 ? & sa & - + ngc4945 xmm5 & 13 05 25.7 & -49 28 30.7 & 772 , 267 , 301 & 4.2 , 1.2 , 1.32 & & sa & - + ngc5204 xmm1 & 13 29 38.5 & + 58 25 03.6 & 9981 , 3352 , 3384 & 62.8 , 17.7 , 17.9 & ngc5204 [ rw2000 ] x-1 ; ixo 77 & sa & 11 + & & & 9231 , 2284 , 2349 & 85.5 , 24.7 , 25.8 & & - & + ngc5204 xmm2 & 13 29 27.4 & + 58 25 31.8 & 573 , 170 , 231 & 3.4 , 0.8 , 1.2 & - & edge sa & - + & & & 772 , 161 , 121 & 5.0 , 1.7 , 1.1 & & - & + m51 xmm1 & 13 29 40 & + 47 12 36.2 & 1102 , 367 , 409 & 6.2 , 1.8 , 2 & ngc5194 [ rw2000 ] x-1 & sa & - + m51 xmm2 & 13 30 07.7 & + 47 11 04.8 & 514 , 540 , 549 & 2.6 , 2.2 , 2.2 & ixo 81 & sa & - + m51 xmm3 & 13 30 01.1 & + 47 13 41.4 & 1004 , 315 , 311 & 4.1 , 1.1 , 1.1 & cxou j133001.0 + 471344 ; ixo 80 & sa & - + m51 xmm4 & 13 30 06 & + 47 15 38.9 & 518 , 183 , 166 & 2.8 , 0.9 , 0.8 & ngc 5195 [ rw2000 ] x-1 & sa & - + m51 xmm5 & 13 29 59.6 & + 47 15 54 & 1079 , 359 , 257 & 5 , 1.5 , 1 & cxou j132959.5 + 471559 & near center & - + m51 xmm6 & 13 29 57.5 & + 47 10 45.3 & 536 , 206 , 247 & 1.9 , 0.6 , 0.8 & cxou j132957.6 + 471048 & sa & - + m51 xmm7 & 13 29 53.6 & + 47 14 31.5 & 452 , 141 , 143 & 2.4 , 0.6 , 0.7 & cxou j132953.8 + 471432 & edge sa & - + m83 xmm1 & 13 37 19.8 & -29 53 49.8 & 3074 , 927 , 987 & 12 , 3.3 , 2.5 & rx j133719 -2953.6 ; ixo 82 & sa & - + m83 xmm2 & 13 36 59.4 & -29 49 57.2 & 1133 , 371 , 397 & 4 , 1.3 , 1 & cxou j133659.4 -294959 & sa & - + m83 xmm3 & 13 37 04.4 & -29 51 24 & 1724 , 576 , 459 & 7.2 , 2.3 , 1.2 & cxou j133704.3 -295121 & sa & - + m83 xmm4 & 13 37 01.5 & -29 53 26 & 1289 , 345 , 401 & 4.9 , 1.3 , 1.0 & cxou j133701.4 -295326 & sa & - + ngc5253 & & & & & & & + m101 xmm1 & 14 03 14.7 & + 54 18 05 & 2690 , 1449 , 1417 & 10.3 , 3.3 , 3.3 & cxou j140313.9 + 541811 ; xmm-2 & sa & 12 + m101 xmm2 & 14 03 03.8 & + 54 27 37 & 2825 , 1623 , 1551 & 10.3 , 3.6 , 3.3 & xmm-1 & edge sa ? & 12 + m101 xmm3 & 14 04 14.6 & + 54 26 04.4 & 1460 , 822 , 717 & 5 , 1.6 , 1.3 & cxou j140414.3 + 542604 ; xmm-3 & edge sa ? & 12 + m101 xmm4 & 14 02 28.5 & + 54 16 26.7 & 1505 , 877 , 757 & 5.1 , 1.7 , 1.5 & cxou j140228.3 + 541626 & sa & 12 + m101 xmm5 & 14 02 22.5 & + 54 17 58 & 516 , 245 , 289 & 1.8 , 0.4 , 0.6 & cxou j140222.2 + 541756 ; xmm-6 & sa & 12 + ngc5408 xmm1 & 14 03 19.8 & -41 22 59.3 & 5932 , 2036 , 2077 & 12.8 , 3.2 , 3.3 & ngc5408 [ kcp2003 ] x-1 & sa & 5 + circinus xmm1 & 14 12 54.2 & -65 22 55.3 & 16220 , 11452 , - & 14.5 , 11.3 , - & - & edge sa ? & - + circinus xmm2 & 14 12 39.2 & -65 23 34.3 & 8741 , 2386 , - & 5.7 , 1.9 , - & - & edge sa ? & - + circinus xmm3 & 14 13 28.3 & -65 18 08.3 & 4873 , 1031 , - & 1.8 , 0.7 , - & - & edge sa ? & - + ngc 247 & 0 46 51.7 & -20 43 30 & qso b044 - 2059 + ngc 300 & 0 55 26.7 & -37 31 25.6 & hd 5403 ( star ) + ngc 625 & 01 34 42.4 & -41 36 15.2 & qso b0132 - 4151 + ngc 1569 & 04 31 16.9 & + 64 49 50 & cxou j043116.8 + 644950 ( star ) + ngc 1569 & 04 31 14.2 & + 64 51 07.9 & cxou 043114.0 + 645107 ( star ) + ngc 1569 & 04 31 25.4 & + 64 51 53.8 & cxou 043125.1 + 645154 ( agn ) + ngc 1705 & 04 54 01.2 & -53 21 12.3 & wga j0454.0 - 5320 ( m star or elliptical galaxy ) + ngc 2403 & 07 35 09 & + 65 40 27.5 & hd 59581 ( star ) + ngc 4258 & 12 18 08.9 & + 47 16 08.3 & qso j1218 + 472 + m83 & 13 36 45.6 & -29 59 13.9 & 2masx j13364579 - 2959122 ( galaxy ) + m83 & 13 36 13.9 & -29 56 13 & rx j133615 - 2957.8 ( galaxy ) + ngc 5253 & 13 39 50.6 & -31 34 11.1 & cd-30 10790 ( star ) + m101 & 14 02 30 & + 54 21 18.2 & [ wip99 ] h13 ( star ) + ngc 5408 & 14 03 27.5 & -41 25 18.5 & ( star ) + ngc247 xmm1 & @xmath503 & @xmath504 & 112.2/95 & 1900 & 2200 + ngc253 xmm1 & @xmath505 & @xmath506 & 262.6/232 & 3.1 & 3.6 + & @xmath507 & @xmath508 & 611.6/582 & 2.9 & 3.3 + ngc253 xmm2 ( obs 2 ) & @xmath509 & @xmath510 & 507.4/500 & 1.2 & 1.4 + ngc253 xmm3 & @xmath511 & @xmath512 & 91.9/83 & 0.73 & 1.2 + & @xmath513 & @xmath514 & 381.8/409 & 0.98 & 1.6 + ngc253 xmm4 & @xmath515 & @xmath516 & 73.6/59 & 0.52 & 0.85 + & @xmath144 & @xmath517 & 321.4/293 & 0.29 & 0.48 + ngc253 xmm5 & @xmath518 & @xmath519 & 31.8/25 & 0.32 & 0.53 + & @xmath520 & @xmath521 & 283.8/298 & 1.1 & 1.3 + ngc253 xmm6 & @xmath522 & @xmath523 & 435/409 & 0.93 & 1.5 + ngc253 xmm7 & @xmath524 & @xmath525 & 357/342 & 1.3 & 2.2 + ngc300 xmm1 & @xmath526 & @xmath527 & 469.8/422 & 0.81 & 0.6 + ngc300 xmm2 & @xmath528 & @xmath529 & 133.98/99 & 0.27 & 0.21 + ngc300 xmm3 & @xmath530 & @xmath531 & 101.9/81 & 0.19 & 0.15 + ngc300 xmm5 & @xmath532 & @xmath533 & 54.2/56 & 0.14 & 0.10 + ngc300 xmm6 & @xmath534 & @xmath535 & 47.6/37 & 0.06 & 0.15 + ngc1313 xmm1 & @xmath166 & @xmath536 & 219.8/203 & 0.42 & 0.88 + ngc1313 xmm2 & @xmath537 & @xmath538 & 464.1/421 & 1.9 & 4.0 + ngc1313 xmm3 & @xmath539 & @xmath540 & 778.3/426 & 3.3 & 6.9 + ic0342 xmm3 & @xmath541 & @xmath542 & 185.8/109 & 1.7 & 3.1 + ngc1705 xmm1 & 0.3 & @xmath543 & 61.9/88 & 0.12 & 0.37 + ngc1705 xmm2 & @xmath544 & @xmath545 & 91/76 & 0.078 & 0.24 + ngc1705 xmm3 & @xmath546 & @xmath547 & 80.9/67 & 0.17 & 0.53 + ngc2403 xmm1 & @xmath548 & @xmath549 & 92.2/81 & 2.2 & 3.6 + ngc2403 xmm2 & @xmath550 & @xmath551 & 179.5/151 & 1.3 & 2.0 + ngc2403 xmm3 & @xmath552 & @xmath553 & 92.6/107 & 0.81 & 1.3 + holmii xmm1 ( obs 1 ) & @xmath554 & @xmath555 & 1134.2/976 & 12 & 10 + holm i xmm1 & @xmath534 & @xmath556 & 102.8/95 & 0.48 & 1.7 + m81 xmm1 & @xmath557 & @xmath558 & 1849.9/1245 & 4.5 & 7.0 + & @xmath559 & @xmath560 & 224.9/208 & 4.3 & 6.7 + m81 xmm2 & 7.3 & 6.13 & 1358.2/618 & 48.5 & 75.2 + m81 xmm3 & @xmath561 & @xmath562 & 81.35/80 & 2.5 & 3.9 + m81 xmm4 & @xmath534 & @xmath563 & 66.4/52 & 0.35 & 0.54 + m81 xmm5 & @xmath564 & @xmath565 & 97.5/82 & 0.44 & 0.68 + holm ix xmm1 & @xmath566 & @xmath567 & 1000.9/882 & 9.4 & 15 + sextans a xmm1 & @xmath568 & @xmath569 & 271.4/275 & 0.56 & 0.13 + ngc4214 xmm2 & @xmath570 & @xmath571 & 50.9/46 & 0.16 & 0.14 + ngc4258 xmm1 & @xmath572 & @xmath573 & 101.4/78 & 0.06 & 0.04 + ngc4258 xmm2 ( obs 1 ) & @xmath574 & @xmath575 & 97.5/63 & 0.43 & 2.6 + ngc4395 xmm1 & @xmath576 & @xmath577 & 195.1/156 & 7.2 & 14 + ngc4395 xmm3 & @xmath534 & @xmath578 & 55.9/58 & 0.25 & 0.48 + ngc4449 xmm1 & @xmath579 & @xmath580 & 103/118 & 1.2 & 1.36 + ngc4449 xmm3 & @xmath581 & @xmath582 & 154/89 & 1.3 & 1.5 + ngc4490 xmm1 & @xmath583 & @xmath584 & 101.5/65 & 1.2 & 8.7 + ngc4490 xmm2 & @xmath585 & @xmath586 & 49.5/56 & 0.92 & 6.7 + ngc4490 xmm3 & @xmath587 & @xmath588 & 76.7/80 & 1.5 & 11 + ngc4631 xmm1 & @xmath589 & @xmath590 & 383.4/347 & 0.76 & 5.1 + ngc4631 xmm2 & @xmath591 & @xmath592 & 119.5/99 & 0.23 & 1.5 + ngc4631 xmm3 & @xmath593 & @xmath594 & 146/98 & 0.15 & 1.0 + ngc4736 xmm1 & @xmath595 & @xmath596 & 62.8/53 & 0.36 & 0.80 + ngc4945 xmm1 & @xmath597 & @xmath598 & 116/122 & 0.9 & 1.0 + ngc4945 xmm2 & @xmath599 & @xmath600 & 114.5/115 & 0.71 & 0.82 + ngc4945 xmm4 & @xmath601 & @xmath602 & 75.5/62 & 0.49 & 0.56 + ngc5204 xmm1 & @xmath603 & @xmath604 & 592.1/561 & 2.0 & 5.5 + & @xmath605 & @xmath606 & 533/498 & 3.0 & 8.3 + m51 xmm1 & @xmath607 & @xmath608 & 110.5/82 & 0.34 & 2.8 + m51 xmm2 & @xmath609 & @xmath610 & 75.2/70 & 0.52 & 3.3 + m51 xmm5 & 2.7 & 3.08 & 256.0/72 & 0.43 & 2.7 + m51 xmm6 & @xmath611 & @xmath612 & 40.97/43 & 0.13 & 0.83 + m51 xmm7 & @xmath613 & @xmath614 & 37.8/31 & 0.11 & 0.66 + m83 xmm1 & @xmath615 & @xmath616 & 210.9/211 & 0.64 & 2.8 + m83 xmm4 & @xmath617 & @xmath618 & 91.4/91 & 0.4 & 1.8 + m101 xmm1 & @xmath619 & @xmath620 & 303/233 & 0.45 & 2.9 + m101 xmm2 & @xmath621 & @xmath622 & 288.8/263 & 0.81 & 5.3 + m101 xmm3 & @xmath623 & @xmath624 & 148.9/133 & 0.51 & 3.4 + m101 xmm4 & @xmath625 & @xmath626 & 165.7/140 & 0.38 & 2.5 + m101 xmm5 & @xmath627 & @xmath628 & 47.9/46 & 0.12 & 0.8 + ngc5408 xmm1 & @xmath629 & @xmath630 & 396.8/339 & 7.04 & 19.4 + circinus xmm1 & @xmath631 & @xmath632 & 762.9/863 & 4.6 & 8.8 + circinus xmm2 & @xmath633 & @xmath634 & 517.9/432 & 2.7 & 5.2 + circinus xmm3 & @xmath635 & @xmath636 & 285.2/262 & 0.32 & 0.61 + llllll ngc 253 xmm2 & @xmath637 & @xmath638 & @xmath639 & 464/498 & 1.47 + ngc 2403 xmm1 & @xmath640 & @xmath641 & @xmath642 & 82.8/85 & 1.4 + ngc 4490 xmm1 & @xmath643 & @xmath644 & @xmath645 & 66.5/64 & 0.66 + ngc 4490 xmm2 & @xmath646 & @xmath647 & @xmath648 & 45.7/55 & 0.67 + m101 xmm2 & @xmath649 & @xmath650 & @xmath651 & 256/262 & 0.65 + m101 xmm3 & @xmath652 & @xmath653 & @xmath654 & 128/132 & 0.41 + circinus xmm2 & @xmath655 & @xmath656 & @xmath657 & 437.2/430 & 0.5 + circinus xmm3 & @xmath658 & @xmath659 & @xmath660 & 273.1/261 & 0.17 + ngc247 & 7.93 & 27.32 & 0.602 & 0.687 & 1 + ngc253 & 998.73 & 1861.67 & 55.92 & 93.10 & 3 + ngc300 & 23.08 & 74.45 & 1.688 & 1.324 & 0 + ngc625 & 5.09 & 9.08 & 0.280 & 0.230 &
0 + ngc1313 & 35.97 & 92.00 & 2.329 & 4.845 & 2 + ic0342 & 255.96 & 661.68 & 16.66 & 30.32 & 3 + ngc1569 & 45.41 & 47.29 & 2.072 & 0.635 & 0 + ngc1705 & 0.970 & 2.580 & 0.064 & 0.199 & 0 + mrk 71 & 3.51 & 4.67 & 0.173 & 0.239 & 1 + ngc2403 & 51.55 & 148.49 & 3.547 & 5.378 & 2 + holmberg ii & 1.15 & 2.62 & 0.070 & 0.061 & 1 + holmberg i & & & & & 1 + m81 & 44.73 & 174.02 & 3.647 & 5.655 & 1 + m82 & 1271.32 & 1351.09 & 58.35 & 106.2 & 1 + holmberg ix & & & & & 1 + sextans a & 0.255 & 0.674 & 0.017 & 0.004 & 0 + ic 2574 & 2.41 & 10.62 & 0.212 & 0.329 & 0 + ngc 4214 & 17.87 & 29.04 & 0.947 & 0.826 & 1 + ngc 4258 & 21.60 & 78.39 & 1.690 & 10.48 & 1 + ngc4395 & 4.21 & 12.90 & 0.299 & 0.573 & 1 + ngc4449 & 37.00 & 58.28 & 1.937 & 2.199 & 1 + ngc4490 & 47.79 & 85.94 & 2.636 & 19.19 & 5 + ngc4631 & 82.90 & 208.66 & 5.324 & 35.83 & 1 + ngc4736 & 62.41 & 135.34 & 3.734 & 8.261 & 4 + ngc4945 & 588.11 & 1415.5 & 36.95 & 42.49 & 0 + ngc 5204 & 2.33 & 5.35 & 0.143 & 0.395 & 2 + m51 & 108.68 & 292.08 & 7.213 & 44.74 & 5 + m83 & 266.03 & 638.63 & 16.69 & 76.79 & 0 + ngc5253 & 30.00 & 30.92 & 1.365 & 16.72 & 0 + m101 & 88.04 & 252.84 & 6.048 & 39.63 & 4 + ngc5408 & 2.825 & 2.958 & 0.129 & 0.356 & 1 + circinus & 248.7 & 315.85 & 12.06 & 23.10 & 4 + | we present the results of an archival _ xmm - newton _ study of the bright x - ray point sources ( l@xmath0ergs@xmath1 ) in 32 nearby galaxies . from our list of approximately 100 point sources ,
we attempt to determine if there is a low - state counterpart to the ultraluminous x - ray ( ulx ) population , searching for a soft - hard state dichotomy similar to that known for galactic x - ray binaries and testing the specific predictions of the imbh hypothesis . to this end , we searched for low - state " objects , which we defined as objects within our sample which had a spectrum well fit by a simple absorbed power law , and
high - state " objects , which we defined as objects better fit by a combined blackbody and a power law . assuming that `` low - state '' objects accrete at approximately 10% of the eddington luminosity @xcite and that high - state " objects accrete near the eddington luminosity we further divided our sample of sources into low and high state ulx sources .
we classify 16 sources as low - state ulxs and 26 objects as high - state ulxs . as in galactic black hole systems , the spectral indices , @xmath2 , of the low - state objects , as well as the luminosities ,
tend to be lower than those of the high - state objects .
the observed range of blackbody temperatures for the high state is 0.1 - 1kev , with the most luminous systems tending toward the lowest temperatures .
we therefore divide our high - state ulxs into candidate imbhs ( with blackbody temperatures of approximately 0.1kev ) and candidate stellar mass bhs ( with blackbody temperatures of approximately 1.0kev ) .
a subset of the candidate stellar mass bhs have spectra that are well - fit by a comptonization model , a property similar of galactic bhs radiating in the very - high " state near the eddington limit . |
let @xmath0 be a symplectic manifold without boundary , and @xmath2 a compact connected lie group with lie algebra @xmath3 .
we fix a hamiltonian action of @xmath2 on @xmath4 and an ( equivariant ) momentum map @xmath5 . throughout this memoir
, we make the following standing assumption : + * hypothesis ( h ) : * _ @xmath2 acts freely on @xmath6 and the momentum map @xmath7 is proper . _
+ then the symplectic quotient @xmath8 is well - defined , smooth and closed ( i.e. , compact and without boundary ) .
the _ kirwan map _ is a canonical ring homomorphism @xmath9 here @xmath10 and @xmath11 denote cohomology and equivariant cohomology with rational coefficients , and the product structures are the cup products .
f. kirwan proved @xcite that this map is surjective .
based on this result , the cohomology ring @xmath12 was described in different ways by l. c. jeffrey and f. kirwan ( * ? ? ?
* theorem 8.1 ) , s. tolman and j. weitsman ( * ? ? ? * theorem 1 ) , and many others .
the present memoir is concerned with the problem of `` quantizing '' the kirwan map , which was first investigated by r. gaio and d. a. salamon .
assuming symplectic asphericity and some other restrictive conditions , in ( * ? ? ?
* corollary a ) these authors constructed a ring homomorphism from @xmath13 to the ( small ) quantum cohomology of @xmath14 , which intertwines the gromov - witten invariants of the symplectic quotient with the symplectic vortex invariants .
their result is based on an adiabatic limit in the symplectic vortex equations .
it was used by k. cieliebak and d. a. salamon in ( * ? ? ?
* theorem 1.3 ) to prove that given a monotone linear symplectic torus action on @xmath15 with minimal first equivariant chern number at least 2 , the quantum cohomology of @xmath14 is isomorphic to the batyrev ring .
the result by gaio and salamon motivates the following conjecture .
we denote by @xmath16 the equivariant quantum cohomology . by this
we mean the @xmath17-vector space of all maps @xmath18 satisfying an equivariant version of the novikov condition , together with a product counting holomorphic maps from @xmath19 to the fibers of the borel construction for the action of @xmath2 on @xmath4 .
the novikov condition states that for every number @xmath20 there are only finitely many classes @xmath21 , such that @xmath22 and @xmath23,b\big{\rangle}\leq c$ ] . here
h_g^2(m)$ ] denotes the cohomology class of the two - cocycle @xmath25 in the cartan model .
the space @xmath16 is naturally a module over the equivariant novikov ring @xmath26 .
satisfying an equivariant version of the novikov condition , analogous to the one above .
the product is given by convolution .
] we denote by @xmath27 the quantum cohomology of @xmath14 with coefficients in this ring .
a map @xmath28 satisfying the equivariant novikov condition , induces a @xmath26-module homomorphism @xmath29 where the sum is over all pairs @xmath30 satisfying @xmath31 .
we denote by @xmath32 the first @xmath2-equivariant chern class of @xmath33 , and by @xmath34 the minimal equivariant chern number .
we call @xmath35 _ semipositive _ iff there exists a constant @xmath36 such that @xmath37,b\big{\rangle}=c\big{\langle}c_1^g(m,{\omega}),b\big{\rangle},\ ] ] for every spherical class @xmath38 , and if @xmath39 then @xmath40 .
[ conj : qk ] assume that ( h ) holds and that @xmath35 is convex at @xmath41-compatible and @xmath2-invariant almost complex structure @xmath42 on @xmath4 , such that the quadruple @xmath43 is convex at @xmath41 in the sense explained before theorem [ thm : bubb ] below . ] and semipositive .
then there exists a map @xmath44 as in ( [ eq : phi ] ) , satisfying the equivariant novikov condition , such that the induced map @xmath45 as in ( [ eq : phi * qh ] ) is a surjective ring homomorphism , and @xmath46,b\big{\rangle}\leq0,\,b\neq0{\,\longrightarrow\,}{\varphi}(b)=0 . & \ ] ] once proven , this conjecture will give rise to a recursion formula for @xmath27 in terms of @xmath16 and @xmath44.,\sum_{i=1}^kb_i\right{\rangle}\,\bigg|\,k\in{\mathbb{n}}\cup\{0\},\,b_i\in h_2^g(m,{\mathbb{z}}):\,{\varphi}(b_i)\neq0\textrm { or } { \operatorname{gw}}^g_{b_i}\neq0,\,i=1,\ldots , k\right\},\ ] ] where @xmath47 denotes the 3-point genus 0 equivariant gromov - witten invariant of @xmath0 in the class @xmath48 . ] as noticed in @xcite , without the semipositivity condition , the conjecture likely needs to be modified as follows : [ conj : qk ] assume that ( h ) holds , and that @xmath35 is convex at @xmath41 .
then there exists a morphism of cohomological field theories ( cohft s ) from the equivariant gromov - witten theory of @xmath0 to the gromov - witten theory of @xmath14 . for the notion of a morphism between two cohft
s @xmath49 and @xmath50 see @xcite .
such a morphism consists of a sequence of @xmath51-invariant multilinear maps @xmath52 satisfying relations involving the composition maps of @xmath49 and @xmath50 . here
@xmath53 denotes the moduli space of stable @xmath54-marked scaled lines .
furthermore , the action of the symmetric group @xmath51 is by permutations of the first @xmath54 arguments of @xmath55 . the map @xmath56 plays the role of @xmath45 as in conjecture [ conj : qk ] .
the map @xmath57 measures how much @xmath56 fails to be a ring homomorphism .
once proven , conjecture [ conj : qk ] will give rise to a recursion formula for the gromov - witten invariants of @xmath14 in terms of the equivariant gromov - witten invariants of @xmath0 and the morphism @xmath58 .
the present memoir is part of a project whose goal is to prove conjectures [ conj : qk ] and [ conj : qk ] .
the approach pursued here was suggested by d. a. salamon .
the idea is to construct the maps of the conjectures by counting symplectic vortices over the complex plane @xmath1 . in a first step
, we will consider the ( symplectically ) aspherical case .
this means that @xmath59 in this case the equivariant quantum cup product is induced by the ordinary cup product on @xmath60 .
to explain the idea of the proofs of conjectures [ conj : qk ] and [ conj : qk ] , we recall the symplectic vortex equations : let @xmath42 be an @xmath61-compatible and @xmath2-invariant almost complex structure on @xmath4 , @xmath62 an invariant inner product on @xmath3 , @xmath63 a riemann surface , and @xmath64 a compatible area form on @xmath65 . and
@xmath64 determine the same orientation of @xmath65 .
] for every smooth ( principal ) @xmath2-bundle @xmath66 over @xmath65 we denote by @xmath67 the affine space of smooth connection one - forms on @xmath66 , and by @xmath68 the set of smooth equivariant maps from @xmath66 to @xmath4
. consider the class @xmath69 the _ symplectic vortex equations _ are the equations @xmath70 for a triple @xmath71 .
to explain these conditions , note that the pullback bundle @xmath72 descends to a complex vector bundle @xmath73 .. ] for every @xmath74 we denote by @xmath75 the infinitesimal action , corresponding to the action of @xmath2 on @xmath4 . with this notation , @xmath76 means the complex antilinear part of @xmath77 , which we think of as a one - form on @xmath65 with values in @xmath73 . in ( [ eq : f a mu ] ) we view the curvature @xmath78 of @xmath79 as a two - form on @xmath65 with values in the adjoint bundle @xmath80 .
acts on @xmath3 in the adjoint way . ] furthermore , identifying @xmath81 with @xmath3 via @xmath62 , we view @xmath82 as a section of @xmath83 . the vortex equations ( [ eq
: bar dd j a u],[eq : f a mu ] ) were discovered by k. cieliebak , a. r. gaio , and d. a. salamon @xcite , and independently by i. mundet i riera @xcite .
acting on @xmath84 by multiplication , and @xmath85 , the corresponding energy functional was introduced previously by v. l. ginzburg and l. d. landau @xcite , in order to model superconductivity . more generally , in the case @xmath86 and @xmath2 a closed subgroup of @xmath87 , the functional appeared in physics in the context of _ gauged linear sigma models _ , starting with the work of e. witten @xcite . ] a solution of these equations is called a _ ( symplectic ) vortex_. two elements @xmath88 are called _ ( gauge ) equivalent _ iff there exists an isomorphism @xmath89 of smooth @xmath2-bundles ( which descends to the identity on @xmath65 ) , such that @xmath90 in this case we write @xmath91 .
we define @xmath92 the equations ( [ eq : bar dd j a u],[eq : f a mu ] ) are invariant under equivalence .
we call an element @xmath93 a _ vortex class _ iff it consists of vortices .
we define the _ energy density _ of a class @xmath93 to be @xmath94 where @xmath95 is any representative of @xmath50 . here
the norms are induced by the riemannian metrics @xmath96 on @xmath65 and @xmath97 on @xmath4 , and by @xmath62 .
this definition does not depend on the choice of @xmath98 .
vortex classes are absolute minimizers of the _ ( yang - mills - higgs ) energy functional _
@xmath99,\quad e(w):=\int_{\sigma}e_w{\omega}_{\sigma}\ ] ] in a given second equivariant homology class .
is closed , and vortices in the given homology class exist .
] we define the _ image _ of a class @xmath93 to be the set of orbits of @xmath100 , where @xmath101 is any representative of @xmath50 .
this is a subset of the orbit space @xmath102 .
consider now the complex plane @xmath85 , equipped with the standard area form @xmath103 .
, see remark [ rmk : trivial ] below . ]
let @xmath104 be a vortex class of finite energy , such that the image of @xmath50 has compact closure ] .
then @xmath50 naturally carries an equivariant homology class @xmath105\in h^g_2(m,{\mathbb{z}})$ ] .
( see section [ sec : homology chern ] . )
let @xmath21 .
we denote by @xmath106 the set of vortex classes @xmath50 representing the class @xmath107 , and by @xmath108 a universal @xmath2-bundle .
there are natural evaluation maps @xmath109 at @xmath110 and @xmath111 .. ] we denote by @xmath112 the poincar duality map . to prove conjecture [ conj : qk ] , heuristically , we define @xmath113 for @xmath38 , @xmath114 , and @xmath115 . under the hypotheses of conjecture [ conj : qk ] ,
this map is `` morally '' well - defined and satisfies the conditions of the conjecture : if @xmath42 is chosen as in the definition of convexity below , then there exists a compact subset of @xmath102 containing the image of every finite energy vortex class @xmath104 whose image has compact closure .
this ensures that for every @xmath38 , the space @xmath106 can be compactified by including holomorphic spheres in @xmath116 and in the fibers of the borel construction @xmath117 . in the transverse case
, it follows that the `` boundary '' of @xmath106 has codimension at least 2 .
this makes the map @xmath44 `` well - defined '' .
it satisfies the equivariant novikov condition as a consequence of the compactification argument , conservation of the equivariant homology class in the limit ( see @xcite ) , and the identity @xmath118,[w]\big{\rangle}.\ ] ] this holds for every vortex class @xmath104 of finite energy , such that the image of @xmath50 has compact closure . and a smoothening argument at @xmath41 .
] the identity also implies conditions ( [ eq : phi 0],[eq : om mu b ] ) . the ring homomorphism property for the induced map @xmath119 follows from an argument involving two marked points on the plane @xmath1 that either move together or infinitely apart .
the semipositivity assumption ensures that in the limit there is no bubbling of vortex classes over @xmath1 without marked points .
in contrast with holomorphic planes , such vortex classes may occur in stable maps in top dimensional strata , even in the transverse case .
this is due to the fact that vortices over @xmath1 `` should not be rotated '' , which is explained below .
surjectivity of @xmath120 will be a consequence of surjectivity of the kirwan map @xmath121 , and the equivariant novikov property .
the idea of the proof of conjecture [ conj : qk ] is to define @xmath122 as above , and for general @xmath123 , @xmath124 in a similar way , using @xmath54 marked points .
the map @xmath125 counts vortex classes over @xmath1 without marked points .
the `` quantum kirwan morphism '' @xmath126 will intertwine the genus 0 symplectic vortex invariants with the gromov - witten invariants of @xmath14 .
this will follow from a bubbling argument for a sequence of vortex classes over the sphere @xmath19 , equipped with an area form that converges to @xmath41 .
may bubble off . ]
the goal of the present memoir is to establish bubbling ( i.e. , `` compactification '' ) and fredholm results for vortices over @xmath1 in the aspherical case .
together with a transversality result ( see @xcite ) , the fredholm result will provide a natural structure of an oriented manifold on the set @xmath106 .
furthermore , the bubbling result will imply that the map @xmath127 is a pseudocycle .
this will give a rigorous meaning to the integral ( [ eq : phi b al bar b ] ) .
the ring homomorphism property and the relations defining a morphism of cohft s will be a consequence of the bubbling result and a suitable gluing result .
to explain the first main result of this memoir , we assume that @xmath0 is aspherical , i.e. , condition ( [ eq : int s 2 u om ] ) is satisfied . in section [ sec :
] we denote by @xmath128 the class of all vortices over @xmath1 and the set of equivalence classes of such vortices .
the latter is equipped with a natural topology .-
topology on compact subsets of @xmath1 . ] consider the subspace of all classes in @xmath129 with fixed finite energy @xmath130 .
there are three sources of non - compactness of this space : consider a sequence @xmath131 , @xmath132 , of classes of energy @xmath133 . in the limit
@xmath134 , the following scenarios ( and combinations ) may occur : + * case 1 .
* the energy density of @xmath135 blows up at some point in @xmath1 . + * case 2 . *
there exists a number @xmath136 and a sequence of points @xmath137 that converges to @xmath41 , such that the energy density of @xmath135 on the ball @xmath138 is bounded above and below by some fixed positive constants .
+ * case 3 .
* the energy densities converge to 0 , i.e. , the energy is spread out more and more .
+ in case 1 , by rescaling @xmath135 around the bubbling point , in the limit @xmath134 , we obtain a non - constant @xmath42-holomorphic map from @xmath1 to @xmath4 . using removal of singularity ,
this is excluded by the asphericity condition . in case 2 ,
we pull @xmath135 back by the translation @xmath139 , and in the limit @xmath134 , obtain a vortex class over @xmath1 .
finally , in case 3 , we `` zoom out '' more and more . in the limit @xmath134 and after removing the singularity at @xmath41 , we obtain a pseudo - holomorphic map from @xmath19 to the symplectic quotient @xmath140 . hence in the limit , passing to a subsequence
, we expect @xmath135 to converge to a new sort of stable map , which consists of vortex classes over @xmath1 and pseudo - holomorphic spheres in @xmath116 . here
an important difference to gromov - convergence for pseudo - holomorphic maps is the following : although the vortex equations are invariant under all orientation preserving isometries of @xmath65 , only translations on @xmath1 should be allowed as reparametrizations used to obtain a vortex class over @xmath1 in the limit .
hence we should disregard some symmetries of the equations .
the reasons are that otherwise the reparametrization group does not always act with finite isotropy on the set of simple stable maps , and that there is no suitable evaluation map on the set of vortex classes , which is invariant under rotation . and [ rmk : isom + ] in sections [ sec : conv ] and [ sec : repara ] . ]
we are now ready to formulate the first main result . here
we say that @xmath141 is _ convex at @xmath41_[convex ] iff there exists a proper @xmath2-invariant function @xmath142 and a constant @xmath143 , such that @xmath144 for every @xmath145 and @xmath146 . here
@xmath147 denotes the levi - civita connection of the metric @xmath97 .
this condition reduces to the existence of a plurisubharmonic function in the case in which @xmath2 is trivial .
it is satisfied e.g. if @xmath4 is closed , and for linear actions on symplectic vector spaces .
is proper is used . ]
[ thm : bubb ] assume that hypothesis ( h ) is satisfied , @xmath0 is aspherical , and @xmath141 is convex at @xmath41 .
let @xmath148 , and for @xmath132 let @xmath131 be a vortex class and @xmath149 be points .
suppose that the closure of the image of each @xmath135 is compact , and @xmath150 then there exists a subsequence of @xmath151 that converges to some genus 0 stable map @xmath152 consisting of vortex classes over @xmath1 and pseudo - holomorphic spheres in @xmath116 , with @xmath153 marked points .
( see definitions [ defi : st ] and [ defi : conv ] in chapter [ chap : bubbling ] .
are explained in remark [ rmk : z 0 ] in section [ sec : conv ] . ] ) the proof of this result combines gromov compactness for pseudo - holomorphic maps with uhlenbeck compactness .
it relies on work @xcite by k. cieliebak , r. gaio , i. mundet i riera , and d. a. salamon .
the idea is the following . in
order to capture all the energy , we `` zoom out rapidly '' , i.e. , rescale the vortices so much that the energies of the rescaled vortices are concentrated near the origin in @xmath1 .
now we `` zoom back in '' in such a way that we capture the first bubble , which may either be a vortex class over @xmath1 or a @xmath42-holomorphic sphere in @xmath116 . in the first case we are done . in the second case we `` zoom in '' further , to obtain a finite number of vortices and spheres that are attached to the first bubble .
iterating this procedure , we construct the limit stable map .
the proof involves generalizations of results for pseudo - holomorphic maps to vortices : a bound on the energy density of a vortex , quantization of energy , compactness with bounded derivatives , and hard and soft rescaling .
the proof that the bubbles connect and no energy is lost between them , uses an isoperimetric inequality for the invariant symplectic action functional , proved in @xcite , based on a version of the inequality by r. gaio and d. a. salamon @xcite .
another crucial point is that when `` zooming out '' , no energy is lost locally in @xmath1 in the limit .
this relies on an upper bound on the `` momentum map component '' of a vortex , due to r. gaio and d. a. salamon .
the space of gauge equivalence classes of symplectic vortices can be viewed as the zero set of a section of an infinite dimensional vector bundle .
formally , the second main result of this memoir states that in the case @xmath154 the vertical differential of this section is fredholm when seen as an operator between suitable weighted sobolev spaces .
we will first state the result and then interpret it in this way .
consider the case @xmath85 and @xmath155 .
let @xmath156 and @xmath157 be real numbers .
and @xmath157 always refer to finite values , unless otherwise stated .
] we define the set @xmath158 as follows . for a measurable function @xmath159
we denote @xmath160 $ ] and define the _
@xmath157-weighted @xmath161-norm of @xmath162 _ to be @xmath163.\ ] ] we define @xmath164 to be the class consisting of all triples @xmath101 , where @xmath165 is a @xmath2-bundle of class @xmath166 , is a topological @xmath2-bundle over @xmath1 , equipped with an atlas of local trivializations whose transition functions lie in @xmath166 .
every such bundle is trivializable , but we do not fix a trivialization here . ]
@xmath79 a connection ( one - form ) of class @xmath167 , and @xmath168 a @xmath2-equivariant map of class @xmath167 .
we call two elements
@xmath169 _ @xmath170-equivalent _ iff there exists an isomorphism @xmath89 of @xmath2-bundles of class @xmath166 ( descending to the identity on @xmath1 ) , such that @xmath171 . in this case
we write @xmath172 .
we define @xmath173 where the energy density @xmath174 is defined as in ( [ eq : e w ] ) .
let @xmath175 .
we define normed vector spaces @xmath176 and @xmath177 as follows .
let @xmath133 be a real vector bundle over @xmath1 .
we denote by @xmath178 the bundle of alternating @xmath179-forms on @xmath1 with values in @xmath133 .
if @xmath133 is a complex vector bundle , then we denote by @xmath180 the bundle of anti - linear one - forms on @xmath1 with values in @xmath133 .
we denote by @xmath181 and @xmath182 the spaces of its @xmath183- and @xmath167-sections , respectively .
we fix @xmath184 , and denote by @xmath185 the adjoint bundle and the quotient of the pullback bundle @xmath72 .
let @xmath186 .
we denote @xmath187 $ ] and by @xmath188 the connection on @xmath189 induced by the levi - civita connection @xmath147 of @xmath97 and @xmath79 , ) in appendix [ sec : add ] . ] and we abbreviate @xmath190 .
we define the weighted norm @xmath191.\ ] ] here the norms are taken with respect to @xmath97 , the standard metric on @xmath1 , and @xmath62 .
we denote by @xmath192 the hodge star operator with respect to the standard metric on @xmath1 , and by @xmath193 the formal adjoint of the twisted differential @xmath194 . for @xmath74
we denote by @xmath195 the adjoint map of the infinitesimal action of @xmath2 on @xmath4 at @xmath196 , with respect to the riemannian metric @xmath97 and the inner product @xmath62 .
the collection @xmath197 induces a map @xmath198 from the space of sections of @xmath189 to the space of sections of @xmath83 .
we define is the formal adjoint for the infinitesimal action of the gauge group on the product of the spaces of connections and equivariant maps from @xmath66 to @xmath4.]@xmath199 @xmath200 here the equivalence relations in ( [ eq : x w],[eq : y w ] ) are defined similarly to the @xmath170-equivalence relation on @xmath201 .
since the energy - density is invariant under the gauge transformations , the gauge group of @xmath66 ] of class @xmath166 naturally acts on the set @xmath203 assume that @xmath204 .
then this action is free . in appendix
[ sec : add ] . ] therefore , @xmath176 is naturally a normed vector space , which can canonically be identified with @xmath205 , for any representative @xmath98 of @xmath50 .
similarly , @xmath177 is naturally a normed vector space , which may be identified with @xmath206 , for any representative @xmath98 of @xmath50 .
consider the operator @xmath207:=\left[\begin{array}{c}\big({\nabla}^av+l_u{{\alpha}}\big)^{0,1}-\frac12j({\nabla}_vj)(d_au)^{1,0 } \\
d_a{{\alpha}}+{\omega}_0\ , d\mu(u)v \end{array}\right ] . &
\ ] ] here the brackets @xmath208 $ ] denote equivalence classes . formally , this is the vertical differential of a section of a banach space bundle over a banach manifold , whose zeros are the gauge equivalence classes of vortices .
( for explanations see the next subsection . )
recall that @xmath209 denotes the equivariant first chern class of @xmath0 .
the second main result of this memoir is the following .
[ thm : fredholm ] let @xmath156 , @xmath210 , and @xmath175 ( defined as in ( [ eq : b p lam ] ) ) .
assume that @xmath211 .
then the following statements hold .
[ thm : fredholm : x y ] if @xmath204 then the normed vector spaces @xmath212 and @xmath213 ( defined as in ( [ eq : x w],[eq : y w ] ) ) are complete .
[ thm : fredholm : ddd ] if @xmath214 then the operator @xmath215 ( defined as in ( [ eq : d p lam w],[eq : d w v al ] ) ) is well - defined and fredholm of real index @xmath216\big{\rangle } , \ ] ] where @xmath105 $ ] denotes the equivariant homology class of @xmath50 ( see section [ sec : homology chern ] ) .
the contraction appearing in formula ( [ eq : ind ] ) can be interpreted as a certain maslov index .
( see proposition [ prop : chern maslov ] in section [ sec : homology chern ] . ) the condition @xmath214 in part ( [ thm : fredholm : ddd ] ) of this result captures the geometry of finite energy vortices over @xmath1 .
( see remark [ rmk : geometry ] below . )
the condition @xmath217 is also needed for the map @xmath215 to have the right fredholm index .
( see remark [ rmk : index ] . )
the definition of the space @xmath212 naturally parallels the definition of @xmath158 .
( see remark [ rmk : x ] . )
note that some naive choices for the domain and target of the operator @xmath218 do not work .
( see remark [ rmk : naive ] . )
the proof of theorem [ thm : fredholm ] is based on a fredholm result for the augmented vertical differential and the existence of a bounded right inverse for @xmath219 .
( see theorems [ thm : fredholm aug ] and [ thm : l w * r ] in section [ subsec : reform ] . )
the proof of theorem [ thm : fredholm aug ] has two main ingredients .
the first one is the existence of a suitable complex trivialization of the bundle @xmath220 .
for @xmath221 large , @xmath222 and @xmath223 such a trivialization respects the splitting @xmath224 where @xmath225 denotes the complexified infinitesimal action , for @xmath74 .
the second ingredient are two propositions stating that the standard cauchy - riemann operator @xmath226 and a related matrix differential operator are fredholm maps between suitable weighted sobolev spaces .
these results are based on the analysis of weighted sobolev spaces carried out by r. b. lockhart and r. c. mcowen @xcite .
a crucial analytical ingredient is a hardy - type inequality ( proposition [ prop : hardy ] in appendix [ sec : weighted ] ) . to put the fredholm result into context ,
let @xmath63 be a connected smooth riemann surface , equipped with a compatible area form @xmath64 .
recall the definitions ( [ eq : bb],[eq : b ] ) of @xmath227 . consider the subclass @xmath228 of triples @xmath101 for which there exists a point @xmath229 , such that the action of @xmath2 at the point @xmath230 is free .
we define @xmath231 where @xmath232 is defined as before the definition ( [ eq : b ] ) .
formally , @xmath233 may be viewed as an infinite dimensional manifold , since for every smooth @xmath2-bundle @xmath66 over @xmath65 , the natural action of the gauge group @xmath234 on the `` infinite dimensional manifold '' @xmath235 is free .
furthermore , the set of vortex classes may be viewed as the zero set of a section of a vector bundle @xmath236 over @xmath233 , with infinite dimensional fiber , as follows .
consider the `` vector bundle '' @xmath237 over @xmath238 , whose fiber over a point @xmath239 is given by @xmath240 the bundle @xmath241 over @xmath233 is now defined to be the quotient of the bundle @xmath242 under the natural equivalence relation lifting the relation @xmath232 on @xmath238 .
finally , @xmath243 is defined to be the section induced by @xmath244 heuristically , @xmath236 is an infinite dimensional vector bundle over @xmath233 , and @xmath245 is a smooth section of @xmath236 .
the zero set @xmath246 consists of all vortex classes over @xmath65 .
assume that @xmath247 .
then formally , there is a canonical map @xmath248 , where @xmath249 denotes the fiber over @xmath50 .
we define the vertical differential of @xmath245 at @xmath50 to be the map @xmath250 heuristically , if this map is fredholm and surjective , for every @xmath247 , then the zero set @xmath251 is a smooth submanifold of @xmath233 .
the dimension of a connected component of this submanifold equals the fredholm index of @xmath252 , where @xmath50 is any point in the connected component . at a formal level , in the case @xmath154 , equipped with @xmath253 ,
the vertical differential ( [ eq : d v s w ] ) coincides with the operator @xmath215 , which was defined in ( [ eq : d p lam w],[eq : d w v al ] ) and occurred in the fredholm result , theorem [ thm : fredholm ] . to see this , let @xmath254 .
we interpret @xmath255 as a quotient , as follows .
let @xmath66 be a smooth @xmath2-bundle over @xmath65 , and @xmath256 . denoting @xmath95 ,
the infinitesimal action at the point @xmath257 , corresponding to the action of @xmath258 on @xmath259 , is given by @xmath260 where @xmath261 $ ] .
defining @xmath262 we may identify @xmath263 where @xmath232 denotes the natural lift of the equivalence relation on @xmath238 .
assume formally that @xmath259 and @xmath264 are equipped with a @xmath258-invariant riemannian metric and a @xmath258-invariant inner product , respectively .
for @xmath256 we denote by @xmath265 the adjoint map of @xmath266 .
then by ( [ eq : xx p a u ] ) , we may identify @xmath267 using this and ( [ eq : x w disj],[eq : eee si w ] ) , the vertical differential ( [ eq : d v s w ] ) at @xmath247 agrees with the map @xmath268 given by ( [ eq : d w v al ] ) , in the case @xmath154 and @xmath253 . here on either side , @xmath232 denotes a natural lift of the equivalence relation on @xmath238 .
[ rmk : bundle ] in some earlier work ( e.g. @xcite and @xcite ) , the @xmath2-bundle @xmath66 was fixed and the vortex equations were seen as equations for a pair @xmath257 rather than a triple @xmath101 . the motivation for making @xmath66 part of the data is twofold : when formulating convergence for a sequence of vortex classes over @xmath1 to a stable map , one has to pull back the vortices by translations of @xmath1 .
( see section [ sec : conv ] . )
if the principal bundle is fixed and vortices are defined as pairs @xmath257 solving ( [ eq : bar dd j a u],[eq : f a mu ] ) , then there is no natural such pullback .
however , there _ is _ a natural pullback if the bundle is made part of the data for a vortex .- bundle @xmath66 over @xmath1 , we may of course choose a trivialization of @xmath66 , and then define a pull back for pairs @xmath257 , using the trivialization .
however , this approach is unnatural , since it depends on the choice of a trivialization . ]
more generally , it is possible to pull back vortex triples @xmath101 by a khler transformation of a riemann surface equipped with a compatible area form .
another motivation is the following : if the area form or the complex structure on the surface @xmath65 vary , then in the limit we may obtain a surface @xmath269 with singularities . it does not make sense to consider @xmath66 as a bundle over @xmath269 .
one way of solving this problem is by decomposing @xmath269 into smooth surfaces , and constructing smooth @xmath2-bundles over these surfaces .
hence the @xmath2-bundle should be viewed as a varying object .
once @xmath66 is made part of the data , it is natural to consider _ equivalence classes _ of triples @xmath270 ( as defined in ( [ eq : m ] ) ) , rather than the triples themselves .
one reason is that all important quantities , like energy density and energy , are invariant ( or equivariant ) under equivalence .
note also that the bubbling and fredholm results are more naturally stated for equivalence classes of vortices .
viewing the equivalence classes as the fundamental objects also matches the physical viewpoint that the `` gauge field '' , i.e. , the connection @xmath79 , is physically relevant only `` up to gauge '' .
@xmath271 [ rmk : trivial ] let @xmath65 be the plane @xmath1 , equipped with the standard area form @xmath272 , and consider the trivial @xmath2-bundle @xmath273 . then the solutions @xmath257 of the vortex equations ( [ eq : bar dd j a u],[eq : f a mu ] ) on @xmath274 bijectively correspond to solutions @xmath275 of the equations @xmath276+\mu(f)=0 . & \ ] ] here we denote by @xmath277 and @xmath278 the standard coordinates in @xmath279 , and in the second equation we identify the lie algebra @xmath3 with its dual via the inner product @xmath62 .
the correspondence maps such a triple @xmath280 to @xmath281 , where @xmath79 denotes the connection on @xmath274 defined by @xmath282 and the map @xmath283 is given by @xmath284 .
the group @xmath285 acts on the set @xmath286 of solutions of ( [ eq : dd s f l f phi],[eq : dd s psi ] ) by @xmath287 where we denote the adjoint representation of an element @xmath288 by @xmath289 .
this group naturally corresponds to the gauge group @xmath290 , and its action to the action @xmath291 since every @xmath2-bundle over @xmath1 is trivializable , it follows that the quotient of @xmath286 by the action of @xmath285 bijectively corresponds to the quotient @xmath292 , consisting of gauge equivalence classes of triples @xmath101 of solutions of ( [ eq : bar dd j a u],[eq : f a mu ] ) .
hence the results of the present memoir can alternatively be formulated in terms of solutions of the equations ( [ eq : dd s f l f phi],[eq : dd s psi ] )
. however , the intrinsic approach using equations ( [ eq : bar dd j a u],[eq : f a mu ] ) seems more natural .
@xmath271 [ rmk : asphericity ] without the asphericity condition one needs to include holomorphic spheres in the fibers of the borel construction in the definition of a stable map .
in this situation , to compactify the space of vortices over @xmath1 with an upper energy bound , one needs to combine the proof of theorem [ thm : bubb ] with the analysis carried out by i. mundet i riera and g. tian in @xcite , or by a. ott in @xcite .
@xmath271 [ rmk : quotient ] the space @xmath176 occurring in theorem [ thm : fredholm ] is a quotient of a disjoint union of normed vector spaces .
it is canonically isomorphic to the space @xmath205 , for every representative @xmath98 of @xmath50 .
similar statements hold for @xmath177 .
the description of the spaces @xmath176 and @xmath177 as such quotients may look unconventional , however , it seems natural , since it does not involve any choice of a representative of @xmath50 . alternatively , one could phrase the fredholm result in terms of the spaces @xmath205 and @xmath206 .
however , in view of the last part of remark [ rmk : bundle ] , this seems less natural than the present formulation .
@xmath271 [ rmk : geometry ] the condition @xmath293 in the definition ( [ eq : bb p lam ] ) of @xmath294 and the requirement @xmath214 in theorem [ thm : fredholm]([thm : fredholm : ddd ] ) capture the geometry of finite energy vortices over @xmath1 , in the following sense .
let @xmath295 be a finite energy vortex such that @xmath100 has compact closure .
( here @xmath164 is defined as at the beginning of section [ sec : fredholm ] . )
then for every @xmath296 there exists a constant @xmath297 such that @xmath298 , for every @xmath299 .
this follows from theorem [ thm : reg bundle ] in appendix [ sec : smooth ] and ( * ? ? ?
* corollary 1.4 ) .
it follows that @xmath300 if @xmath217 . this bound is sharp . to see this ,
let @xmath301 and @xmath302 , equipped with the standard symplectic form @xmath303 , complex structure @xmath304 , and the action of the trivial group @xmath305 .
consider the inclusion @xmath306 .
since this map is holomorphic , the triple @xmath307 is a finite energy vortex whose image has compact closure .
it does not lie in @xmath294 .
on the other hand , every @xmath308 has finite energy whenever @xmath156 and @xmath204 .
] where @xmath309 .
the first estimate is hlder s inequality and the second one follows from a calculation in radial coordinates . ]
the latter condition is sharp . to see this ,
consider @xmath310 .
we choose a smooth map @xmath311 , such that @xmath312 then the triple @xmath307 lies in @xmath294 for every @xmath156 and @xmath313 .
however , it has infinite energy .
satisfying @xmath314 for every @xmath156 and @xmath313 , is @xmath315 , where @xmath316 , for every @xmath317 .
however , the closure of the image of such a map @xmath318 is non - compact , since it contains the set @xmath319 .
therefore , @xmath98 does not lie in @xmath294 for any @xmath170 and @xmath157 . ]
@xmath271 [ rmk : index ] the condition @xmath217 in part ( [ thm : fredholm : ddd ] ) of theorem [ thm : fredholm ] is needed for the map @xmath215 to have the right fredholm index .
namely , let @xmath204 be such that @xmath320 , and @xmath175 .
then the proof of theorem [ thm : fredholm ] shows that @xmath215 is fredholm with index equal to @xmath321\big{\rangle},\ ] ] where @xmath322 is the largest integer less than @xmath323 . in particular
, the index changes when @xmath157 passes the value @xmath324 .
note also that the condition @xmath204 is needed , in order for the homology class @xmath105 $ ] to be well - defined .
( see remark [ rmk : w well - defined ] in section [ sec : homology chern ] . )
@xmath271 [ rmk : x ] the definition of the space @xmath212 naturally parallels the definition ( [ eq : b p lam ] ) of @xmath158 .
namely , by linearizing with respect to @xmath79 and @xmath318 the terms @xmath325 and @xmath82 occurring in the energy density @xmath174 , we obtain the terms @xmath326 and @xmath327 .
these expressions occur in @xmath328 ( defined as in ( [ eq : ze w p lam ] ) ) , except for the factor @xmath329 in @xmath327 . in appendix
[ sec : proofs homology ] that this factor is irrelevant . ]
the expression @xmath330 is needed in order to make @xmath331 non - degenerate .
@xmath271 [ rmk : naive ] consider the situation in which the norm ( [ eq : ze w p lam ] ) is replaced by the usual @xmath332-norm , and the norm @xmath333 ( defining @xmath206 ) is replaced by the usual @xmath161-norm .
then in general , the map defined by ( [ eq : d w v al ] ) does not have closed image , and hence it is not fredholm .
note also that the 0-th order terms @xmath334 and @xmath335 in ( [ eq : d w v al ] ) are not compact ( neither with respect to @xmath176 and @xmath177 , nor with respect to the usual @xmath332- and @xmath161-norms ) .
the reason is that the embedding of @xmath336 ( for @xmath156 ) into the space of bounded continuous functions on @xmath1 is not compact .
because of these terms , the map ( [ eq : d w v al ] ) is not well - defined between spaces that look like the standard weighted sobolev spaces in `` logarithmic '' coordinates @xmath337 ( with @xmath338 ) .
this is in contrast with the situation in which @xmath65 is the infinite cylinder @xmath339 , equipped with the standard complex structure and area form .
in that situation the splitting ( [ eq : im l c ] ) is unnecessary , and standard weighted sobolev spaces in `` logarithmic '' coordinates can be used . in the relative setting , with the cylinder replaced by the infinite strip @xmath340 $ ] , this was worked out by u. frauenfelder ( * ? ? ?
* proposition 4.7 ) .
the proof of the fredholm result then relies on results @xcite by j. robbin and d. a. salamon .
@xmath271 [ [ quantum - kirwan - maps ] ] quantum kirwan maps + + + + + + + + + + + + + + + + + + + the history of conjectures [ conj : qk ] and [ conj : qk ] is as follows .
assume that ( h ) holds , @xmath0 is aspherical , @xmath35 is convex at @xmath41 and monotone , and @xmath60 is generated by classes of degrees less than @xmath341 , where @xmath342 is the minimal equivariant chern number ( defined as in ( [ eq : n inf ] ) ) . in this case
r. gaio and d. a. salamon @xcite proved that there exists a ring homomorphism from @xmath60 to the quantum cohomology of @xmath14 that agrees with the kirwan map on classes of degrees less than @xmath341 , see ( * ? ? ?
* corollary a ) .
the idea of the proof of this result is to fix an area form @xmath343 on @xmath19 and to relate symplectic vortices for the area form @xmath344 with pseudo - holomorphic spheres in @xmath116 , for sufficiently large @xmath345 .
the authors noticed that in general , this correspondence does not work , since in the limit @xmath346 , vortices over @xmath1 may bubble off .
accordingly , the calculation of the quantum cohomology of monotone toric manifolds in @xcite , which is based on @xcite , does not extend to the situation of a general toric manifold .
this follows from examples by h. spielberg @xcite .
based on these observations , salamon suggested to construct a ring homomorphism from @xmath13 to the quantum cohomology of @xmath14 , by counting symplectic vortices over the plane @xmath1 , provided that @xmath0 is aspherical and @xmath35 is convex at @xmath41 .
this homomorphism should intertwine the symplectic vortex invariants and the gromov - witten invariants of @xmath14 .
this gave rise to the ph.d .-
thesis @xcite , which served as a basis for the present memoir .
there it is observed that in the definition of convergence for vortices over @xmath1 to a stable map , only translations should be allowed as reparametrizations used to obtain a vortex component in the limit . in section [ sec : conv ] , the reason for this is that the evaluation of a vortex class at a point in @xmath1 is not invariant under rotations . ] c. woodward realized that with this restriction , vortices over @xmath1 without marked points may appear in the bubbling argument used in the proof of the ring homomorphism property of the quantum kirwan map .
( this may happen even in the transverse case . ) as a solution , he suggested to interpret the quantum kirwan map as a morphism of cohomological field theories .
( see @xcite and conjecture [ conj : qk ] above . ) on the other hand , under the semipositivity introduced above , the vortices without marked points can be excluded in the transverse case .
this gave rise to conjecture [ conj : qk ] .
in his recent article @xcite c. woodward developed these ideas in an algebraic geometric setting .
he defined a quantum kirwan map in the case of a smooth projective variety with an action of a complex reductive group .
( see ( * ? ? ?
* theorem 1.3 ) . )
theorem [ thm : bubb ] of the present memoir is used in the proof of that result to show properness of the deligne - mumford stack of stable scaled gauged maps in @xmath0 of genus 0 .
( see ( * ? ? ?
* theorem 5.25 ) . ) in @xcite e. gonzalez and c. woodward used woodward s definition to calculate the quantum cohomology of a compact toric orbifold with projective coarse moduli space .
furthermore , in @xcite they used it to formulate a quantum version of kalkman s wall - crossing formula .
[ [ bubbling - and - fredholm - results - for - vortices ] ] bubbling and fredholm results for vortices + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + assume that @xmath65 is closed , ( h ) holds , and @xmath4 is symplectically aspherical and equivariantly convex at @xmath41 . in this case , in ( * ? ? ? * theorem 3.4 ) , k. cieliebak et al . proved compactness of the space of vortex classes with energy bounded above by a fixed constant . in the case in which @xmath4 and @xmath65 are closed , in ( * ? ? ? * theorem 4.4.2 ) i. mundet i riera compactified the space of bounded energy vortex classes with fixed complex structure on @xmath65 . assuming also that @xmath347 ,
this was extended by i.
mundet i riera and g. tian in ( * ? ? ?
* theorem 1.4 ) to the situation of varying complex structure . that work is based on a version of gromov - compactness for continuous almost complex structures , proved by s. ivashkovich and v. shevchishin in @xcite . in ( * ? ? ?
* theorem 1.8 ) a. ott compactified the space of bounded energy vortex classes in a different way , for a general lie group and closed @xmath4 and @xmath65 , the last with fixed complex structure .
he used the approach to gromov - compactness by d. mcduff and d. a. salamon in @xcite . in the case
in which @xmath65 is an infinite strip , equipped with the standard area form and complex structure , the compactification was carried out in a relative setting by u. frauenfelder in ( * ? ? ?
* theorem 4.12 ) .
( see also @xcite . ) in @xcite r. gaio and d. a. salamon investigated the vortex equations with area form @xmath348 in the limit @xmath346 . here
@xmath65 is a closed surface equipped with a fixed area form @xmath64 .
they proved that three types of objects may bubble off : a holomorphic sphere in @xmath116 , a vortex over @xmath1 , and a holomorphic sphere in @xmath4 .
( see the proof of ( * ? ? ? * theorem a ) . ) for a _ closed _ riemann surface @xmath65 , in @xcite k. cieliebak et al
. proved that the augmented vertical differential of the vortex equations is fredholm .
[ [ other - related - work ] ] other related work + + + + + + + + + + + + + + + + + + in @xcite s. venugopalan and c. woodward establish a hitchin - kobayashi correspondence for symplectic vortices over the plane @xmath1 .
this memoir is organized as follows .
chapter [ chap : bubbling ] contains the bubbling analysis for vortices over the plane @xmath1 . in section [ sec : stable ] we define the notion of a stable map consisting of vortex classes over @xmath1 , pseudo - holomorphic spheres in @xmath116 , and marked points . in section [ sec : conv ] we formulate convergence of a sequence of vortex classes over @xmath1 to such a stable map .
stable maps and convergence are explicitly described in the ginzburg - landau setting in section [ sec : example ] .
section [ sec : repara ] covers an additional topic , which will be relevant in a future definition of the quantum kirwan map .
the main result of section [ sec : comp ] is that given a sequence of rescaled vortices with uniformly bounded energies , there exists a subsequence that converges up to gauge , modulo bubbling at finitely many points .
section [ sec : soft ] contains a result that tells how to find the next bubble in the bubbling tree , at a bubbling point of a given sequence of rescaled vortices . based on these results , the bubbling result , theorem [ thm : bubb ] ,
is proven in section [ sec : proof : thm : bubb ] .
section [ sec : proof : prop : conv s 1 c ] contains the proof of the result characterizing convergence in the ginzburg - landau setting .
chapter [ chap : fredholm ] contains the fredholm theory for vortices over the plane @xmath1 . in section [ sec : homology chern ] the equivariant homology class of an equivalence class of triples @xmath101 is defined , and the contraction appearing in formula ( [ eq : ind ] ) is interpreted as a certain maslov index .
section [ sec : proof : fredholm ] contains the proof of the fredholm result , theorem [ thm : fredholm ] . in the appendix we recollect some auxiliary results , which are used in the proofs of the main results .
this memoir arose from my ph.d .- thesis .
i would like to thank my adviser , dietmar a. salamon , for the inspiring topic .
i highly profited from his mathematical insight .
i am very much indebted to chris woodward for his interest in my work , for sharing his ideas with me , and for his continuous encouragement .
it was he who coined the term `` quantum kirwan map '' .
i would also like to thank urs frauenfelder , kai cieliebak , eduardo gonzalez , andreas ott , and bumsig kim for stimulating discussions .
finally , i am grateful to the referees for their valuable suggestions .
in this chapter , stable maps consisting of vortex classes over the plane @xmath1 , holomorphic spheres in the symplectic quotient , and marked points , are defined , and the first main result of this memoir , theorem [ thm : bubb ] , is proven .
this result states that given a sequence of vortex classes over @xmath1 , with uniformly bounded energies , and sequences of marked points , there exists a subsequence that converges to some stable map .
we also describe stable maps and convergence in the simplest interesting example , the ginzburg - landau setting .
let @xmath349 be as in chapter [ chap : main ] .
our standing hypothesis ( h ) implies that the symplectic quotient @xmath350 is well - defined and closed .
the structure @xmath42 induces an @xmath351-compatible almost complex structure on @xmath116 as follows . for every @xmath74
we denote by @xmath75 the infinitesimal action at @xmath196 .
we define the _ horizontal distribution _ @xmath352 by @xmath353 here @xmath354 denotes the orthogonal complement with respect to the metric @xmath97 on @xmath4 .
we denote by @xmath355 the canonical projection .
we define @xmath356 to be the unique isomorphism of @xmath357 such that @xmath358 we identify @xmath359 with @xmath19 .
the ( connectedness ) condition in the definition of a stable map below will involve the evaluation of a vortex class at the point @xmath360 . in order to make sense of this , we need the following .
we denote by @xmath361 the orbit of a point @xmath74 .
let @xmath66 be a smooth ( principal ) @xmath2-bundle over @xmath1 and @xmath362 a map .
we denote by @xmath102 the orbit space of the action of @xmath2 on @xmath4 , and define @xmath363 where @xmath229 is an arbitrary point in the fiber over @xmath364 . for @xmath93
we define @xmath365 where @xmath366 is any representative of @xmath50 .
this is well - defined , i.e. , does not depend on the choice of @xmath98 .
recall the definition ( [ eq : m ] ) of @xmath129 , and that by the image of a class @xmath367 we mean the set of orbits of @xmath100 , where @xmath101 is any vortex representing @xmath50 .
we define the set @xmath368 [ prop : bar u ] if @xmath369 then the map @xmath370 extends continuously to a map @xmath371 , such that @xmath372 .
this follows from the estimate ( [ eq : sup z z bar d ] ) with @xmath373 in proposition [ prop : en conc ] ( section [ sec : soft ] below ) .
[ defi : barev ] we define the _ evaluation map _
@xmath374 to be the map from the disjoint union of @xmath375 and @xmath376 to @xmath102 , given as follows .
for @xmath377 we define @xmath378 furthermore , for @xmath369 we define @xmath379 where @xmath162 is as in proposition [ prop : bar u ] . by a tree relation on a set @xmath380 we mean a symmetric , anti - reflexive relation on @xmath380 , such that each two points in @xmath380 are connected by a unique simple path of edges .
a central definition of this memoir is the following .
[ defi : st ] for every @xmath381 a _ ( genus 0 ) stable map consisting of vortex classes over @xmath1 and pseudo - holomorphic spheres in @xmath116 , with @xmath153 marked points _ , is a tuple @xmath382 where @xmath383 is a finite set for @xmath384 , @xmath133 is a tree relation on @xmath385 , @xmath386 ( for @xmath387 ) , @xmath388 is a @xmath356-holomorphic map ( for @xmath389 ) , @xmath390 is a point for each edge @xmath391 , @xmath392 is a vertex , and @xmath393 is a point , for @xmath394 , such that the following conditions hold .
[ defi : st comb]*(combinatorics ) * * we have @xmath395 . * for every @xmath396 there exist an integer @xmath397 and vertices @xmath398 such that @xmath399 and for every @xmath400 we have @xmath401 [ defi : st dist]*(special points ) * * if @xmath402 then @xmath403 . * if @xmath387 and @xmath404 are such that @xmath405 then @xmath406 . * fix @xmath396 .
then the points @xmath407 with @xmath408 such that @xmath391 and the points @xmath409 with @xmath394 such that @xmath410 , are all distinct .
[ defi : st conn]*(connectedness ) * let @xmath411 be such that @xmath391 .
then @xmath412 here @xmath374 is defined as in ( [ eq : barev ] ) and ( [ eq : barev w ] ) and we set @xmath413 if @xmath389 .
[ defi : st st]*(stability ) * let @xmath396 . * if @xmath387 and @xmath414 then the set @xmath415 contains at least two points .
if either @xmath389 and @xmath416 , or @xmath417 , then the set ( [ eq : special ] ) consists of at least three points .
this definition is modeled on the notion of a genus 0 pseudo - holomorphic stable map , as introduced by kontsevich in @xcite .
roughly speaking , a stable map in the sense of definition [ defi : st ] can be thought of as a collection of vortex classes over @xmath1 , pseudo - holomorphic spheres in the symplectic quotient @xmath116 , `` ghost spheres of type 0 '' corresponding to the vertices of @xmath418 , and marked and nodal points .
a vortex class may be connected to a sphere in @xmath116 at the nodal point @xmath41 , and to `` ghost spheres of type 0 '' at points in @xmath1 .
furthermore , spheres of the same type may be connected at nodal points .
the `` ghost spheres of type 0 '' should be thought of as constant spheres in the borel construction @xmath117 .
they are needed for the bubbling result ( theorem [ thm : bubb ] ) to capture colliding marked points in @xmath1 .
evaluate to points in @xmath117 at points in @xmath1 .
therefore , identifying each `` ghost sphere of type 0 '' with a point in @xmath117 , it makes sense to ask that the sphere is connected to a vortex class over @xmath1 at a given nodal point . ]
figure [ fig : stable map ] shows an example of a stable map . here
the `` teardrops '' correspond to vortex classes over @xmath1 , the solid and dashed spheres to pseudo - holomorphic spheres in @xmath116 , and the dotted spheres to `` ghost spheres of type 0 '' .
the solid objects have positive energy , and the dashed and dotted spheres are `` ghosts '' , i.e. , their energy vanishes .
each `` teardrop '' is connected to a sphere ( in @xmath116 ) via a nodal point at its vertex , which corresponds to the point @xmath419 . to explain the stability condition ( [ defi : st st ] ) , we fix @xmath396 .
we define the _ set of nodal points on @xmath420 _ to be @xmath421 the set of _ marked points on @xmath420 _ to be @xmath422 and the _ set @xmath423 of special points on @xmath420 _ to be the union of these two sets .
the stable map of figure [ fig : stable map ] carries ten marked points , which are drawn as dots .
the stability condition says the following .
assume that @xmath396 is a `` ghost component '' , i.e. , @xmath417 or @xmath424 carries zero energy ( in the case @xmath425 ) .
then the following holds : if @xmath387 then it carries at least one special point in @xmath1 .. ] furthermore , if @xmath426 then @xmath420 carries at least three special points .
this condition ensures that the action of a natural reparametrization group on the set of _ simple _ stable maps of a given type is free . in section [ sec : repara ] below . ] in a future definition of the quantum kirwan map this will be needed in order to show that the evaluation map on the set of non - trivial vortex classes ( with marked points ) is a pseudo - cycle .
condition ( [ defi : st comb ] ) implies that if @xmath427 is empty then so is @xmath418 , and hence a stable map in the sense of definition [ defi : st ] is a genus 0 stable map of @xmath356-holomorphic spheres in @xmath116 . if @xmath402 then @xmath428 and @xmath429 .
this follows from the second part of condition ( [ defi : st comb ] ) for @xmath389 , using the last condition in ( [ eq : be i be i 1 ] ) . hence in this case a stable map consists of a single vortex class and special points . if @xmath430 then the sets @xmath431 and @xmath432 are subtrees of @xmath380 , and every element of @xmath427 is adjacent to a unique vertex in @xmath431 and to no vertex in @xmath427 .
in particular , each element of @xmath427 is a leaf of the tree @xmath432 .
these statements follow from condition ( [ defi : st comb ] ) and the fact that @xmath380 does not contain any simple cycle .
the vertices in @xmath418 are not adjacent to those in @xmath431 .
furthermore , for each connected component of @xmath418 there exists a unique vertex in @xmath427 that is adjacent to some element of the connected component .
these assertions follow from condition ( [ defi : st comb ] ) and the fact that @xmath380 does not contain any simple cycle .
@xmath271 [ rmk : z i]if @xmath433 is such that @xmath434 then @xmath435 .
this follows from condition ( [ defi : st dist ] ) and the fact that either @xmath402 or every vertex in @xmath427 is adjacent to some vertex in @xmath431 .
@xmath271 [ rmk : general ] without the asphericity assumption , a stable map should also include holomorphic maps from the sphere @xmath19 to the fibers of the borel construction @xmath117 .
these occur if in a sequence of vortices over @xmath1 energy is concentrated around some point in @xmath1 .
the necessary analysis was worked out by i. mundet i riera and g. tian in @xcite , and by a. ott in @xcite .
@xmath271 the simplest example of a stable map consists of the tree with one vertex @xmath436 , a vortex class @xmath437 , the marked point @xmath438 and a finite number of distinct points @xmath439 , @xmath440 , where @xmath441 if @xmath442 .
@xmath271 [ xpl : stable ] we set @xmath443 , choose an integer @xmath444 , and define @xmath445 @xmath446 let @xmath447 , @xmath448 be distinct points , @xmath449 a @xmath356-holomorphic sphere , and @xmath450 be a vortex class , for @xmath448 .
assume that @xmath451 for every @xmath179 , and if @xmath452 then @xmath449 is nonconstant .
then the tuple @xmath453 is a stable map .
( see figure [ fig : example stable ] . )
@xmath271 let @xmath454 , for @xmath132 let @xmath455 be a vortex class and @xmath456 be points , and let @xmath457 be a stable map .
in order to define convergence of @xmath458 to @xmath152 , we need the following notations . for a @xmath356-holomorphic map @xmath459
we denote its energy by @xmath460 let @xmath461 and @xmath394 .
we define @xmath462 as follows .
if @xmath463 then we set @xmath464 otherwise we define @xmath465 to be the first special point encountered on a path of edges from @xmath420 to @xmath466 .
to explain this , we denote by @xmath404 the unique vertex such that the chain of vertices of @xmath380 running from @xmath420 to @xmath466 is given by @xmath467 .
( @xmath468 is also allowed . )
we define @xmath469 let @xmath65 be a compact connected smooth surface with non - empty boundary . recall the definition ( [ eq : b ] ) of the set @xmath470 of equivalence classes of triples @xmath101 .
we define the _ @xmath471-topology @xmath472 _ on this set as follows : we fix a ( smooth ) @xmath2-bundle @xmath66 over @xmath65 , and denote by @xmath258 its gauge group . since by hypothesis , @xmath2 is connected , every @xmath2-bundle over @xmath65 is trivializable .
it follows that the map @xmath473\in{\mathcal{b}}_{\sigma}\ ] ] is a bijection . denotes the affine space of smooth connection one - forms on @xmath66 .
we use the simplified notation @xmath474 $ ] for the equivalence class @xmath475 $ ] . ]
[ defi : tau si ] we define @xmath472 to be the pushforward of the quotient topology of the @xmath471-topology on @xmath476 under the map ( [ eq : a c g ] ) .
let @xmath65 be a smooth surface , @xmath477\in{\mathcal{b}}_{\sigma}$ ] , and @xmath478 an open subset with compact closure and smooth boundary .
we define the _ restriction @xmath479 _ to be the equivalence class of the pullback of @xmath101 under the inclusion map @xmath480 . for @xmath477\in{\mathcal{b}}_{\sigma}$ ] and @xmath44 a translation on @xmath1
, we define the pullback of @xmath50 by @xmath44 to be @xmath481}{\varphi}^*[p , a , u]:=\big[{\varphi}^*p,\phi^*(a , u)\big],\ ] ] where @xmath482 is defined by @xmath483 . has the form @xmath484 , where @xmath485 and @xmath170 lies in the fiber of @xmath66 over @xmath486 .
] we define @xmath487 note that @xmath488 by our standing hypothesis ( h ) .
[ defi : conv ] the sequence @xmath489 is said to converge to @xmath152 , as @xmath134 , iff the limit @xmath490 exists , @xmath491 and there exist mbius transformations @xmath492 , for @xmath396 , @xmath132 , such that the following conditions hold .
[ defi : conv phi z ] * if @xmath387 then @xmath493 is a translation on @xmath1 . * for every @xmath389 we have @xmath494 , where @xmath495 is defined as in ( [ eq : z al i ] ) , ( [ eq : z al i ] ) .
* let @xmath389 and @xmath496 be a mbius transformation such that @xmath497 .
then the derivatives @xmath498 converge to @xmath41 , for every @xmath485 .
[ defi : conv al be ] if @xmath499 are such that @xmath391 then @xmath500 , uniformly on compact subsets of @xmath501 .
[ defi : conv w ] * let @xmath387 and @xmath502 be an open connected subset with compact closure and smooth boundary . then the restriction @xmath503 converges to @xmath504 , with respect to the topology @xmath505 ( as in definition [ defi : tau si ] ) .
* fix @xmath389 .
let @xmath506 be a compact subset of @xmath507 .
for @xmath508 large enough , we have @xmath509 and @xmath510 converges to @xmath511 in @xmath512 on @xmath506 .
( here @xmath513 and @xmath514 are defined as in ( [ eq : z al],[eq : bar u w ] ) . )
[ defi : conv z ] we have @xmath515 for every @xmath440 .
the meaning of this definition is illustrated by figure [ fig : convergence ] .
it is based on the notion of convergence of a sequence of pseudo - holomorphic spheres to a genus 0 pseudo - holomorphic stable map .
an example in which it can be understood more explicitly , is discussed in the next section . + the condition in the first part of ( [ defi : conv w ] ) , that @xmath516 with respect to @xmath505 , is equivalent to the requirement that there exist representatives @xmath517 of @xmath503 ( for @xmath132 ) and @xmath98 of @xmath504 such that @xmath517 converges to @xmath98 in the @xmath518-topology , for every @xmath397 .
this follows from a straight - forward argument , using lemma [ le : x g ] ( appendix [ sec : add ] ) . @xmath271
the last part of condition ( [ defi : conv phi z ] ) and the second part of condition ( [ defi : conv w ] ) capture the idea of catching a pseudo - holomorphic sphere in @xmath116 by `` zooming out '' : fix @xmath389 , and consider the case @xmath519 .
then there exist @xmath520 and @xmath521 such that @xmath522 .
it follows from a direct calculation that @xmath523 is a vortex class with respect to the area form @xmath524 , where @xmath272 denotes the standard area form on @xmath1 .
the last part of condition ( [ defi : conv phi z ] ) means that @xmath525 , for @xmath134 .
hence in the limit @xmath134 we obtain the equations @xmath526 these correspond to the @xmath356-cauchy - riemann equations for a map from @xmath1 to @xmath116 .
( see proposition [ prop : bar del j ] in appendix [ sec : add ] . )
the second part of ( [ defi : conv w ] ) imposes that in fact the sequence of rescaled vortex classes converges ( in a suitable sense ) to a @xmath356-holomorpic sphere and that this sphere equals @xmath511 .
it is unclear whether the bubbling result , theorem [ thm : bubb ] , remains true if we replace the @xmath512-convergence in this part of condition ( [ defi : conv w ] ) by @xmath471-convergence .
( compare to remark [ rmk : c infty ] in section [ sec : comp ] . ) @xmath271 the `` energy - conservation '' condition ( [ eq : e v bar t ] ) has the important consequence that the stable map @xmath152 represents the same equivariant homology class as the vortex class @xmath135 , for @xmath508 large enough .
( see ( * ? ? ?
* proposition 5.4 ) and @xcite . ) @xmath271 [ rmk : z 0]the purpose of the additional marked point @xmath527 is to be able to formulate the second part of condition ( [ defi : conv w ] ) . for @xmath389 and @xmath132 the map @xmath528
is only defined on the subset @xmath529 . since by condition ( [ defi : conv phi z ] ) we have @xmath494 , the composition @xmath530 is well - defined for each compact subset @xmath531 .
hence the second part of condition ( [ defi : conv w ] ) makes sense . as another motivation ,
note that the bubbling result , theorem [ thm : bubb ] , is in general wrong , if we do not introduce the additional marked points @xmath532 and @xmath527 .
see example [ xpl : conv ] in section [ sec : example ] below .
@xmath271 [ rmk : conv rot]one conceptual difficulty in defining the notion of convergence is the following . consider the group @xmath533 of orientation preserving isometries of @xmath65 ( with respect to the metric @xmath96 ) . that preserve the pair @xmath534 . ]
this group acts on @xmath470 ( defined as in ( [ eq : b ] ) ) , as in ( [ eq : phi [ ] ] ) .
the set @xmath535 of finite energy vortex classes is invariant under this action .
hence naively , in the definition of convergence , for @xmath387 one would allow @xmath493 to be an orientation preserving isometry of @xmath1 , rather than just a translation .
the problem is that with this modification , there is no evaluation map on the set of stable maps , that is continuous with respect to convergence .
such a map is needed for the definition of the quantum kirwan map .
note here that we can not define evaluation of a vortex class @xmath50 at some point @xmath536 by choosing a representative @xmath101 of @xmath50 and evaluating @xmath318 at some point in the fiber over @xmath364 , thus obtaining a point in @xmath4 , since this point depends on the choices .
instead , @xmath50 evaluates at @xmath364 to a point in the borel construction for the action of @xmath2 on @xmath4 .
another reason for allowing only translations as reparametrizations ( for @xmath387 ) is that the action of @xmath533 on the set of vortex classes with positive energy is not always free .
( see example [ xpl : rot ] in section [ sec : repara ] below . )
this means that the action of the reparametrization group on the set of simple stable maps is not always free , if we allow reparametrizations in @xmath533 .
( compare to section [ sec : repara ] . )
@xmath271 recall the definition ( [ eq : me ] ) of the set @xmath535 of finite energy vortex classes over the plane @xmath154 , whose image has compact closure . in this section
we describe this set , stable maps , and convergence of a sequence of vortex classes to a stable map , in the simplest interesting example : let @xmath537 .
we equip @xmath538 with the standard inner product , and consider the action of @xmath539 on @xmath540 by multiplication of complex numbers .
we define a momentum map @xmath541 for this action by @xmath542 in this setting , the energy functional ( [ eq : e b ] ) was introduced by v. l. ginzburg and l. d. landau @xcite , in order to model superconductivity . following the book @xcite by a. jaffe and c. taubes ,
the set @xmath535 can be described as follows .
let @xmath543 ( defined as in ( [ eq : bb ] ) ) and recall the definition ( [ eq : e w ] ) of the energy density @xmath174 .
we denote by @xmath544 the energy of @xmath98 . in the present situation
the condition on the image in the definition of @xmath535 is superfluous .
this means that if @xmath98 solves the vortex equations ( [ eq : bar dd j a u],[eq : f a mu ] ) and has finite energy , then the closure of the image of @xmath318 is compact . to see this
, we fix such a solution .
if the energy of @xmath98 vanishes , then @xmath545 , therefore the image of @xmath318 equals @xmath546 and is thus compact .
hence the statement is a consequence of the following result .
[ prop : s 1 c image ] if the energy of @xmath98 is positive ( and finite ) then the image of @xmath318 is contained in the open unit ball @xmath547 .
an elementary calculation shows that @xmath257 solves the euler - lagrange equations corresponding to the energy functional @xmath548 $ ] .
therefore , it follows from ( * ? ? ?
iii , theorem 8.1 ) that @xmath549 for every @xmath229 .
this proves proposition [ prop : s 1 c image ] .
in fact , under the hypothesis of this result the image of @xmath318 equals @xmath550 , see corollary [ cor : s 1 c image ] below .
( however , this fact will not be used . )
consider @xmath369 .
we define the _ local degree map _ @xmath551 as follows .
we choose a representative @xmath101 of @xmath50 .
since @xmath552 , it follows from ( * ? ? ? * chapter iii , theorem 2.2 ) that for every smooth section @xmath553 , the map @xmath554 has only finitely many zeros .
let @xmath485 .
we define @xmath555 where @xmath553 is any smooth section and @xmath296 is so small that the closed ball @xmath556 intersects @xmath557 only in the point @xmath364 . here
@xmath558 denotes the circle of radius @xmath559 , centered at @xmath364 .
this definition does not depend on the choice of @xmath560 nor @xmath559 .
we now define the map ( [ eq : deg w ] ) by @xmath561 by an elementary argument the right hand side is independent of the choice of the representative @xmath101 of @xmath50 , and hence @xmath562 is well - defined .
furthermore , it follows from ( * ? ? ?
* theorem 2.2 ) that @xmath563 takes on nonnegative values .
we define the _
( total ) degree of @xmath50 _ to be @xmath564 ( only finitely many terms in this sum are non - zero , hence the sum makes sense . )
this number is proportional to the energy of @xmath50 : [ prop : e d ] we have @xmath565 we choose a representative @xmath95 of @xmath50 and a smooth section @xmath553 as in proposition [ prop : e int u om ] in appendix [ sec : vort ] .
we denote @xmath566 .
let @xmath567 be so large that @xmath568 if @xmath569 .
we denote by @xmath570 the radial coordinate in @xmath571 and by @xmath572 the standard angular one - form on @xmath573 . over any circle centered at the origin . ]
the one - form @xmath574 is a primitive of @xmath272 , and therefore , by stokes theorem , @xmath575 by elementary arguments , we have @xmath576 ( here in the last equality we used the assumption that @xmath568 if @xmath569 . )
it follows that @xmath577 on the other hand , using the estimates @xmath578 and @xmath579 , lemma [ le : a priori ] in appendix [ sec : vort ] implies that @xmath580 converges to 0 , uniformly in @xmath581 , as @xmath582 . combining this with ( [ eq : pi deg w],[eq : int b r u om],[eq : int b r e ] ) , equality ( [ eq : e d ] ) follows .
this proves proposition [ prop : e d ] .
this result has the following consequence .
[ cor : s 1 c image ] let @xmath95 be a smooth vortex over @xmath1 with positive and finite energy .
then the image of @xmath318 contains the open unit ball @xmath547 .
consider the set @xmath583 this set is connected , and hence an interval . since @xmath578 , for every @xmath584 there exists a point @xmath229 such that @xmath585 .
on the other hand , positivity of the energy and proposition [ prop : e d ] imply that @xmath318 vanishes somewhere .
it follows that @xmath586 contains the interval @xmath587 .
since the image of @xmath318 is invariant under the @xmath588-action , it follows that it contains the ball @xmath550 .
this proves corollary [ cor : s 1 c image ] .
proposition [ prop : s 1 c image ] and corollary [ cor : s 1 c image ] imply that the image of @xmath318 equals @xmath550 , for every smooth vortex over @xmath1 with positive and finite energy .
fix now @xmath589 .
we denote by @xmath590 the @xmath591-fold symmetric product . by definition
this is the quotient topological space for the action of the symmetric group @xmath592 on @xmath593 given by @xmath594 we identify @xmath590 with the set @xmath595 of all maps @xmath596 such that @xmath597 for only finitely many points @xmath485 , and @xmath598 by assigning to @xmath599\in{\operatorname{sym}}^d({\mathbb c})$ ] the _ multiplicity _
map @xmath600 , given by @xmath601 we can now characterize vortex classes with energy @xmath602 as follows .
[ prop : class ] the map @xmath603 is a bijection .
this follows from ( * ? ? ?
iii , theorem 1.1 ) . as a consequence of proposition [ prop : class ] ,
we obtain a classification of stable maps in the sense of definition [ defi : st ] in the present setting : here the symplectic quotient @xmath604 consists of a single point , the orbit @xmath605 .
hence every holomorphic sphere in @xmath116 is constant .
stable maps are thus classified in terms of their combinatorial structure @xmath606 , the location of the special points , and for each @xmath387 , a point in some symmetric product of @xmath1 .
convergence to a stable map is explicitly described by the following result . here
we will use the inclusion @xmath607):=\big[z_1,\ldots , z_{d'},\infty,\ldots,\infty\big],&\end{aligned}\ ] ] where we identify @xmath608 .
we drop the constant maps to the symplectic quotient from the notation for a stable map , since no information gets lost .
[ prop : conv s 1 c ] let @xmath148 , for @xmath132 let @xmath455 be a vortex class and @xmath149 be points , and let @xmath609 be a stable map .
then the following conditions are equivalent .
[ prop : conv s 1 c w ] the sequence @xmath151 converges to @xmath152 . [ prop : conv s 1 c deg ] for large enough @xmath508 we have @xmath610 furthermore , there exist mbius transformations @xmath492 for @xmath396 and @xmath132 such that conditions ( [ defi : conv phi z],[defi : conv al be],[defi : conv z ] ) of definition [ defi : conv ] are satisfied , and for every @xmath387 the point in the symmetric product @xmath611 converges to @xmath612 , as @xmath134 .
the proof of proposition [ prop : conv s 1 c ] is based on proposition [ prop : cpt mod ] ( compactness modulo bubbling and gauge ) in section [ sec : comp ] below , and will be given in section [ sec : proof : prop : conv s 1 c ] .
[ xpl : conv ] let @xmath443 , and for @xmath132 let @xmath613 be the unique vortex satisfying @xmath614 let @xmath615 be the unique vortex satisfying @xmath616 and @xmath617 be the unique vortex satisfying @xmath618 then @xmath619 converges to the stable map consisting of the sets @xmath620 , @xmath621 , @xmath622 , the tree relation @xmath623 , the vortices @xmath624 and @xmath625 , the unique constant @xmath356-holomorphic sphere @xmath449 in @xmath116 , the nodal points @xmath626 , @xmath627 , @xmath628 , and the marked point @xmath629 .
this follows from proposition [ prop : conv s 1 c ] .
it follows from this example that in general , the additional marked points @xmath532 are needed in the bubbling result , theorem [ thm : bubb ] . without these points , no subsequence of @xmath135 as above converges to a stable map .. ] this follows from an elementary argument .
@xmath271 this section covers an additional topic , which will not be used in this memoir , but will be relevant in a future definition of the quantum kirwan map .
namely , we introduce a natural group of reparametrizations and show that this group acts freely on the set of simple stable maps consisting of vortex classes over @xmath1 and pseudo - holomorphic spheres in the symplectic quotient .
the relevance of this result is the following .
for the definition of the quantum kirwan map it will be necessary to show that a certain natural evaluation map on the set of vortex classes over @xmath1 ( see ( * ? ? ?
* proposition 6.1 ) and @xcite ) is a pseudo - cycle .
this will rely on the fact that its omega limit set has codimension at least two . in order to show this
, one needs to cut down the dimension of each `` boundary stratum '' by dividing by the action of the reparametrization group .
heuristically , the freeness of the action of this group implies that the quotient is a smooth manifold , hence providing a meaning to this procedure .
we fix finite sets @xmath630 and a tree relation @xmath133 on the disjoint union @xmath385 .
we define the _ reparametrization group _ @xmath631 as follows .
we define @xmath632 to be the subgroup of all automorphisms @xmath162 of the tree @xmath633 , satisfying @xmath634 , for @xmath384 .
we denote by @xmath635 the group of mbius transformations and by @xmath636 the group of translations of the plane @xmath1 .
we define @xmath637 we denote by @xmath638 the set of collections @xmath639 , such that @xmath640 , for every @xmath396 .
the group @xmath641 acts on @xmath638 by @xmath642 [ defi : g t ] we define @xmath643 to be the semi - direct product of @xmath641 and @xmath638 induced by this action .
the group @xmath635 acts on the set of @xmath356-holomorphic maps @xmath644 by @xmath645 furthermore , the group @xmath636 acts on the set @xmath535 as in ( [ eq : phi [ ] ] ) . by the _ combinatorial type _ of a stable map @xmath152 as in ( [ eq : w z ] ) we mean the tuple @xmath606 .
we denote by @xmath646 the set of all _ stable maps of ( combinatorial ) type @xmath380_. @xmath631 acts on @xmath647 as follows . for every @xmath648 and @xmath649
we define @xmath650 ( here we set @xmath413 if @xmath389 .
furthermore , for @xmath651 we set @xmath652 . )
[ defi : action g t ] we define @xmath653 this defines an action of @xmath631 on @xmath647 .
let now @xmath654 be an almost complex manifold . recall that a @xmath42-holomorphic map @xmath655 is called _ multiply covered
_ iff there exists a holomorphic map @xmath656 of degree at least two , and a @xmath42-holomorphic map @xmath657 , such that @xmath658 .
otherwise , @xmath318 is called _
simple_. we call a stable map @xmath152 _ simple _ iff the following conditions hold : for every @xmath389 the @xmath356-holomorphic map @xmath511 is constant or simple . furthermore , if @xmath659 are such that @xmath660 and @xmath661 , and @xmath651 , then @xmath662 .
moreover , if @xmath663 are such that @xmath660 and @xmath511 is nonconstant , and if @xmath664 , then @xmath665 . we denote by @xmath666 the subset of all _ simple stable maps_. the action of @xmath631 on @xmath647 leaves @xmath667 invariant . [ prop : simple ] the action of @xmath631 on @xmath667 is free .
the proof of this result uses the following lemma .
[ le : trans ] the action of @xmath636 on @xmath535 is free . assume that @xmath50 is a smooth vortex class over @xmath1 and @xmath668 is a vector , such that defining @xmath669 by @xmath670 , we have @xmath671
. then @xmath672 for every @xmath485 and @xmath673 .
it follows that @xmath674 , @xmath675 , or @xmath676 .
lemma [ le : trans ] follows from this .
this follows from an elementary argument , using the stability condition ( [ defi : st st ] ) , lemma [ le : trans ] , and the fact that every simple holomorphic sphere is somewhere injective ( see ( * ? ? ?
* proposition 2.5.1 ) ) .
[ rmk : isom + ] the action of @xmath636 on @xmath535 extends to an action of the group @xmath677 of orientation preserving isometries of @xmath1 .
hence one may be tempted to adjust the definition of the reparametrization group @xmath631 and its action on @xmath667 accordingly . however , for the purpose of defining the quantum kirwan map , this is not possible .
the reason is that in general there is no continuous evaluation map on @xmath535 that is invariant under the action of @xmath677 . by definition
such an evaluation map @xmath678 assigns a point in the borel construction @xmath117 to each pair @xmath679 , where @xmath50 is a vortex class and @xmath485 , in such a way that @xmath680 . here
@xmath681 is defined as in ( [ eq : bar u w ] ) and @xmath682 denotes the canonical projection .
hence when studying the vortex equations over @xmath1 we need to restrict our attention to a subgroup of the group of global symmetries of the equations .
this is a crucial difference between vortices and pseudo - holomorphic curves .
note also that the action of @xmath677 on the set of vortex classes of positive energy is not always free , as the next example shows .
@xmath271 [ xpl : rot ] consider the action of @xmath683 on @xmath84 by multiplication .
let @xmath589 be an integer .
by proposition [ prop : class ] there exists a unique finite energy vortex class @xmath50 over @xmath1 such that @xmath684 for every rotation @xmath685 we have @xmath686 where @xmath687 acts in a natural way on @xmath590 .
thus the action of @xmath677 on the set of vortex classes of positive energy is not free .
@xmath271 in this section we formulate and prove a crucial ingredient ( proposition [ prop : cpt mod ] ) of the proof of theorem [ thm : bubb ] , which states the following .
consider a sequence of rescaled vortices over @xmath1 with image in a fixed compact subset of @xmath4 and uniformly bounded energies .
we assume that @xmath0 is aspherical .
then there exists a subsequence that away from finitely many bubbling points and up to regauging , converges to a rescaled vortex over @xmath1 .
if the rescalings converge to @xmath41 , then the limit object corresponds to a @xmath356-holomorphic sphere in @xmath116 .
the proof of this result is based on compactness for rescaled vortices over the punctured plane with uniformly bounded energy densities ( proposition [ prop : cpt bdd ] below ) .
it also uses the fact that at each bubbling point at least the energy @xmath688 is lost , which is the minimal energy of a vortex over @xmath1 or pseudo - holomorphic sphere in @xmath116 .
this is the content of proposition [ prop : quant en loss ] below , which is proved by a hard rescaling argument , using proposition [ prop : cpt bdd ] and hofer s lemma .
another ingredient of the proof of proposition [ prop : cpt mod ] is lemma [ le : conv e ] below , which says that the energy densities of a convergent sequence of rescaled vortices converge to the density of the limit . in order to explain the main result of this section ,
let @xmath349 , @xmath689 , and @xmath64 be as in chapter [ chap : main ] .
we fix a triple @xmath690 ( defined as in ( [ eq : bb ] ) ) .
recall the definition ( [ eq : e w ] ) of the energy density @xmath691 .
[ rmk : trafo e vort ] this density has the following transformation property : let @xmath269 be another surface , and @xmath692 a smooth immersion .
consider the pullback @xmath693 where the bundle isomorphism @xmath482 is defined by @xmath483 .
a straight - forward calculation shows that @xmath694 note also that @xmath98 is a vortex with respect to @xmath534 if and only if @xmath695 is a vortex with respect to @xmath696 .
@xmath271 [ rmk : e vort ] if @xmath98 is a vortex ( with respect to @xmath534 ) then @xmath697 where @xmath698 denotes the complex linear part of @xmath699 , viewed as a one - form on @xmath65 with values in the complex vector bundle @xmath73 .. ] this follows from the vortex equations ( [ eq : bar dd j a u],[eq : f a mu ] ) .
@xmath271 let @xmath700 $ ] and @xmath701 .
consider first the case @xmath702 .
then we define the _ @xmath221-energy density of @xmath98 _ to be @xmath703 this means that @xmath704 where the subscript `` @xmath64 '' means that the norms are taken with respect to the metric @xmath96 .
if @xmath705 or @xmath41 then we define @xmath706 we define the _ @xmath221-energy of @xmath98 _ on a measurable subset @xmath707 to be @xmath708.\ ] ] consider the case @xmath709 , equipped with the standard area form @xmath272 .
assume that @xmath702 , and consider the map @xmath669 defined by @xmath710
. then equality ( [ eq : e phi * ] ) implies that @xmath711 hence in the present setting the @xmath221-energy transforms via @xmath712 the _ ( symplectic ) @xmath221-vortex equations _ are the equations ( [ eq : bar dd j a u],[eq : f a mu ] ) with @xmath64 replaced by @xmath713 , i.e. , the equations @xmath714 in the case @xmath373 we interpret the equation ( [ eq : f a r ] ) as @xmath715 we call a solution @xmath716 of equations ( [ eq : r dd j a u],[eq : f a r ] ) an _ @xmath221-vortex _ over @xmath65 .
consider the case @xmath709 , equipped with @xmath272 , and let @xmath702 .
we define the map @xmath669 by @xmath710 .
then @xmath717 is a vortex if and only if @xmath695 is an @xmath221-vortex .
the rescaled energy density has the following important property .
let @xmath718 be a sequence that converges to some @xmath719 $ ] , and for @xmath720 let @xmath721 be an @xmath722-vortex .
if on compact sets @xmath723 converges to @xmath724 in @xmath725 and @xmath726 converges to @xmath727 in @xmath512 then @xmath728 in @xmath725 on compact sets .
( see lemma [ le : conv e ] below . ) in the proof of theorem [ thm : bubb ] , this will be used in order to show that locally on @xmath1 no energy is lost in the limit @xmath134 .
@xmath271 we define the _ minimal energy @xmath729 _ as follows .
recall from ( [ eq : m ] ) that @xmath730 denotes the class consisting of smooth vortices , and that the energy of a @xmath356-holomorphic map @xmath459 is given by @xmath731 .
we define.]@xmath732 @xmath733 @xmath734 assume that @xmath4 is equivariantly convex at @xmath41 .
then corollary [ cor : quant ] in appendix [ sec : vort ] implies that @xmath735 .
furthermore , our standing assumption ( h ) implies that @xmath116 is closed .
it follows that @xmath736 .
hence the number @xmath729 is positive .
the infima ( [ eq : e v ] ) and ( [ eq : emin ] ) are attained , and hence the name `` minimal energy '' for @xmath729 is justified .
( this fact is not used anywhere in this memoir . ) that ( [ eq : emin ] ) is attained follows from the fact that for every @xmath20 there exist only finitely many homotopy classes @xmath737 with @xmath738,{{\overline{b}}}{\rangle}\leq c$ ] that can be represented by a @xmath739-holomorphic map @xmath644 . that ( [ eq : e v ] ) is attained
follows from the fact that for every @xmath20 there exist only finitely many homology classes @xmath21 with @xmath23,b\big{\rangle}\leq c$ ] that can be represented by a finite energy vortex whose image has compact closure .
this is a consequence of theorem [ thm : bubb ] and ( * ? ? ?
* proposition 5.4 ) ( conservation of equivariant homology class ) .
@xmath271 the results of this and the next section are formulated for connections and maps of sobolev regularity .
this is a natural setup for the relevant analysis .
furthermore , we restrict our attention to the trivial bundle @xmath740 .
is trivializable , this suffices for the proof of the main result . ]
we fix @xmath156 , unless otherwise stated . ] and naturally identify the affine space of connections on @xmath740 of local sobolev class @xmath167 with the space of one - forms on @xmath65 with values in @xmath3 , of class @xmath167 .
furthermore , we identify the space of @xmath2-equivariant maps from @xmath740 to @xmath4 of class @xmath167 with @xmath741 . finally , we identify the gauge group on @xmath740 of class @xmath166 with @xmath742 .
we denote @xmath743 @xmath744 the gauge group @xmath742 acts on @xmath745 by @xmath746 where @xmath747 denotes the adjoint action of an element @xmath748 . for @xmath136
we denote by @xmath749 the open ball of radius @xmath570 , around 0 .
[ prop : cpt mod ] assume that @xmath0 is aspherical .
let @xmath718 be a sequence that converges to some @xmath750 $ ] , @xmath751 a sequence that converges to @xmath41 , and for every @xmath132 let @xmath752 be an @xmath722-vortex ( with respect to @xmath753 ) .
assume that there exists a compact subset @xmath754 such that @xmath755 , for every @xmath508 .
suppose also that @xmath756 then there exist a finite subset @xmath757 , an @xmath758-vortex @xmath759 , and , passing to some subsequence , gauge transformations @xmath760 , such that the following conditions hold .
[ prop : cpt mod : < ] if @xmath761 then @xmath762 and the sequence @xmath763 converges to @xmath764 in @xmath471 on every compact subset of @xmath1 .
[ prop : cpt mod:= ] if @xmath765 then on every compact subset of @xmath766 , the sequence @xmath767 converges to @xmath724 in @xmath725 , and the sequence @xmath768 converges to @xmath727 in @xmath512 . [
prop : cpt mod lim nu e eps ] fix a point @xmath769 and a number @xmath770 so small that @xmath771 .
then for every @xmath772 the limit @xmath773 exists and @xmath774 furthermore , the function @xmath775 is continuous
. convergence in conditions ( [ prop : cpt mod:<],[prop : cpt mod:= ] ) should be understood as convergence of the subsequence labelled by those indices @xmath508 for which @xmath776 contains the given compact set .
@xmath271 this proposition will be proved on page .
the strategy of the proof is the following .
assume that the energy densities @xmath777 are uniformly bounded on every compact subset of @xmath1 .
then the statement of proposition [ prop : cpt mod ] with @xmath762 follows from an argument involving uhlenbeck compactness , an estimate for @xmath778 , elliptic bootstrapping ( for statement ( [ prop : cpt mod : < ] ) ) , and a patching argument .
if the densities are not uniformly bounded then we rescale the maps @xmath517 by zooming in near a bubbling point @xmath779 in a `` hard way '' , to obtain a positive energy @xmath780-vortex in the limit , with @xmath781 .
if @xmath761 then @xmath782 , and we obtain a @xmath42-holomorphic sphere in @xmath4 .
this contradicts symplectic asphericity , and thus this case is impossible .
if @xmath765 then either @xmath783 or @xmath784 , and hence either a vortex over @xmath1 or a pseudo - holomorphic sphere in @xmath116 bubbles off . therefore , at least the energy @xmath729 is lost at @xmath779 .
our assumption that the energies of @xmath517 are uniformly bounded implies that there can only be finitely many bubbling points . on the complement of these points
a subsequence of @xmath517 converges modulo gauge .
the bubbling part of this argument is captured by proposition [ prop : quant en loss ] below , whereas the convergence part is the content of the following result .
[ prop : cpt bdd ] let @xmath757 be a finite subset , @xmath785 be a sequence of numbers that converges to some @xmath786 $ ] , @xmath787 open subsets such that @xmath788 , and for @xmath132 let @xmath789 be an @xmath722-vortex .
assume that there exists a compact subset @xmath754 such that for @xmath508 large enough @xmath790 suppose also that for every compact subset @xmath791 , we have @xmath792 then there exists an @xmath758-vortex @xmath759 , and passing to some subsequence , there exist gauge transformations @xmath760 , such that the following conditions are satisfied .
[ prop : cpt bdd < infty ] if @xmath761 then @xmath793 converges to @xmath764 in @xmath471 on every compact subset of @xmath766 .
[ prop : cpt bdd = infty ] if @xmath765 then on every compact subset of @xmath766 , @xmath767 converges to @xmath724 in @xmath725 , and @xmath768 converges to @xmath727 in @xmath512 . the proof of this result is an adaption of the argument of step 5 in the proof of theorem a in the paper by r. gaio and d. a. salamon @xcite .
the proof of statement ( [ prop : cpt bdd < infty ] ) is based on a compactness result for the case of a compact surface @xmath65 ( possibly with boundary ) .
( see theorem [ thm : cpt cpt ] in appendix [ sec : vort ] .
that result follows from an argument by k. cieliebak et al . in @xcite . )
the proof also involves a patching argument for gauge transformations , which are defined on an exhausting sequence of subsets of @xmath766 . to prove statement ( [ prop : cpt bdd = infty ] )
, we will show that curvatures of the connections @xmath723 are uniformly bounded in @xmath332 .
this uses the second rescaled vortex equations and a uniform upper bound on @xmath794 ( lemma [ le : mu u ] in appendix [ sec : vort ] ) , due to r. gaio and d. a. salamon .
the statement then follows from uhlenbeck compactness with compact base , compactness for @xmath778 , and a patching argument .
we may assume w.l.o.g . that there exists a @xmath2-invariant compact subset @xmath754 satisfying ( [ eq : u nu om k ] ) .
( to see this , we choose a compact subset @xmath795 satisfying this condition and consider the set @xmath796 . )
we choose @xmath797 so big that the balls @xmath798 , @xmath769 , are disjoint and contained in @xmath799 . for every @xmath800
we define @xmath801 we prove * statement ( [ prop : cpt bdd < infty])*. assume that @xmath761 . using the hypotheses ( [ eq : u nu om k],[eq : sup e ] ) , it follows from theorem [ thm : cpt cpt ] in appendix [ sec : vort ] ( with @xmath802 ) that there exist an infinite subset @xmath803 and gauge transformations @xmath804 ( @xmath805 ) , such that @xmath806 and @xmath807 is smooth , for every @xmath805 , and the sequence @xmath808 converges to some @xmath758-vortex @xmath809 , in @xmath471 on @xmath810 . iterating this argument , for every @xmath811 there
exist an infinite subset @xmath812 and gauge transformations @xmath813 ( @xmath814 ) , such that @xmath815 and @xmath816 is smooth , for every @xmath814 , and the sequence @xmath817 converges to some @xmath758-vortex @xmath818 , in @xmath471 on @xmath819 .
let @xmath820 .
for every @xmath814 we define @xmath821 .
we have @xmath822 .
furthermore , the sequences @xmath823 and @xmath824 are bounded in @xmath825 on @xmath819 , for every @xmath397 . hence it follows from lemma [ le : g smooth ] ( appendix [ sec : add ] ) that the sequence @xmath826 is bounded in @xmath825 on @xmath819 , for every @xmath397 .
hence , using morrey s embedding theorem and the arzel - ascoli theorem , it has a subsequence that converges to some gauge transformation @xmath827 , in @xmath471 on @xmath819 .
note that @xmath828 we choose a map @xmath829 such that @xmath830 on @xmath831 .
we define @xmath832 , and recursively , @xmath833 using ( [ eq : h i w i ] ) and the fact @xmath834 on @xmath835 , we have , for every @xmath811 , @xmath836 it follows that there exists a unique @xmath837 that restricts to @xmath838 on @xmath819 , for every @xmath820 .
let @xmath820 .
we choose @xmath839 such that @xmath840 and a map @xmath841 that is the identity on @xmath831 .
we define @xmath842 the sequence @xmath843 converges to @xmath98 , in @xmath471 on every compact subset of @xmath766 .
( here we use the @xmath471-convergence on @xmath819 of @xmath817 to @xmath844 and the facts @xmath845 and @xmath846 . )
statement ( [ prop : cpt bdd < infty ] ) follows .
we prove * statement ( [ prop : cpt bdd = infty])*. assume that @xmath765 .
[ claim : kappa nu p ] for every compact subset @xmath791 we have @xmath847 let @xmath848 be an open subset containing @xmath506 such that @xmath849 is compact and contained in @xmath766 . hypothesis ( [ eq : sup e ] )
implies that @xmath850 it follows from our standing hypothesis ( h ) that there exists @xmath851 such that @xmath2 acts freely on @xmath852 since @xmath7 is proper the set @xmath795 is compact .
recall that @xmath75 denotes the infinitesimal action at @xmath196 .
it follows that @xmath853 using the second @xmath722-vortex equation , we have @xmath854 hence by hypothesis ( [ eq : sup e ] ) and the assumption @xmath855 , we have @xmath856 , for @xmath508 large enough . using ( [ eq : sup nu vert],[eq : sup xi ] ) , lemma [ le : mu u ] in appendix [ sec : vort ] implies that @xmath857 estimate ( [ eq : sup kappa nu ] ) follows from this and the second @xmath722-vortex equation .
this proves claim [ claim : kappa nu p ] .
using claim [ claim : kappa nu p ] , theorem [ thm : uhlenbeck compact ] ( uhlenbeck compactness ) in appendix [ sec : add ] implies that there exist an infinite subset @xmath803 and , for each @xmath805 , a gauge transformation @xmath804 , such that @xmath806 and the sequence @xmath858 converges to some @xmath332-connection @xmath859 over @xmath810 , weakly in @xmath332 on @xmath810 . by morrey s embedding theorem and the arzel - ascoli theorem , shrinking @xmath860
, we may assume that @xmath861 converges ( strongly ) in @xmath725 on @xmath810 .
iterating this argument , for every @xmath811 there exist an infinite subset @xmath812 and gauge transformations @xmath862 , for @xmath814 , such that @xmath815 , for every @xmath814 , and the sequence @xmath863 converges to some @xmath332-connection @xmath864 over @xmath819 , weakly in @xmath332 and in @xmath725 on @xmath819 .
let @xmath820 .
for @xmath814 we define @xmath821 . an argument as in the proof of statement ( [ prop : cpt bdd < infty ] ) , using lemma [ le : g smooth ] ( appendix [ sec : add ] ) , implies that the sequence @xmath865 has a subsequence that converges to some gauge transformation @xmath866 , weakly in @xmath867 on @xmath819 . repeating the construction in the proof of statement ( [ prop : cpt bdd < infty ] ) and using the weak @xmath332- and strong @xmath725-convergence of @xmath868 on @xmath819
, we obtain an index @xmath869 and a gauge transformation @xmath870 , for every @xmath820 , such that @xmath839 and @xmath871 converges to some @xmath332-connection @xmath79 over @xmath766 , weakly in @xmath332 and in @xmath725 on every compact subset of @xmath766 . passing to the subsequence @xmath872
, we may assume w.l.o.g .
that @xmath723 converges to @xmath79 , weakly in @xmath332 and in @xmath725 on every compact subset of @xmath766 .
[ claim : hyp : prop : compactness delbar ] the hypotheses of proposition [ prop : compactness delbar ] ( appendix [ sec : add ] ) with @xmath873 are satisfied .
let @xmath874 be an open subset with compact closure , and @xmath875 be such that @xmath876 .
since the sequence @xmath877 converges to @xmath79 , weakly in @xmath878 , we have @xmath879 * condition ( [ eq : u nu k ] ) * is satisfied by assumption ( [ eq : u nu om k ] ) .
we check * condition ( [ eq : du nu ] ) * : we denote by @xmath880 the area of @xmath881 and choose a constant @xmath345 such that @xmath882 , for every @xmath883 and @xmath884 . for @xmath885
, we have @xmath886 here the second inequality uses the hypothesis ( [ eq : u nu om k ] ) .
combining this with ( [ eq : sup e ] ) and ( [ eq : a nu w 1 p ] ) , condition ( [ eq : du nu ] ) follows .
* condition ( [ eq : sup k p ] ) * follows from the first vortex equation , ( [ eq : a nu w 1 p ] ) , ( [ eq : du nu ] ) , and hypothesis ( [ eq : u nu om k ] ) .
this proves claim [ claim : hyp : prop : compactness delbar ] . by claim [ claim : hyp :
prop : compactness delbar ] , we may apply proposition [ prop : compactness delbar ] , to conclude that , passing to some subsequence , @xmath726 converges to some map @xmath887 , weakly in @xmath867 and in @xmath512 on every compact subset of @xmath766 .
the pair @xmath888 solves the first vortex equation .
furthermore , multiplying the second @xmath722-vortex equation with @xmath889 , it follows that @xmath890 .
this means that @xmath98 is an @xmath41-vortex .
[ claim : g a u smooth ] there exists a gauge transformation @xmath891 such that @xmath892 is smooth . since @xmath766 continuously deformation retracts onto a wedge of circles , and @xmath2
is connected , there exists a continuous lift @xmath893 of @xmath894 . by proposition [ prop :
bar del j ] in appendix [ sec : add ] the map @xmath895 is @xmath356-holomorphic . hence it is smooth .
it follows that we may approximate @xmath896 by some smooth lift @xmath897 of @xmath894 .
we define @xmath898 to be the unique solution of @xmath899 , for every @xmath900 .
since the infinitesimal action at every @xmath901 is injective , the equation @xmath902 and smoothness of @xmath903 imply that @xmath904 is smooth .
this proves claim [ claim : g a u smooth ] .
we choose @xmath905 as in claim [ claim : g a u smooth ] .
regauging @xmath723 by @xmath905 , statement ( [ prop : cpt bdd = infty ] ) follows .
this completes the proof of proposition [ prop : cpt bdd ] .
one can try to circumvent the patching argument for the gauge transformations in this proof by choosing an extension @xmath906 of @xmath907 to @xmath766 , and defining @xmath908 .
however , the sequence @xmath909 does not have the required properties , since @xmath793 does not necessarily converge on compact subsets of @xmath766 .
the reason is that for @xmath910 the transformation @xmath911 does not need to restrict to @xmath907 on @xmath819 .
@xmath271 [ rmk : c infty ] it is not clear if in the case @xmath765 the @xmath912 s can be chosen in such a way that @xmath793 converges in @xmath471 on every compact subset of @xmath766 . to prove this ,
one approach is to fix an open subset of @xmath1 with smooth boundary and compact closure , which is contained in @xmath766 .
we can now try mimic the proof of ( * ? ? ?
* theorem 3.2 ) . in step 3 of that proof the first and second vortex equations ( and relative coulomb gauge )
are used iteratively in an alternating way .
this iteration fails in our setting , because of the factor @xmath913 in the second vortex equations , which converges to @xmath41 by assumption .
@xmath271 the second ingredient of the proof of proposition [ prop : cpt mod ] says that if the energy densities of a sequence of rescaled vortices are not uniformly bounded on some compact subset @xmath506 , then at least the energy @xmath729 is lost in the limit , at some point in @xmath506 . here
@xmath729 is defined as in ( [ eq : emin ] ) .
[ prop : quant en loss ] assume that @xmath0 is aspherical .
let @xmath848 be an open subset , @xmath914 a sequence such that @xmath915 , and @xmath916 an @xmath722-vortex , for @xmath132 .
assume that there exists a compact subset @xmath754 such that @xmath917 for every @xmath508 and that @xmath918 .
then the following conditions hold . [
prop : hard r e ] for every compact subset @xmath919 we have @xmath920 [ prop : hard limsup ] if there exists a compact subset @xmath919 such that @xmath921 then there exists @xmath922 with the following property . for every @xmath296 so small that @xmath923 we have @xmath924 the proof of proposition [ prop : quant en loss ] is built on a bubbling argument as in step 5 in the proof of ( * ? ? ?
* theorem a ) .
the idea is that under the assumption of ( [ prop : hard limsup ] ) we may construct either a @xmath356-holomorphic sphere in @xmath116 or a vortex over @xmath1 , by rescaling the sequence @xmath517 in a `` hard way '' .
this means that after rescaling the energy densities are bounded .
we need the following two lemmata .
[ lemma : hofer ] let @xmath925 be a metric space , @xmath926 a continuous function , @xmath927 , and @xmath851 .
assume that the closed ball @xmath928 is complete .
then there exists @xmath929 and a number @xmath930 such that @xmath931 see ( * ? ? ?
* lemma 4.6.4 ) .
the next lemma ensures that for a suitably convergent sequence of rescaled vortices in the limit @xmath134 no energy gets lost on any compact set .
apart from proposition [ prop : quant en loss ] , it will also be used in the proofs of propositions [ prop : cpt mod ] and [ prop : soft ] , and theorem [ thm : bubb ] .
[ le : conv e ] let @xmath932 be a surface without boundary , equipped with an area form and a compatible complex structure , @xmath933 , @xmath132 , a sequence of numbers that converges to some @xmath719 $ ] , and for @xmath720 let @xmath934 be an @xmath722-vortex .
assume that on every compact subset of @xmath65 , @xmath723 converges to @xmath724 in @xmath725 and @xmath726 converges to @xmath727 in @xmath512 .
then we have @xmath935 in @xmath725 on every compact subset of @xmath65 . in the * case @xmath761 *
the statement of the lemma is a consequence of equality ( [ eq : e r w 1 2 ] ) and the second rescaled vortex equation ( [ eq : f a r ] ) .
consider the * case @xmath765*. it follows from our standing hypothesis ( h ) that there exists a constant @xmath851 such that @xmath2 acts freely on @xmath936 properness of @xmath7 implies that @xmath795 is compact .
let @xmath937 be a compact subset .
the convergence of @xmath726 and the fact @xmath938 imply that for @xmath508 large enough , we have @xmath939 .
furthermore , our hypotheses about the convergence of @xmath723 and @xmath726 imply that @xmath940 .
finally , since @xmath795 is compact and @xmath2 acts freely on it , we have @xmath941 therefore , we may apply lemma [ le : mu u ] ( appendix [ sec : vort ] ) , to conclude that @xmath942 since @xmath943 , and @xmath944 , the convergence ( [ eq : e r nu w nu e f ] ) follows .
this completes the proof of lemma [ le : conv e ] . in the proof of proposition [ prop : quant en loss ] we will also use the following . [
rmk : bar j]let @xmath945 be an @xmath41-vortex , i.e. , a solution of the equations @xmath946 and @xmath890 . then by proposition [ prop : bar del j ] ( appendix [ sec : add ] ) the map @xmath947 is @xmath356-holomorphic , and @xmath948 .
if this energy is finite , then by removal of singularities the map @xmath949 extends to a @xmath356-holomorphic map @xmath950 .
it follows that @xmath951 , provided that @xmath952 .
@xmath271 we write @xmath953 .
consider the function @xmath954 suppose that the hypotheses of proposition [ prop : quant en loss ] are satisfied and that there exists a sequence @xmath955 that converges to some @xmath956 , such that @xmath957 .
then there exists a number @xmath958 and an @xmath959-vortex @xmath960 , such that @xmath961 for every @xmath296 so small that @xmath923 .
* construction of @xmath959 : * we define @xmath962 . for @xmath508
large enough we have @xmath963 .
we pass to some subsequence such that this holds for every @xmath508 . by lemma [ lemma : hofer ] , applied with @xmath964 , there exist @xmath965 and @xmath966 , such that @xmath967 since by assumption @xmath957 , it follows from ( [ eq : ze z nu ] ) that the sequence @xmath968 converges to @xmath779 .
we define @xmath969 note that @xmath970 is an @xmath971-vortex . passing to some subsequence
we may assume that @xmath971 converges to some @xmath972 $ ] .
since @xmath973 , it follows from ( [ eq : eps nu ] ) that @xmath974 .
it follows that the * second inequality in ( [ eq : r 0 limsup ] ) * holds for the original sequence .
* construction of @xmath764 : * we choose a sequence @xmath975 of open sets such that @xmath976 and @xmath977 , for every @xmath132 .
we check the conditions of proposition [ prop : cpt bdd ] with these sets , @xmath978 , and @xmath979 replaced by @xmath980 : * condition ( [ eq : u nu om k ] ) * is satisfied by hypothesis .
we check * condition ( [ eq : sup e ] ) * : a direct calculation involving ( [ eq : sup b ] ) shows that @xmath981 it follows from ( [ eq : eps nu ] ) and the fact @xmath957 , that @xmath982 . combining this with ( [ eq : d wt a nu wt u nu ] ) , condition ( [ eq : sup e ] )
follows , for every compact subset @xmath983 .
therefore , applying proposition [ prop : cpt bdd ] , there exists an @xmath959-vortex @xmath984 and , passing to some subsequence , there exist gauge transformations @xmath985 , with the following properties . for every compact subset @xmath983
, @xmath986 converges to @xmath724 in @xmath725 on @xmath506 , and @xmath987 converges to @xmath727 in @xmath512 on @xmath506 .
we prove the * first inequality in ( [ eq : limsup e w 0 ] ) * : by lemma [ le : conv e ] we have @xmath988 in @xmath989 for every compact subset @xmath983 .
since @xmath990 it follows that @xmath991 .
this implies that @xmath992 .
this proves the first inequality in ( [ eq : limsup e w 0 ] ) .
we prove the * second inequality in ( [ eq : limsup e w 0 ] ) * : let @xmath296 be so small that @xmath923 , and @xmath993 .
it follows from ( [ eq : e r0 w0 ] ) that @xmath994 . by hypothesis
this supremum is finite .
therefore , there exists @xmath567 such that @xmath995 since @xmath996 the convergence ( [ eq : e r0 w0 ] ) implies that @xmath997 on the other hand , since @xmath998 and @xmath999 , for @xmath508 large enough the ball @xmath1000 is contained in @xmath1001 . combining this with ( [ eq : e r nu ] )
, we obtain @xmath1002 since this holds for every @xmath851 , the second inequality in ( [ eq : limsup e w 0 ] ) ( for the original sequence ) follows . it remains to prove the * first inequality in ( [ eq : r 0 limsup ] ) * , i.e. , that @xmath1003 .
assume by contradiction that @xmath1004 .
for a map @xmath1005 we denote by @xmath1006 its ( dirichlet-)energy . on @xmath4 .
] by the second @xmath221-vortex equation with @xmath1007 we have @xmath1008 . therefore , by proposition
[ prop : ka 0 ] ( appendix [ sec : add ] ) there exists @xmath1009 such that @xmath1010 . by the first vortex equation the map @xmath1011 is @xmath42-holomorphic .
let @xmath296 be such that @xmath923 . using the second inequality in ( [ eq : limsup e w 0 ] )
, we have @xmath1012 combining this with the hypothesis @xmath1013 , it follows that the energy @xmath1014 is finite .
hence by removal of singularities , @xmath1015 extends to a smooth @xmath42-holomorphic map @xmath657 . by the first inequality in ( [ eq : limsup e w 0 ] ) we have @xmath1016 this contradicts asphericity of @xmath0 .
hence @xmath959 must be positive .
this concludes the proof of the claim .
* statement ( [ prop : hard r e ] ) * of proposition [ prop : quant en loss ] follows from the claim , considering a sequence @xmath1017 , such that @xmath1018 , and using ( [ eq : r 0 limsup ] ) .
we prove * statement ( [ prop : hard limsup])*. assume that there exists a compact subset @xmath919 such that @xmath1019 .
let @xmath1017 be such that @xmath957 .
we choose a pair @xmath1020 as in the claim . using the first inequality in ( [ eq : limsup e w 0 ] ) and remark [ rmk : bar j ] ( in the case @xmath1021 )
, we have @xmath1022 . combining this with the second inequality in ( [ eq : limsup e w 0 ] ) , inequality ( [ eq : limsup emin ] )
this proves ( [ prop : hard limsup ] ) and concludes the proof of proposition [ prop : quant en loss ] .
we are now ready to prove proposition [ prop : cpt mod ] ( p. ) .
[ prop : cpt mod proof ] we abbreviate @xmath1023 . for every @xmath444
there exists a finite subset @xmath1024 such that the following holds . if @xmath761 then we have @xmath1025 . furthermore , if @xmath1026 then we have @xmath1027 for every compact subset @xmath1028 .
moreover , for every @xmath1029 and every @xmath296 the inequality ( [ eq : limsup emin ] ) holds . for @xmath1030
the assertion holds with @xmath1031 .
we fix @xmath1032 and assume by induction that there exists a finite subset @xmath1033 such that the assertion with @xmath1034 replaced by @xmath1035 holds . if ( [ eq : sup vert e nu ] ) is satisfied for every compact subset @xmath1036 , then the statement for @xmath1034 holds with @xmath1037 .
assume that there exists a compact subset @xmath1036 , such that ( [ eq : sup vert e nu ] ) does not hold .
it follows from the induction hypothesis that @xmath1038 there exists a point @xmath922 such that inequality ( [ eq : limsup emin ] ) holds , for every @xmath296 .
we set @xmath1039 .
it follows from the fact that ( [ eq : sup vert e nu ] ) does not hold and statement ( [ prop : hard r e ] ) of proposition [ prop : quant en loss ] that @xmath1040 . furthermore , since @xmath1041 , ( [ eq : z ell 1 ] ) implies that @xmath1042 .
it follows that the statement of the claim for @xmath1034 is satisfied . by induction
, the claim follows .
we fix an integer @xmath1043 and a finite subset @xmath1044 that satisfies the conditions of the claim .
it follows from the inequality ( [ eq : limsup emin ] ) that @xmath1045 .
hence by the statement of the claim , the hypothesis ( [ eq : sup e ] ) of proposition [ prop : cpt bdd ] is satisfied with @xmath1046 . applying that result and passing to some subsequence
, there exist an @xmath758-vortex @xmath1047 and gauge transformations @xmath760 , such that the * statements ( [ prop : cpt mod:<],[prop : cpt mod:= ] ) * of proposition [ prop : cpt mod ] are satisfied .
( here we use that @xmath762 if @xmath761 . )
we prove * statement ( [ prop : cpt mod lim nu e eps])*. passing to some `` diagonal '' subsequence , the limit @xmath1048 exists , for every @xmath820 and @xmath769 .
let now @xmath769 and @xmath296 .
we choose @xmath820 bigger than @xmath1049 .
for @xmath1050 we denote @xmath1051 by lemma [ le : conv e ] the limit @xmath1052 exists and equals @xmath1053 .
it follows that the limit @xmath773 exists .
inequality ( [ eq : limsup emin ] ) implies that @xmath1054 .
since @xmath1053 depends continuously on @xmath559 , the same holds for @xmath1055 .
this proves statement ( [ prop : cpt mod lim nu e eps ] ) and completes the proof of proposition [ prop : cpt mod ] . in the above proof
the set of bubbling points @xmath1056 is constructed by `` terminating induction '' . intuitively , this is induction over the number of bubbling points .
the `` auxiliary index '' @xmath1034 in the claim is needed to make this idea precise .
inequality ( [ eq : limsup emin ] ) ensures that the `` induction stops '' . @xmath271
the following proposition will be used inductively in the proof of theorem [ thm : bubb ] to find the next bubble in the bubbling tree , at a bubbling point of a given sequence of rescaled vortex classes .
it is an adaption of ( * ? ? ?
* proposition 4.7.1 . ) to vortices .
[ prop : soft ] assume that @xmath0 is aspherical .
let @xmath136 , @xmath1057 , @xmath1058 be a sequence that converges to @xmath41 , @xmath156 , and for every @xmath132 let @xmath1059 be an @xmath722-vortex , such that the following conditions are satisfied .
[ prop : soft k ] there exists a compact subset @xmath754 such that @xmath1060 for every @xmath508 .
[ prop : soft e ] for every @xmath1061 the limit @xmath1062 exists and @xmath1063 .
furthermore , the function @xmath1064\ni{\varepsilon}\mapsto e({\varepsilon})\in{\mathbb{r}}\ ] ] is continuous . then there
exist @xmath1065 , a finite subset @xmath757 , an @xmath758-vortex @xmath759 , and , passing to some subsequence , there exist sequences @xmath1066 , @xmath1067 , and @xmath760 , such that , defining @xmath1068 the following conditions hold .
[ prop : soft z ] if @xmath1069 then @xmath762 and @xmath1070 .
if @xmath765 and @xmath1071 then @xmath1072 .
[ prop : soft eps ] the sequence @xmath1073 converges to @xmath779 .
furthermore , if @xmath1069 then @xmath1074 for every @xmath508 , and if @xmath765 then @xmath1075 converges to 0 and @xmath1076 converges to @xmath41 .
[ prop : soft conv ] if @xmath1069 then the sequence @xmath1077 converges to @xmath764 in @xmath471 on every compact subset of @xmath766 .
furthermore , if @xmath765 then on every compact subset of @xmath766 , the sequence @xmath1078 converges to @xmath724 in @xmath725 , and the sequence @xmath1079 converges to @xmath727 in @xmath512 .
[ prop : soft lim nu e eps ] fix @xmath769 and a number @xmath770 such that @xmath771 .
then for every @xmath772 the limit @xmath1080 exists and @xmath1081 .
furthermore , the function @xmath1082 is continuous .
[ prop : soft en ] we have @xmath1083 roughly speaking , this result states that given a sequence of `` zoomed out '' vortices for which a positive amount of energy is concentrated around some point @xmath779 , after `` zooming back in '' , some subsequence converges either to a vortex over @xmath1 or a holomorphic sphere in the symplectic quotient , up to bubbling at finitely many points .
furthermore , no energy is lost in the limit between the original sequence and the `` zoomed in '' subsequence . to explain this ,
note that condition ( [ prop : soft e ] ) in the hypothesis implies that given an arbitrarily small ball around the point @xmath779 , the energy of the vortex @xmath517 on the ball is bounded below by a positive constant arbitrarily close to @xmath729 , provided that @xmath508 is large enough .
this means that in the limit @xmath134 , a positive amount of energy bubbles off around the point @xmath779 .
it follows that there exists a sequence of points @xmath137 converging to @xmath779 , such that the energy density of @xmath517 at @xmath1073 converges to @xmath41 . in the conclusion of the proposition , condition ( [ prop : soft eps ] ) says that we are `` zooming in '' around @xmath779 , and it specifies how fast we do so .
condition ( [ prop : soft conv ] ) states that the `` zoomed in '' subsequence converges to either a vortex over @xmath1 or a holomorphic sphere in @xmath116 , depending on how fast we `` zoom back in '' .
condition ( [ prop : soft z ] ) implies that the limit object is nontrivial or there are at least two new bubbling points .
( the latter can only happen if the limit is a holomorphic sphere in @xmath116 . ) in the proof of theorem [ thm : bubb ] , this will guarantee that the new bubble is stable .
analogously to condition ( [ prop : soft e ] ) , condition ( [ prop : soft lim nu e eps ] ) implies that at each point in @xmath1056 in the limit @xmath134 , at least the energy @xmath729 bubbles off in the sequence of rescaled vortices @xmath1084 . in the proof of theorem [ thm : bubb ] , this will be used to prove that the inductive construction of the bubble tree terminates after finitely many steps .
finally , condition ( [ prop : soft en ] ) will ensure that at each step no energy is lost between the old and new bubble . in condition ( [ prop : soft conv ] )
the pullback @xmath1084 is defined over the set @xmath1085 .
this set contains any given compact subset @xmath791 , provided that @xmath508 is large enough ( depending on @xmath506 ) .
hence condition ( [ prop : soft conv ] ) makes sense . @xmath271
the proof of proposition [ prop : soft ] is given on page .
it is based on the following result , which states that the energy of a vortex over an annulus is concentrated near the ends , provided that it is small enough . for @xmath1086
we denote the open annulus around 0 with radii @xmath1087 by @xmath1088 note that @xmath1089 , and @xmath1090 in the case @xmath1091 .
we define @xmath1092\ ] ] to be the distance function induced by the riemannian metric @xmath97 .
is disconnected then @xmath591 attains the value @xmath41 . ]
we define @xmath1093,\qquad \bar d(\bar x,\bar y):=\min_{x\in\bar x,\,y\in\bar y}d(x , y).\ ] ] by lemma [ le : metr ] in appendix [ sec : add ] this is a distance function on @xmath102 which induces the quotient topology .
[ prop : en conc ] there exists a constant @xmath1003 such that for every compact subset @xmath754 and every @xmath296 there exists a constant @xmath1094 , such that the following holds .
assume that @xmath1095 , @xmath156 , and @xmath1096 is a vortex ( with respect to @xmath753 ) , such that @xmath1097 then we have we have @xmath1098 , hence ( [ eq : e w a ar],[eq : sup z z bar d ] ) are non - trivial only for @xmath1099.]@xmath1100 ( here @xmath1101 denotes the orbit of a point @xmath74 . )
the proof of this result is modeled on the proof of ( * ? ? ?
* theorem 1.3 ) , which in turn is based on the proof of ( * ? ? ?
* proposition 11.1 ) .
it is based on an isoperimetric inequality for the invariant symplectic action functional ( theorem [ thm : isoperi ] in appendix [ sec : action ] ) .
it also relies on an identity relating the energy of a vortex over a compact cylinder with the actions of its end - loops ( proposition [ prop : en act ] in appendix [ sec : action ] ) .
the proof of ( [ eq : sup z z bar d ] ) also uses the following remark .
[ rmk : bar d bar x 0 bar x 1 ] let @xmath1102 be a riemannian manifold , @xmath2 a compact lie group that acts on @xmath4 by isometries , @xmath66 a @xmath2-bundle over @xmath1103 $ ] , @xmath1104 a connection , and @xmath362 a map .
we define @xmath1105 where @xmath699 , and the norm is taken with respect to the standard metric on @xmath1103 $ ] and @xmath1106 . furthermore , we define @xmath1107\to m / g,\quad\bar u(t):=gu(p),\ ] ] where @xmath229 is any point over @xmath278 .
we denote by @xmath591 the distance function induced by @xmath1106 , and define @xmath1108 as in ( [ eq : bar d ] ) .
then for every pair of points @xmath1109 , we have @xmath1110 this follows from a straight - forward argument .
@xmath271 for every subset @xmath1111 we define @xmath1112 where the norms are with respect to @xmath97 and @xmath62 .
we set @xmath1113 let @xmath754 be a compact subset and @xmath296 . replacing @xmath795 by @xmath796 , we may assume w.l.o.g . that @xmath795 is @xmath2-invariant .
an elementary argument using our standing hypothesis ( h ) shows that there exists a number @xmath1114 such that @xmath2 acts freely on @xmath1115 , and @xmath1116 we choose a constant @xmath1117 as in theorem [ thm : isoperi ] in appendix [ sec : action ] ( isoperimetric inequality ) , corresponding to @xmath1118 shrinking @xmath1117 we may assume that it satisfies the condition of proposition [ prop : en act ] in appendix [ sec : action ] ( energy action identity ) for @xmath1119 .
we choose a constant @xmath1120 as in lemma [ le : a priori ] in appendix [ sec : vort ] ( called @xmath1094 there ) , corresponding to @xmath795 .
we define @xmath1121 assume that @xmath1122 are as in the hypothesis . without loss of generality , we may assume that @xmath1123 .
consider first the * case @xmath1124 * , and assume that @xmath98 extends to a smooth vortex over the compact annulus of radii @xmath570 and @xmath221 .
we show that * inequality ( [ eq : e w a ar ] ) * holds .
we define the function @xmath1125 for every @xmath1126 we have @xmath1127 using the fact @xmath1128 and ( [ eq : e w a e 0],[eq : e 0 de 0 de ] ) , it follows from lemma [ le : a priori ] in appendix [ sec : vort ] ( with @xmath570 replaced by @xmath1129 ) that @xmath1130 we define @xmath1131 ( here we identify @xmath1132 , where @xmath1133 , for every @xmath673 . ) let @xmath1134 $ ] . combining ( [ eq : e w z 0 8 ] ) with the fact @xmath579 and remark [ rmk : bar d bar x 0 bar x 1 ]
, it follows that @xmath1135 hence the hypotheses of theorem [ thm : isoperi ] ( appendix [ sec : action ] ) are satisfied with @xmath795 replaced by @xmath1119 and @xmath1136 . by the statement of that result the loop @xmath1137 is admissible , and defining @xmath1138 we have @xmath1139 here @xmath1140 denotes the invariant symplectic action , as defined in appendix [ sec : action ] .
furthermore , the @xmath1141-norms are with respect to the standard metric on @xmath1142 , the metric @xmath97 on @xmath4 , and the operator norm @xmath1143 , induced by @xmath62 . by ( [ eq : r 0 2 pi m],[eq : m k 1 eps ] ) and the fact @xmath1144
, we have @xmath1145 here the norm @xmath1146 is with respect to the standard metric on @xmath1142 , and we used the fact that for every @xmath1147 , @xmath1148 where @xmath884 is determined by @xmath1149 .
we fix @xmath1126 . recalling ( [ eq : e s ] ) , we have @xmath1150 combining this with ( [ eq : a iota s 0],[eq : frac 1 2 2 eps ] ) , it follows that @xmath1151 using ( [ eq : wt u bar ell ] ) , the hypotheses of proposition [ prop : en act ] are satisfied with @xmath795 replaced by @xmath1119 . applying that result , we have @xmath1152 combining this with ( [ eq : a iota ] ) , inequality ( [ eq : d ds e ] ) follows .
this proves the claim .
by the claim the derivative of the function @xmath1153 is non - positive , and hence this function is non - increasing .
inequality ( [ eq : e w a ar ] ) follows .
* we prove ( [ eq : sup z z bar d])*. let @xmath1154 . using ( [ eq : e w a e 0 ] ) and the fact @xmath1155 , it follows from lemma [ le : a priori ] ( with @xmath570 replaced by @xmath1129 ) that @xmath1156 we define @xmath1157 .
then @xmath1158 and @xmath1159 is contained in @xmath1160 .
therefore , by ( [ eq : e w a ar ] ) we have @xmath1161 combining this with ( [ eq : e w z ] ) and the fact @xmath1162 , it follows that @xmath1163 where @xmath1164 .
a similar argument shows that @xmath1165 let now @xmath1166 and @xmath1167 .
assume that @xmath1168 .
( this is no real restriction . )
we define @xmath1169\to{\mathbb c}$ ] to be the radial path of constant speed , such that @xmath1170 and @xmath1171 .
furthermore , we choose an angular path @xmath1172\to{\mathbb c}$ ] of constant speed , such that @xmath1173 , @xmath1174 , and @xmath1175 has minimal length among such paths .
( see figure [ fig : path ] . )
consider the `` twisted length '' of @xmath1176 , given by @xmath1177 it follows from ( [ eq : d a u c 2r],[eq : d a u c r ] ) and the fact @xmath1168 , that this length is bounded above by @xmath1178 .
similarly , it follows that the `` twisted length '' of @xmath1179 is bounded above by @xmath1180 .
therefore , using remark [ rmk : bar d bar x 0 bar x 1 ] , inequality ( [ eq : sup z z bar d ] ) with @xmath559 replaced by @xmath1181 follows .
assume now that @xmath98 is not smooth . by theorem [ thm : reg gauge bdd ] in appendix
[ sec : vort ] the restriction of @xmath98 to any compact cylinder contained in @xmath1182 is gauge equivalent to a smooth vortex .
hence the inequalities ( [ eq : e w a ar],[eq : sup z z bar d ] ) follow from what we just proved , using the @xmath2-invariance of @xmath795 . in the * case @xmath373 * the inequalities ( [ eq : e w a ar],[eq :
sup z z bar d ] ) follow from what we just proved , by considering an arbitrarily large radius @xmath1124 .
this completes the proof of proposition [ prop : en conc ] .
we are now ready for the proof of proposition [ prop : soft ] ( p. ) .
[ proof prop : soft ] the function @xmath133 as in ( [ eq : eps e eps ] ) is increasing and , by hypothesis ( [ prop : soft e ] ) , bounded below by @xmath729 .
hence the limit @xmath1183 exists and is bounded below by @xmath729 .
we fix a compact subset @xmath754 as in hypothesis ( [ prop : soft k ] ) .
we choose a constant @xmath1184 as in lemma [ le : a priori ] in appendix [ sec : vort ] ( bound on energy density ) , depending on @xmath795 . considering the pullback of @xmath517 by the translation by @xmath779
, we may assume w.l.o.g .
that @xmath1185 .
[ claim : soft wlog ] we may assume w.l.o.g .
that @xmath1186 [ proof of claim [ claim : soft wlog ] ] suppose that we have already proved the proposition under this additional assumption , and let @xmath1187 be as in the hypotheses of the proposition .
we choose @xmath1188 so small that @xmath1189 for @xmath132 we choose @xmath1190 such that @xmath1191 [ claim : z nu ] the sequence @xmath1192 converges to @xmath1193 .
recall that @xmath1182 denotes the open annulus of radii @xmath570 and @xmath221 .
let @xmath1194 .
inequality ( [ eq : e hhat r ] ) implies that there exists @xmath1195 such that @xmath1196 for every @xmath1197 .
hence by lemma [ le : a priori ] with @xmath1198 we have @xmath1199 we define @xmath1200 increasing @xmath1201 , we may assume that @xmath1202 for every @xmath1197 , and therefore @xmath1203 combining this with ( [ eq : e r ze],[eq : e 4 c ] ) and the fact @xmath1204 , it follows that @xmath1205 , for every @xmath1197 .
this proves claim [ claim : z nu ] . by claim [ claim : z nu ] we may pass to some subsequence such that @xmath1206 for every @xmath508 .
we define @xmath1207 then ( [ eq : e c 0 ] ) with @xmath1208 replaced by @xmath1209 is satisfied . by elementary arguments the hypotheses of proposition [ prop : soft ] are satisfied with @xmath1210 replaced by @xmath1211 . assuming that we have already proved the statement of the proposition for @xmath970 , a straight - forward argument using claim [ claim : z nu ] shows that it also holds for @xmath517 .
this proves claim [ claim : soft wlog ] .
* so we assume w.l.o.g . that ( [ eq : e c 0 ] ) holds . * + * construction of @xmath1212 and @xmath764 : * recall the definition ( [ eq : m 0 lim ] ) of @xmath1213 , and that we have chosen @xmath1184 as in lemma [ le : a priori ] .
we choose constants @xmath959 and @xmath1214 as in proposition [ prop : en conc ] , the latter ( called @xmath1094 there ) corresponding to the compact set @xmath795 and @xmath1215 .
we fix a constant @xmath1216 we pass to some subsequence such that @xmath1217 for every @xmath132 , there exists @xmath1218 , such that @xmath1219 it follows from the definition of @xmath1213 that @xmath1220 [ claim : inf nu hhat eps nu ] we have @xmath1221 equality ( [ eq : e c 0 ] ) implies that @xmath1222 the hypotheses @xmath855 , ( [ prop : soft k ] ) , and ( [ prop : soft e ] ) imply that the hypotheses of proposition [ prop : quant en loss ] ( quantization of energy loss ) are satisfied with @xmath1223 .
thus by assertion ( [ prop : hard r e ] ) of that proposition with @xmath1224 , we have @xmath1225 combining this with ( [ eq : e eps nu c],[eq : e r = ] ) and the fact @xmath1226 , inequality ( [ eq : hhat eps nu r nu ] ) follows .
this proves claim [ claim : inf nu hhat eps nu ] . passing to some subsequence
, we may assume that the limit @xmath1227\ ] ] exists .
by claim [ claim : inf nu hhat eps nu ] we have @xmath1228 .
we define @xmath1229 @xmath1230 by proposition [ prop : cpt mod ] with @xmath1231 @xmath517 replaced by @xmath971 , @xmath970 and @xmath1232 there exist a finite subset @xmath757 and an @xmath758-vortex @xmath1233 , and passing to some subsequence , there exist gauge transformations @xmath760 , such that the conclusions of that proposition hold .
we check the * conditions of proposition [ prop : soft ] * with @xmath1234 : * condition [ prop : soft]([prop : soft eps ] ) * holds by ( [ eq : eps nu 0],[eq : hhat eps r],[eq : r 0 eps nu ] ) .
* condition [ prop : soft]([prop : soft conv ] ) * follows from [ prop : cpt mod]([prop : cpt mod:<],[prop : cpt mod:= ] ) , and * condition [ prop : soft]([prop : soft lim nu e eps ] ) * follows from [ prop : cpt mod]([prop : cpt mod lim nu e eps ] ) .
we prove * condition [ prop : soft]([prop : soft en ] ) * : we define @xmath1235 we choose @xmath1061 so small that @xmath1236 furthermore , we choose an integer @xmath1237 so large that for @xmath885 , we have @xmath1238 .
we fix @xmath885 . using ( [ eq : e r = ] , [ eq : de min ] )
, it follows that @xmath1239 it follows that the requirements of proposition [ prop : en conc ] are satisfied with @xmath570 , @xmath221 , @xmath517 replaced by @xmath1240 , @xmath1241 , @xmath1242 .
therefore , we may apply that result ( with `` @xmath559 '' equal to @xmath1243 ) , obtaining @xmath1244 using ( [ eq : hhat eps r],[eq : r 0 eps nu ] ) and the fact @xmath1245 , the equality ( [ eq : en cons ] ) follows .
this proves [ prop : soft]([prop : soft en ] ) . to see that * condition [ prop : soft]([prop : soft z ] ) * holds , assume first that @xmath1069
. then @xmath762 by statement ( [ prop : cpt mod : < ] ) of proposition [ prop : cpt mod ] . the same statement and lemma [ le : conv e ] imply that @xmath1246 it follows from the convergence @xmath1247 and ( [ eq : e r = ] , [ eq : de min ] ) that this limit is positive .
this proves condition [ prop : soft]([prop : soft z ] ) in the case @xmath1069 .
* assume now that @xmath765 and @xmath1071 .
* then condition [ prop : soft]([prop : soft z ] ) is a consequence of the following two claims .
[ claim : z b 1 ] the set @xmath1056 is not contained in the open ball @xmath550 . by [ prop : soft]([prop : soft en ] ) ( which we proved above ) there exists @xmath567 so that @xmath1248 ( here we used that @xmath1249 . ) since @xmath765 , we have @xmath1250 . hence it follows from ( [ eq : e r = ] ) and ( [ eq : e eps lim ] ) that @xmath1251 by the definition ( [ eq : m 0 lim ] ) of @xmath1213 , the right hand side is bounded below by @xmath1117 . therefore , combining ( [ eq : lim e r nu ] ) with ( [ eq : limsup de ] ) , it follows that @xmath1252 suppose by contradiction that @xmath1253 .
then by [ prop : cpt mod]([prop : cpt mod:= ] ) , the connection @xmath986 converges to @xmath724 in @xmath725 on @xmath1254 , and the map @xmath987 converges to @xmath727 in @xmath512 on @xmath1255 .
hence lemma [ le : conv e ] implies that @xmath1256 combining this with ( [ eq : liminf e a ] ) , we arrive at a contradiction to our assumption @xmath1071 .
this proves claim [ claim : z b 1 ] .
[ claim:0 z ] the set @xmath1056 contains @xmath1193 . by claim [ claim : z b 1 ] the set @xmath1257 is nonempty .
we choose a point @xmath1258 and a number @xmath770 so small that @xmath771 .
we fix @xmath772 .
since @xmath1259 ( as @xmath134 ) , ( [ eq : e c 0 ] ) implies that @xmath1260 for @xmath508 large enough . combining this with condition [ prop : soft]([prop : soft lim nu e eps ] ) ( which we proved above ) , it follows that @xmath1261 since @xmath1262 is arbitrary , it follows that @xmath1263 if @xmath1193 did not belong to @xmath1056 , then by [ prop : cpt mod]([prop : cpt mod:= ] ) and lemma [ le : conv e ] the energy density @xmath1264 would converge to @xmath1265 , a contradiction to ( [ eq : e wt r nu wt w nu 0 ] ) .
this proves claim [ claim:0 z ] , and completes the proof of [ prop : soft]([prop : soft z ] ) and therefore of proposition [ prop : soft ] .
[ rmk : soft proof ] assume that @xmath1266 are constructed as in the proof of condition ( [ prop : soft z ] ) of proposition [ prop : soft ] , and that @xmath765 and @xmath1071 . then @xmath1267 ( and hence @xmath1268 by claim [ claim : z b 1 ] ) .
this follows from the inequalities @xmath1269 here the first inequality is a consequence of condition ( [ eq : e r = ] ) .
@xmath271 based on the results of the previous sections , we are now ready to prove the first main result of this memoir , theorem [ thm : bubb ] .
the proof is an adaption of the proof of ( * ? ? ?
* theorem 5.3.1 ) to the present setting .
the strategy is the following : consider first the case @xmath1270 , i.e. , the only marked point is @xmath1271 .
we rescale the sequence @xmath135 so rapidly that all the energy is concentrated around the origin in @xmath1 .
then we `` zoom back in '' in a soft way , to capture the bubbles ( spheres in @xmath116 and vortices over @xmath1 ) in an inductive way .
( see claim [ claim : tree ] below . )
next we show that at each stage of this construction , the total energy of the components of the tree plus the energy loss at the unresolved bubbling points equals the limit of the energies @xmath1272 . here
we call a bubbling point `` unresolved '' if we have not yet constructed any sphere or vortex which is attached to this point .
( see claim [ claim : f e ] . )
furthermore , we prove that the number of vertices of the tree is uniformly bounded above .
( see inequality ( [ eq : n e ] ) . )
this implies that the inductive construction terminates at some point .
we also show that the vortices over @xmath1 and the bubbles in @xmath116 have the required properties and that the data fits together to a stable map , which is the limit of a subsequence of @xmath1273 .
( see claims [ claim : mmm bar u i ] and [ claim : st conv ] . ) for @xmath441 we then prove the statement of the theorem inductively , using the statement for @xmath1270 . at each induction step
we need to handle one additional marked point in the sequence of vortex classes and marked points . in the limit
there are three possibilities for the location of this point : + * ( i ) * it does not coincide with any special point . *
( ii ) * it coincides with the marked point @xmath409 ( lying on the vertex @xmath466 ) , for some @xmath179 . *
( iii ) * it lies between two already constructed bubbles . + in case ( i ) we just include the new marked point into the bubble tree . in case ( ii )
, we introduce either * a `` ghost bubble '' , which carries the two marked points and is connected to @xmath466 , or * a `` ghost vortex '' , which is connected to @xmath466 , and a `` ghost bubble of type 0 '' , which is connected to the `` ghost vortex '' and carries the two marked points . the second option will only occur if @xmath1274 . which of the two options we choose depends on the speed
at which the two marked points come together . in case ( iii )
we introduce a `` ghost bubble '' between the two bubbles , which carries the new marked point .
[ thm : bubb proof]we consider first the * case @xmath1270 .
* let @xmath135 be a sequence of vortex classes as in the hypothesis . for each @xmath132
we choose a representative @xmath1275 of @xmath135 , such that @xmath1276 . passing to some subsequence
we may assume that @xmath1277 converges to some constant @xmath133 .
the hypothesis @xmath1278 ( for every @xmath508 ) implies that @xmath1279 .
we choose a sequence @xmath1280 such that @xmath1281 we define @xmath1282 the next claim provides an inductive construction of the bubble tree .
( some explanations are given below .
see also figure [ fig : claim ] . )
[ claim : tree ] for every integer @xmath1283 , passing to some subsequence , there exist an integer @xmath1284 and tuples
@xmath1285 where @xmath1286 , @xmath1287 is a finite subset , @xmath1288 is an @xmath1289-vortex , @xmath1290 , @xmath1291 , @xmath1292 , and @xmath1293 , such that the following conditions hold .
[ claim : tree z dist ] for every @xmath1294 we have @xmath1295 .
moreover , if @xmath1296 are such that @xmath1297 and @xmath1298 then @xmath1299 . [ claim : tree stab ] let @xmath1300 .
if @xmath1301 then @xmath1302 and @xmath1303 .
if @xmath1304 and @xmath1305 then @xmath1306 .
[ claim : tree r ] fix @xmath1300 . if @xmath1301 then @xmath1307 for every @xmath508 , and if @xmath1304 then @xmath1308 .
furthermore , @xmath1309 in the following we set @xmath1310 , for @xmath1311 and @xmath132 .
[ claim : tree conv ] for every @xmath1300 there exist gauge transformations @xmath1312 such that the following holds . if @xmath1301 then @xmath1313 converges to @xmath1314 in @xmath471 on every compact subset of @xmath1 . furthermore
, if @xmath1304 then on every compact subset of @xmath1315 the sequence @xmath1316 converges to @xmath1317 in @xmath725 , and the sequence @xmath1318 converges to @xmath1319 in @xmath512 .
[ claim : tree lim nu e eps ] let @xmath1300 , @xmath1320 and @xmath770 be such that @xmath1321 .
then for every @xmath772 the limit @xmath1322 exists , and @xmath1081 .
furthermore , the function @xmath1323 is continuous .
[ claim : tree b b]for every @xmath1300 , we have @xmath1324 [ claim : tree compl ] if @xmath1325 then for every @xmath1326 we have @xmath1327 to understand this claim , note that the collection @xmath1328 describes a tree with vertices the numbers @xmath1329 and unordered edges @xmath1330 . attached to the vertices of this tree are vortex classes and @xmath41-vortex classes .
( the latter will give rise to holomorphic spheres in @xmath116 . ) each pair @xmath1331 defines a rescaling @xmath1332 , which is used to obtain the @xmath179-th limit vortex or @xmath41-vortex .
( see condition ( [ claim : tree conv ] ) . )
the point @xmath409 is the nodal point on the @xmath1333-th vertex , at which the @xmath179-th vertex is attached .
the corresponding nodal point on the @xmath179-th vertex is @xmath41 .
the set @xmath1334 consists of the nodal points except @xmath41 ( if @xmath811 ) on the @xmath179-th vertex together with the bubbling points that have not yet been resolved .
condition ( [ claim : tree z dist ] ) implies that the nodal points at a given vertex are distinct .
condition ( [ claim : tree stab ] ) guarantees that once all bubbling points have been resolved , the @xmath179-th component will be stable .
there is another nodal point at @xmath41 , and for @xmath1335 there will be a marked point at @xmath41 , which is the limit of the sequence @xmath1336 . ]
condition ( [ claim : tree r ] ) implies that the rescalings @xmath1332 `` zoom out '' less than the rescalings @xmath1337 .
a consequence of condition ( [ claim : tree lim nu e eps ] ) is that at every nodal or unresolved bubbling point at least the energy @xmath729 concentrates in the limit .
condition ( [ claim : tree b b ] ) means that no energy is lost in the limit between each pair of adjacent bubbles .
finally , condition ( [ claim : tree compl ] ) means that in the case @xmath1325 all bubbling points have been resolved .
we show that the statement holds for @xmath1338 .
we check the conditions of proposition [ prop : soft ] ( soft rescaling ) with @xmath1339 , @xmath1340 and @xmath722 , @xmath517 replaced by @xmath1341 , @xmath1342 .
condition [ prop : soft]([prop : soft k ] ) follows from proposition [ prop : bounded ] in appendix [ sec : vort ] , using the hypothesis that @xmath4 is equivariantly convex at @xmath41 .
condition [ prop : soft]([prop : soft e ] ) follows from the facts @xmath1343 the first condition is a consequence of the facts @xmath1344 , @xmath1345 , and ( [ eq : e b e ] ) . thus by proposition [ prop : soft ] , there exist @xmath1346 , a finite subset @xmath757 , and an @xmath758-vortex @xmath1347 , and passing to some subsequence , there exist sequences @xmath1066 , @xmath1073 , and @xmath912 , such that the conclusions of proposition [ prop : soft ] with @xmath979 replaced by @xmath1348 hold .
we define @xmath1349 we check * conditions ( [ claim : tree z dist])-([claim : tree compl ] ) * of claim [ claim : tree ] with @xmath1350 : conditions ( [ claim : tree z dist],[claim : tree compl ] ) are void .
furthermore , conditions ( [ claim : tree stab])-([claim : tree b b ] ) follow from [ prop : soft]([prop : soft z])-([prop : soft en ] ) .
this proves the statement of the claim for @xmath1350 .
let @xmath1283 and assume , by induction , that we have already proved the statement of claim [ claim : tree ] for @xmath1034 .
we show that it holds for @xmath1351 . by assumption
there exists a number @xmath1352 and there exist collections @xmath1353 such that conditions ( [ claim : tree z dist])-([claim : tree compl ] ) hold . if @xmath1354 then conditions ( [ claim : tree z dist])-([claim : tree compl ] ) hold with @xmath1355 , and we are done . hence assume that there exists a @xmath1356 such that @xmath1357 we set @xmath1358 and choose an element @xmath1359 we fix a number @xmath136 so small that @xmath1360 .
we apply proposition [ prop : soft ] with @xmath1361 and @xmath722 , @xmath517 replaced by @xmath1362 , @xmath1363 condition [ prop : soft]([prop : soft k ] ) holds by hypothesis .
furthermore , by condition ( [ claim : tree lim nu e eps ] ) for @xmath1034 , condition [ prop : soft]([prop : soft e ] ) is satisfied . hence passing to some subsequence ,
there exist @xmath1065 , a finite subset @xmath757 , an @xmath758-vortex @xmath1364 , and sequences @xmath1066 , @xmath1073 , such that the conclusion of proposition [ prop : soft ] holds .
we define @xmath1365 we check * conditions ( [ claim : tree z dist])-([claim : tree compl ] ) * of claim [ claim : tree ] with @xmath1034 replaced by @xmath1351 , i.e. , @xmath342 replaced by @xmath1366 .
condition ( [ claim : tree z dist ] ) follows from the induction hypothesis and ( [ eq : z n 1 ] ) .
conditions ( [ claim : tree stab])-([claim : tree b b ] ) follow from [ prop : soft]([prop : soft z])-([prop : soft en ] ) .
we show that ( [ claim : tree compl ] ) holds with @xmath342 replaced by @xmath1366 : by the induction hypothesis , it holds for @xmath342 .
hence ( [ eq : z j 0 ] ) implies that @xmath1367 .
so there is nothing to check .
this completes the induction and the proof of claim [ claim : tree ] .
let @xmath1283 be an integer and @xmath1352 , @xmath1368 , @xmath1369 , @xmath1331 be as in claim [ claim : tree ] . recall that @xmath1370 and @xmath1371 .
we fix @xmath1311 .
we define @xmath1372 , for every measurable subset @xmath1373 we denote @xmath1374 furthermore , for @xmath1320 we define @xmath1375 for @xmath1376 it follows from ( [ eq : e b e ] ) and @xmath1344 that the limit @xmath1377 exists and equals @xmath133 . for @xmath1300
it follows from condition ( [ claim : tree lim nu e eps ] ) that the limit ( [ eq : m i z ] ) exists and that @xmath1378 . for @xmath1379 we define @xmath1380 ( this is the set of points on the @xmath1381-th sphere that have not been resolved after the construction of the @xmath322-th bubble . )
we define the function @xmath1382 [ claim : f e ] @xmath1383 we show by induction that @xmath1384 for every @xmath1385 .
claim [ claim : f e ] is a consequence of this with @xmath1386 . for the proof of equality ( [ eq : k f e ] ) we need the following .
[ claim : e z ] for every @xmath1300 we have @xmath1387 let @xmath1300 .
we choose a number @xmath296 so small that @xmath1388 and if @xmath1389 are points in @xmath1334 then @xmath1390 . by condition ( [ claim : tree lim nu e eps ] ) of claim [ claim : tree ] , for each @xmath1320 the limit @xmath1391 exists .
lemma [ le : conv e ] implies that @xmath1392 combining this with condition ( [ claim : tree b b ] ) of claim [ claim : tree ] , equality ( [ eq : m j i z i ] ) follows from a straight - forward argument .
this proves claim [ claim : e z ] . since @xmath1393 , equality ( [ eq :
k f e ] ) for @xmath873 follows from claim [ claim : e z ] and the fact @xmath1394 .
let now @xmath1395 and assume that we have proved ( [ eq : k f e ] ) for @xmath322 .
an elementary argument using claim [ claim : e z ] with @xmath1396 shows ( [ eq : k f e ] ) with @xmath322 replaced by @xmath153 . by induction ,
claim [ claim : f e ] follows .
consider the tree relation @xmath133 on @xmath1397 defined by @xmath1398 iff @xmath1399 or @xmath1400 .
lemma [ le : weight ] in appendix [ sec : add ] with @xmath162 as in ( [ eq : f i ] ) , @xmath1401 , @xmath1402 , and @xmath1403 , implies that @xmath1404 ( hypothesis ( [ eq : f al e ] ) follows from conditions ( [ claim : tree stab],[claim : tree lim nu e eps ] ) of claim [ claim : tree ] . )
assume now that we have chosen @xmath1405 .
by ( [ eq : n e ] ) we have @xmath1325 , and therefore by condition ( [ claim : tree compl ] ) of claim [ claim : tree ] , equality ( [ eq : z j ] ) holds , for every @xmath1326 .
we define @xmath1406 and the tree relation @xmath133 on @xmath380 by @xmath1407 furthermore , for @xmath1408 such that @xmath1409 we define the nodal points @xmath1410 moreover , we define the marked point @xmath1411 [ claim : mmm bar u i ] let @xmath1412 . if @xmath1413 then @xmath1414 and @xmath1415 has compact closure . furthermore , if @xmath1416 then the map @xmath1417 extends to a smooth @xmath356-holomorphic map @xmath1418 we choose gauge transformations @xmath1312 as in condition ( [ claim : tree conv ] ) of claim [ claim : tree ] , and define @xmath1419 .
* assume that @xmath1413 .
* it follows from fatou s lemma that @xmath1420 furthermore , since by hypothesis @xmath4 is equivariantly convex at @xmath41 , by proposition [ prop : bounded ] in appendix [ sec : vort ] there exists a @xmath2-invariant compact subset @xmath1421 such that @xmath1422 , for every @xmath132 .
since @xmath1423 converges to @xmath1319 pointwise , it follows that @xmath1424 .
hence @xmath1314 has the required properties . *
assume now that @xmath1416 .
* by proposition [ prop : bar del j ] in appendix [ sec : add ] the map @xmath1425 is @xmath356-holomorphic , and @xmath1426 .
it follows from fatou s lemma that @xmath1427 therefore , by removal of singularities , it follows that @xmath1428 extends to a smooth @xmath356-holomorphic map @xmath1429 .
this proves claim [ claim : mmm bar u i ] .
[ claim : st conv ] the tuple @xmath1430)_{i\in t_1},(\bar u_i)_{i\in t_\infty},(z_{ii'})_{iei'},({{\alpha}}_0:=1,z_0:=\infty)\big)\ ] ] is a stable map in the sense of definition [ defi : st ] , and the sequence @xmath1431,z_0^\nu:=\infty)$ ] converges to @xmath152 in the sense of definition [ defi : conv ] .
( here @xmath1432 $ ] denotes the gauge equivalence class of @xmath1314 . )
we check the conditions of definition [ defi : st ] .
* condition ( [ defi : st comb ] ) * is a consequence of the definitions of @xmath1433 and @xmath431 .
* condition ( [ defi : st dist ] ) * follows from condition ( [ claim : tree z dist ] ) of claim [ claim : tree ] and the fact @xmath1302 , for @xmath1413 .
( this follows from condition ( [ claim : tree stab ] ) of claim [ claim : tree ] . )
* condition ( [ defi : st conn ] ) * follows from an elementary argument using claim [ claim : tree]([claim : tree r],[claim : tree conv],[claim : tree b b ] ) and proposition [ prop : en conc ] .
* condition ( [ defi : st st ] ) * follows from claim [ claim : tree]([claim : tree stab ] ) .
hence all conditions of definition [ defi : st ] are satisfied .
we check the * conditions of definition [ defi : conv]*. * condition ( [ eq : e v bar t ] ) * follows from claim [ claim : f e ] , using condition ( [ claim : tree compl ] ) of claim [ claim : tree ] .
* condition [ defi : conv]([defi : conv phi z ] ) * follows from a straight - forward argument , using claim [ claim : tree]([claim : tree r ] ) .
* condition [ defi : conv]([defi : conv al be ] ) * follows from claim [ claim : tree]([claim : tree r ] ) by an elementary argument .
* condition [ defi : conv]([defi : conv w ] ) * follows from claim [ claim : tree]([claim : tree conv ] ) . finally
, * condition [ defi : conv]([defi : conv z ] ) * is void , since @xmath1270 .
this proves claim [ claim : st conv ] .
thus we have proved theorem [ thm : bubb ] in the case @xmath1270 .
+ * we prove that the theorem holds for every @xmath441 : * we prove by induction that for every @xmath148 and every tuple @xmath1434 as in the hypotheses of theorem [ thm : bubb ] , there exists a stable map @xmath152 as in ( [ eq : w z ] ) and a collection @xmath1435 of mbius transformations , such that the conditions of definition [ defi : conv ] hold , @xmath1436 @xmath1437 and for every edge @xmath405 such that @xmath425 and @xmath1438 lies in the chain of vertices from @xmath420 to @xmath1439 , we have @xmath1440 for @xmath1270 we proved this above . in that construction condition ( [ eq :
al t infty q ] ) follows from statement ( [ prop : soft conv ] ) of proposition [ prop : soft ] .
furthermore , condition ( [ eq : lim de limsup nu ] ) is a consequence of condition ( [ claim : tree b b ] ) of claim [ claim : tree ] with @xmath1335 , and the facts @xmath1441 and @xmath1344 .
finally , condition ( [ eq : lim limsup be al ] ) follows from claim [ claim : tree]([claim : tree b b ] ) .
let now @xmath397 and assume that we have proved the statement for @xmath1442 . passing to some subsequence
, we may assume that for every @xmath400 , the limit @xmath1443 exists .
we set @xmath1444 passing to a further subsequence , we may assume that the limit @xmath1445 exists , for every @xmath461 .
there are three cases .
+ * case ( i ) * there exists a vertex @xmath396 , such that @xmath1446 is not a special point of @xmath152 at @xmath420 . + * case ( ii ) * there exists an index @xmath1447 such that @xmath1448 . +
* case ( iii ) * there exists an edge @xmath405 such that @xmath1449 and @xmath1450 .
+ these three cases exclude each other . for the combination of the cases ( ii ) and ( iii )
this follows from the last part of condition ( [ defi : st dist ] ) ( distinctness of the special points ) in definition [ defi : st ] .
[ claim : i iii ] one of the three cases always applies .
this follows from an elementary argument , using that @xmath380 is finite and does not contain cycles .
* assume that case ( i ) holds .
* we fix a vertex @xmath396 such that @xmath1446 is not a special point .
we define @xmath1451 and introduce a new marked point @xmath1452 on the @xmath1453-sphere .
then @xmath152 augmented by @xmath1454 is again a stable map and the sequence @xmath1455 converges to this new stable map via @xmath1456 .
the induction step in case ( i ) follows .
+ * assume that case ( ii ) holds .
* we fix an index @xmath1457 such that @xmath1448 .
( it is unique . ) + consider first * case ( iia ) : @xmath1458 and the condition @xmath1459 or @xmath1460 holds*. in this case we extend the tree @xmath380 as follows .
we define @xmath1461 and introduce an additional vertex @xmath572 , which has type @xmath1381 and carries the new marked point .
this vertex is adjacent to @xmath466 in the new tree .
we move the @xmath179-th marked point from the vertex @xmath466 to the vertex @xmath572 and introduce an additional marked point on @xmath572 .
more precisely , we define @xmath1462 if @xmath1463 , i.e. , @xmath1464 , then we introduce the following `` ghost vortex '' : we choose a point @xmath1465 in the orbit @xmath1466 , and define @xmath1467 and @xmath1468 to be the map which is constantly equal to @xmath1465 .
we identify @xmath1469 with a connection on @xmath1470 and @xmath1471 with a @xmath2-equivariant map @xmath1472 , and set @xmath1473 . \ ] ] if @xmath1474 , i.e. , @xmath1475 , then we define @xmath1476 to be the constant map @xmath1477 we denote by @xmath1478 the tuple obtained from @xmath152 by making the changes described above . by elementary arguments
this is again a stable map .
we define the sequence of mbius transformations @xmath1479 by @xmath1480 [ claim : w new ii ] there exists a subsequence of @xmath1481 that converges to @xmath1478 via the mbius transformations @xmath1482
. condition ( [ eq : e v bar t ] ) ( energy conservation ) holds for every subsequence , since the new component @xmath572 carries no energy .
* conditions ( [ defi : conv phi z],[defi : conv al be ] ) of definition [ defi : conv ] * hold ( for the new collection of mbius transformations ) , by elementary arguments . * we check condition [ defi : conv]([defi : conv w ] ) * up to some subsequence . for every @xmath1483
we write @xmath1484 where @xmath1485 is defined as in ( [ eq : bar u w ] ) . in the * case @xmath1486 *
condition [ defi : conv]([defi : conv w ] ) does not contain any new requirement .
* assume that @xmath1463 .
* it follows from proposition [ prop : cpt mod ] ( compactness modulo bubbling and gauge ) with @xmath1487 , @xmath1488 and @xmath135 replaced by @xmath1489 , that there exists a vortex class @xmath1490 over @xmath1 such that passing to some subsequence , the sequence @xmath1489 converges to @xmath1490 , with respect to @xmath1491 ( as in definition [ defi : tau si ] ) , and the sequence @xmath1492 converges to @xmath1493 , uniformly on every compact subset of @xmath1 . [
claim : wt w ga w ga ] we have @xmath1494 ( defined as in ( [ eq : w ga ] ) ) .
a straight - forward argument implies that @xmath1495 .
therefore , condition [ defi : conv]([defi : conv w ] ) for the sequence @xmath1496 implies that the sequence @xmath1497 converges to @xmath1498 , in @xmath512 on some neighborhood of @xmath409 .
( here we used that @xmath1458 . )
since @xmath1499 converges to @xmath1500 , uniformly on every compact subset of @xmath1501 , it follows that @xmath1502 uniformly on every compact subset of @xmath1 .
it follows that @xmath1503 .
we choose representatives @xmath1504 of @xmath1490 and @xmath1505 of @xmath1506 . by hypothesis ( h )
the action of @xmath2 on @xmath6 is free .
hence , after regauging , we may assume that @xmath1507 .
( here we use lemma [ le : g smooth ] ( appendix [ sec : add ] ) , which ensures that the gauge transformation is smooth . )
it follows from the first vortex equation that @xmath1508 .
this shows that @xmath1494 and hence proves claim [ claim : wt w ga w ga ] .
this proves condition [ defi : conv]([defi : conv w ] ) in the case @xmath1463 .
* assume now that @xmath1474 .
* we have again @xmath1495 .
hence condition [ defi : conv]([defi : conv w ] ) for the sequence @xmath1496 implies that the sequence @xmath1497 converges to @xmath1498 , in @xmath512 on some neighborhood of @xmath409 .
( here we used that @xmath1458 . )
let @xmath1509 be a compact subset .
since @xmath1510 converges to @xmath409 , in @xmath471 on @xmath506 , it follows that @xmath1511 converges to @xmath1512 in @xmath512 on @xmath506 , as required .
this proves condition [ defi : conv]([defi : conv w ] ) in all cases .
* condition [ defi : conv]([defi : conv z ] ) * is a consequence of the definition ( [ eq : phi ga nu ] ) of @xmath1513 .
this proves claim [ claim : w new ii ] . to see that condition ( [ eq :
al t infty q ] ) holds , observe that we need to prove this only in the case @xmath1474 and then only for @xmath1514 . in this case
it follows from the same condition for @xmath463 and [ defi : conv]([defi : conv al be ] ) with @xmath463 and @xmath1515 .
( here we used that @xmath435 and it does not coincide with any nodal point at @xmath466 . )
the same conditions also imply ( [ eq : lim limsup be al ] ) for @xmath1516 and @xmath1517 .
( for the other pairs of adjacent vertices ( [ eq : lim limsup be al ] ) follows from the induction hypothesis . )
the induction step in case ( iia ) follows .
+ consider the * case ( iib ) : @xmath1376*. we define @xmath1518 the tuple @xmath1478 obtained from @xmath152 by making these changes , is again a stable map .
it follows from our assumption ( [ eq : phi al nu infty ] ) that there exist @xmath1519 and @xmath137 , such that @xmath1520 .
we define the sequence of mbius transformations @xmath1513 by @xmath1521 we show that some subsequence of @xmath1481 converges to @xmath1478 via the mbius transformations @xmath1482 : condition ( [ eq : e v bar t ] ) holds with @xmath380 replaced by @xmath1522 , since the map @xmath1523 is constant .
conditions [ defi : conv]([defi : conv phi z],[defi : conv al be],[defi : conv z ] ) follow from straight - forward arguments .
we show that * condition [ defi : conv]([defi : conv w ] ) * holds .
recall the definition ( [ eq : w al nu ] ) of @xmath1524 .
an elementary argument using condition [ defi : conv]([defi : conv w ] ) with @xmath1525 , our assumption ( [ eq : lim de limsup nu ] ) , and proposition [ prop : en conc ] , shows that for every @xmath296 there exist numbers @xmath1526 and @xmath875 such that @xmath1527 where @xmath1528 if @xmath430 , and @xmath1529 is defined as in ( [ eq : barev],[eq : barev w ] ) .
the sequence @xmath1530 converges to @xmath1531 , uniformly on every compact subset of @xmath1532 .
it follows that @xmath1492 converges to the constant map @xmath1533 , uniformly on every compact subset of @xmath1534 .
condition [ defi : conv]([defi : conv w ] ) with @xmath1514 is a consequence of the following .
[ claim : conv c 1 ] passing to some subsequence the convergence is in @xmath512 on every compact subset of @xmath1535 .
we denote @xmath1536 , and choose a representative @xmath1537 of @xmath1489 .
this is an @xmath722-vortex .
our assumption @xmath1538 implies that @xmath1539 .
hence by proposition [ prop : cpt mod ] there exists a finite subset @xmath757 and an @xmath41-vortex @xmath1540 , and passing to some subsequence , there exist gauge transformations @xmath1541 , such that the assertions [ prop : cpt mod]([prop : cpt mod:=],[prop : cpt mod lim nu e eps ] ) hold with @xmath1542 .
it follows that @xmath1492 converges in @xmath512 on every compact subset of @xmath766 .
furthermore , it follows from ( [ eq : lim de limsup nu ] ) and [ prop : cpt mod]([prop : cpt mod lim nu e eps ] ) that @xmath1543 .
claim [ claim : conv c 1 ] follows . conditions ( [ eq : al t infty q],[eq : lim de limsup nu ] ) ( for @xmath1544 ) and ( [ eq : lim limsup be al ] ) ( for @xmath1525 and @xmath1545 ) follow from condition ( [ eq : lim de limsup nu ] ) for @xmath1439 , which holds by the induction hypothesis .
this proves the induction step in case ( iib ) .
+ consider now * case ( iic ) : @xmath1546 and @xmath1274*. then we introduce a vertex @xmath1547 adjacent to @xmath466 and a vertex @xmath1548 adjacent to @xmath572 , such that @xmath1117 carries the two new marked points .
more precisely , we define @xmath1549 and @xmath1550 as in case ( iia ) with @xmath1463 .
we thus obtain a new stable map .
the proof of convergence to this stable map and of conditions ( [ eq : al t infty q ] ) ( for @xmath572 ) and ( [ eq : lim limsup be al ] ) ( for @xmath1514 and @xmath468 ) is now analogous to the proof in case ( iia ) .
this proves the induction step in case ( iic ) .
+ the cases ( iia , b , c ) cover all instances of case ( ii ) , hence we have proved the induction step in case ( ii ) . +
* assume now that case ( iii ) holds .
* in this case we introduce a new vertex @xmath572 between @xmath420 and @xmath1438 .
hence @xmath420 and @xmath1438 are no longer adjacent , but are separated by @xmath572 .
we define @xmath1551 if @xmath420 or @xmath1438 lies in @xmath418 , and @xmath1552 , otherwise . or @xmath1438 lies in @xmath431 . ]
we may assume w.l.o.g .
that @xmath1438 lies in the chain of vertices from @xmath420 to @xmath1439 .
we define @xmath1553 in the case @xmath1474 we define @xmath1554 to be the constant map equal to @xmath1555 . again , we obtain a new stable map . by condition ( [ eq : phi al nu infty ] ) there exist @xmath520 and @xmath521 , such that @xmath1556 .
we define @xmath1557 the proof of convergence proceeds now analogously to the proof in case ( iib ) , using ( [ eq : lim limsup be al ] ) for the proof of condition [ defi : conv]([defi : conv w ] ) rather than ( [ eq : lim de limsup nu ] ) .
furthermore , conditions ( [ eq : al t infty q ] ) with @xmath1514 and ( [ eq : lim limsup be al ] ) with @xmath1558 replaced by @xmath1559 and @xmath1560 follow from condition ( [ eq : lim limsup be al ] ) for @xmath1558 , which holds by the induction hypothesis .
this proves the induction step in case ( iii ) , and hence terminates the proof of theorem [ thm : bubb ] in the case @xmath441 . in the above proof for @xmath1270 the stable map @xmath152
is constructed by `` terminating induction '' . intuitively , this is induction over the integer @xmath342 occurring in claim [ claim : tree ] .
the `` auxiliary index '' @xmath1034 in claim [ claim : tree ] is needed to make this idea precise .
condition ( [ claim : tree compl ] ) and the inequality ( [ eq : n e ] ) ensure that the `` induction stops '' .
@xmath271 using compactness modulo bubbling and gauge for ( rescaled ) vortices , which was established in section [ sec : comp ] ( proposition [ prop : cpt mod ] ) , in this section we prove proposition [ prop : conv s 1 c ] . the proof is based on the following result .
let @xmath683 act on @xmath84 by multiplication , with momentum map @xmath1561 given by @xmath1562 , and let @xmath589 .
recall the definition ( [ eq : iota sym ] ) of the map @xmath1563 the definitions ( [ eq : m d sym],[eq : deg w z ] ) of the set @xmath1564 of finite energy vortex classes of degree @xmath591 and of the local degree map @xmath1565 , and the definition [ defi : tau si ] of the `` @xmath471-topology @xmath472 on @xmath470 '' . [ prop : loc conv s 1 c ] let @xmath1566 , @xmath1567 and @xmath1568 , for @xmath132 .
then the following conditions are equivalent .
[ prop : loc conv g ] for every open subset @xmath848 with compact closure and smooth boundary the restriction @xmath1569 converges to @xmath479 to @xmath849 , as @xmath134 , with respect to the topology @xmath505 .
[ prop : loc conv deg ] the point in the symmetric product @xmath1570 converges to @xmath1571 . in the proof of this result we will use the following .
let @xmath586 be a topological space and @xmath589 .
we denote by @xmath1572 the @xmath591-fold symmetric product of @xmath586 , and canonically identify it with the set @xmath1573 of all maps @xmath1574 such that @xmath1575 . to vanish , except for finitely many points @xmath927 .
] we endow @xmath1576 with the quotient topology . for every subset @xmath1577 and @xmath1578
, we define @xmath1579 where @xmath1580 denotes the boundary of @xmath1581 .
[ le : subbasis ] let @xmath1582 be a basis for the topology of @xmath586 . if @xmath586 is hausdorff then the sets @xmath1583 , @xmath1584 , @xmath1578 , form a subbasis for the topology of @xmath1585 .
we denote by @xmath1586 the canonical projection . for every @xmath1584 and @xmath1578
, we have @xmath1587 where @xmath1588 denotes the action of @xmath592 on @xmath1589 . since @xmath1590 ,
this set is open , i.e. , @xmath1583 is open .
now let @xmath1591 be an open set , and @xmath1592 .
it suffices to show that there exists a finite set @xmath1593 , such that the intersection @xmath1594 contains @xmath1595 and is contained in @xmath49 . to see this , we denote by @xmath1596 the distinct points
at which @xmath1595 does not vanish , and abbreviate @xmath1597 .
for @xmath440 we choose a neighborhood @xmath1598 of @xmath1599 , such that the sets @xmath1600 are disjoint , and @xmath1601 here the factor @xmath1602 occurs @xmath1603 times .
( the sets exist by hausdorffness of @xmath586 , openness of @xmath1604 , the definition of the product topology on @xmath1589 , and the fact @xmath1605 we define @xmath1606 .
we have @xmath1607
combining this with ( [ eq : u 1 u k ] ) , it follows that @xmath1608 . on the other hand , since @xmath1609 , we have @xmath1610 .
hence the set @xmath1611 has the required properties .
this proves lemma [ le : subbasis ] .
we will also use the following . [
le : x c 0 ] let @xmath1612 be a loop . then there exists a @xmath725-neighborhood @xmath1613 of @xmath196 , such that every @xmath1614 is homotopic to @xmath196 . we define @xmath1582 to be the open @xmath725-ball around @xmath196 , with radius @xmath1615 .
let @xmath1614 .
we define @xmath1616{\times}s^1\to{\mathbb c},\quad h(t , z):=tx'(z)+(1-t)x(z).\ ] ] this map does not vanish anywhere , and therefore is a homotopy between @xmath196 and @xmath1617 inside @xmath1535 .
this proves lemma [ le : x c 0 ] .
we canonically identify the set @xmath1618 of gauge equivalence classes of smooth triples @xmath101 ( as defined in ( [ eq : b ] ) ) with the set of gauge equivalence classes of smooth pairs @xmath1619 we show that * ( [ prop : loc conv g ] ) implies ( [ prop : loc conv deg])*. assume that ( [ prop : loc conv g ] ) holds .
we denote by @xmath1620 the map given by @xmath563 on @xmath1 , and @xmath1621 .
we define @xmath1622 this is a basis for the topology of @xmath19 .
let @xmath1623 be such that @xmath563 vanishes on @xmath1624 .
then there exists @xmath875 such that for every @xmath885 we have @xmath1625 , where @xmath1626 let @xmath1627 be as in the hypothesis of the claim .
consider first the * case @xmath1628 * , for some @xmath485 , @xmath136 .
we choose a ( smooth ) representative @xmath1629 of @xmath1630 .
since by assumption , @xmath563 vanishes on @xmath1631 , @xmath318 is nonzero on @xmath1632 .
the @xmath471-topology on the set of smooth connections on @xmath1633 is second - countable .
furthermore , the action of the gauge group on this space is continuous .
therefore , by assumption ( [ prop : loc conv g ] ) and lemma [ le : x g ] in appendix [ sec : add ] , for each @xmath132 , there exists a representative @xmath1634 of @xmath135 , such that @xmath517 converges to @xmath98 in @xmath471 , as @xmath134 .
hence , using lemma [ le : x c 0 ] , there exists @xmath875 such that for @xmath885 the restriction of @xmath726 to @xmath1632 does not vanish anywhere , and is homotopic to @xmath1635 , as a map with target @xmath1535 .
it follows that @xmath1636 for @xmath885 . by elementary arguments ,
the left hand side of this equality agrees with @xmath1637 , and the right hand side equals @xmath1638 . since by definition , @xmath1639 and @xmath1640
, it follows that @xmath1641 and therefore , @xmath1625 .
this proves the claim in the case @xmath1642 .
consider now the * case @xmath1643 * , for some @xmath1644 . by what we just proved
, there exists an index @xmath1237 , such that for every @xmath885 @xmath1645 it follows that @xmath1646 and therefore , @xmath1625 , for every @xmath885 .
this proves the claim in the case @xmath1643 .
the claim implies that if @xmath1647 and @xmath1584 are such that @xmath1648 then there exists @xmath875 such that for every @xmath885 we have @xmath1625 . combining this with lemma [ le : subbasis ] ( with @xmath1649 ) , condition ( [ prop : loc conv deg ] ) follows .
this proves that * ( [ prop : loc conv g ] ) implies ( [ prop : loc conv deg])*. we show the * opposite implication * : assume that ( [ prop : loc conv deg ] ) is satisfied .
we prove that given an open subset @xmath848 with compact closure and smooth boundary , the class @xmath1569 converges to @xmath479 , as @xmath134 , with respect to the topology @xmath505 . by lemma [ le : subsubseq ] ( appendix [ sec : add ] ) it suffices to show that for every such @xmath881 and every subsequence @xmath1650 there exists a further subsequence @xmath1651 such that @xmath1652 converges to @xmath479 , as @xmath1653 .
let @xmath1654 be a subsequence .
let @xmath820 .
we choose a representative @xmath1655 of @xmath1656 .
it follows from proposition [ prop : s 1 c image ] that the image of @xmath1319 is contained in the closed ball @xmath1657 .
, since then the image of @xmath1319 equals @xmath588 . ] furthermore , by proposition [ prop : e d ] , we have @xmath1658 .
therefore , proposition [ prop : cpt mod ] ( compactness modulo bubbling and gauge ) implies that there exist a smooth vortex @xmath1659 over @xmath1 , a subsequence @xmath1651 , and gauge transformations @xmath1660 ( for @xmath1661 ) , such that the pair @xmath1662 is smooth and converges to @xmath764 , as @xmath1653 , in @xmath471 on every compact subset of @xmath1 .
we denote by @xmath1663 the equivalence class of @xmath764 .
it follows from lemma [ le : g smooth ] ( appendix [ sec : add ] ) that @xmath1664 is smooth , and hence @xmath1665 represents @xmath1666 , for every @xmath1381 .
it follows that @xmath1667 converges to @xmath1668 with respect to @xmath505 , as @xmath1653 , for every open subset @xmath848 with compact closure and smooth boundary .
it remains to show that @xmath1669 .
the energy densities @xmath1670 converge to @xmath1671 in @xmath471 on compact subsets of @xmath1 .
therefore , we have @xmath1672 by proposition [ prop : e d ] we have @xmath1673 , and therefore , @xmath1674 .
hence @xmath1675 is well - defined .
it follows from the implication ( [ prop : loc conv g])@xmath1676([prop : loc conv deg ] ) ( which we proved above ) , that the point @xmath1677 converges to @xmath1678 , as @xmath1653 . by our assumption ( [ prop : loc conv deg ] ) , @xmath1679 also converges to @xmath1680 .
it follows that @xmath1681 , hence @xmath1682 , and therefore , @xmath1669 .
( here in the last step we used that the map ( [ eq : m d sym ] ) is injective . )
this proves that * ( [ prop : loc conv deg ] ) implies ( [ prop : loc conv g ] ) * and concludes the proof of proposition [ prop : loc conv s 1 c ] . for the proof of proposition [ prop : conv s 1 c ] we also need the following lemma .
[ le : q q ] let @xmath322 be a positive integer , @xmath1683 be sequences of mbius transformations , and let @xmath1684 be points .
assume that @xmath1685 , and that @xmath1686 uniformly on compact subsets of @xmath1687 , for @xmath1688 .
let @xmath1689 , @xmath1690 be compact subsets .
then for @xmath508 large enough we have @xmath1691 this follows from an elementary argument .
. then proposition [ prop : e d ] and ( [ eq : e v bar t ] ) imply that equality ( [ eq : deg w nu ] ) holds for @xmath508 large enough , as claimed . the second part of condition ( [ prop : conv s 1 c deg ] ) follows from proposition [ prop : loc conv s 1 c ] .
* suppose * now on the contrary that * condition ( [ prop : conv s 1 c deg ] ) holds*. then proposition [ prop : e d ] and equality ( [ eq : deg w nu ] ) imply ( [ eq : e v bar t ] ) .
let @xmath493 ( for @xmath396 ) be as in condition ( [ prop : conv s 1 c deg ] ) .
we show [ defi : conv]([defi : conv w ] ) : the first part of this condition ( concerning @xmath387 ) follows from proposition [ prop : loc conv s 1 c ] .
we prove the second part of the condition : let @xmath389 , and @xmath1692 be a compact subset .
we denote by @xmath1506 the orbit @xmath539 .
[ claim : u nu i j ] for every subsequence @xmath1650 there exists a further subsequence @xmath1651 , such that @xmath1693 we fix a subsequence @xmath1650 .
we choose a mbius transformation @xmath1694 such that @xmath1695 . by condition [ defi :
conv]([defi : conv phi z ] ) we have @xmath1696 and hence @xmath1697 restricts to an automorphism of @xmath1 .
we define @xmath1698 then @xmath1699 is an @xmath1700-vortex class over @xmath1 . we choose a representative @xmath1701 of @xmath1699 .
we check the hypotheses of proposition [ prop : cpt mod ] with the sequences @xmath1702 , @xmath1703 , and @xmath1704 . by condition [ defi :
conv]([defi : conv phi z ] ) , @xmath1705 converges to @xmath41 , as @xmath1706 .
furthermore , by proposition [ prop : s 1 c image ] the image of @xmath1707 is contained in the compact set @xmath1708 .
finally , proposition [ prop : e d ] implies that the energies of the vortices @xmath1709 ( @xmath820 ) are uniformly bounded .
hence the hypotheses of proposition [ prop : cpt mod ] are satisfied . applying this result ,
there exist a finite subset @xmath757 , an @xmath41-vortex @xmath759 , a subsequence @xmath1710 , and gauge transformations @xmath1711 , such that assertions of the proposition hold , with the sequence @xmath1712 replaced by @xmath1713 . by condition [ prop : cpt mod]([prop : cpt mod:= ] ) the map @xmath1714 converges to @xmath1715 in @xmath512 on every compact subset of @xmath1716 .
therefore , using the equality @xmath1695 , the statement of claim [ claim : u nu i j ] is a consequence of the following : [ claim : z z al ] the set @xmath1717 is contained in @xmath513 .
we choose a number @xmath567 so large that @xmath1718 let @xmath1719 be a vertex .
we denote @xmath1720 . by what we have already proved
, the restriction @xmath1721 converges to @xmath1722 , as @xmath134 , with respect to the topology @xmath1723 .
therefore , using lemmas [ le : x g ] ( appendix [ sec : add ] ) and [ le : conv e ] , it follows that @xmath1724 as @xmath134 .
we choose an index @xmath1725 so large that for @xmath1726 and every @xmath1719 we have @xmath1727 let @xmath1728 be vertices .
we denote by @xmath1729 the chain of vertices connecting @xmath1438 with @xmath572 , and write @xmath1730 , @xmath1731 . by conditions [ defi : st]([defi : st dist ] ) and [ defi : conv]([defi : conv al be ] ) the hypothesis of lemma [ le : q q ] with @xmath1732 are satisfied .
it follows that for every compact subset @xmath1733 and @xmath1734 , for @xmath508 large enough , we have @xmath1735 applying this several times with @xmath1736 and @xmath1737 , it follows that for @xmath508 large enough the sets @xmath1738 , @xmath1719 , are disjoint .
increasing @xmath1725 we may assume w.l.o.g .
that this holds for @xmath1739 .
therefore , for @xmath1726 , we have @xmath1740 here in the second line we used ( [ eq : e w nu be ] ) and in the third line we used ( [ eq : be v emin ] ) . our assumption ( [ eq : deg w nu ] ) and proposition [ prop : e d ]
imply that @xmath1741 for @xmath508 large enough .
hence ( [ eq : pi d emin ] ) implies that @xmath1742 for @xmath1381 large enough .
let now @xmath1743 .
we show that @xmath364 does not belong to @xmath1717 .
we choose a number @xmath296 so small that @xmath1744 .
we define @xmath1745 . by ( [ eq :
phi be nu q ] ) , we have @xmath1746 for @xmath508 large enough .
therefore , by ( [ eq : e w nu i j ] ) implies that @xmath1747 for @xmath1381 large enough . on the other hand ,
if @xmath364 belonged to @xmath1717 , then by condition [ prop : cpt mod]([prop : cpt mod lim nu e eps ] ) we would have @xmath1748 this contradiction proves that @xmath1749 .
it follows that @xmath1750 . since @xmath1695 , claim [ claim : z z al ] follows .
this concludes the proof of claim [ claim : u nu i j ] .
claim [ claim : u nu i j ] and an elementary argument imply that @xmath1751 for @xmath508 large enough .
furthermore , it follows from the same claim and lemma [ le : subsubseq ] in appendix [ sec : add ] that @xmath1752 in @xmath512 on @xmath506 , as @xmath134 .
this proves the second part of [ defi : conv]([defi : conv w ] ) .
hence all conditions of definition [ defi : conv ] are satisfied , and therefore , * condition [ prop : conv s 1 c]([prop : conv s 1 c w ] ) holds*. this concludes the proof of proposition [ prop : conv s 1 c ] .
in this chapter the equivariant homology class @xmath105 $ ] of an equivalence class @xmath50 of triples @xmath101 is defined , and the equivariant chern number of @xmath105 $ ] is interpreted as a certain maslov index .
this number entered the index formula ( [ eq : ind ] ) .
furthermore , the second main result of this memoir , theorem [ thm : fredholm ] , is proven .
it states the fredholm property for the vertical differential of the vortex equations over the plane @xmath1 , viewed as equations for equivalence classes @xmath50 of triples @xmath101 .
let @xmath1753 and @xmath42 be as in chapter [ chap : main ] , ) is satisfied .
] and @xmath1754 , equipped with the standard area form @xmath1755 .
we fix numbers @xmath1756 then every equivalence class @xmath175 of triples @xmath1757 ( defined as in ( [ eq : bb p lam ] ) ) carries an equivariant homology class @xmath105\in h_2^g(m,{\mathbb{z}})$ ] , whose definition is based on the following lemma .
let @xmath66 be a topological ( principal ) @xmath2-bundle over @xmath1 , and @xmath168 a continuous equivariant map . by an extension of the pair @xmath1758 to @xmath19 we mean a triple @xmath1759 , where @xmath1760 is a topological @xmath2-bundle over @xmath19 , @xmath1761 a morphism of topological @xmath2-bundles which descends to the inclusion @xmath1762 , and @xmath1763 a continuous @xmath2-equivariant map satisfying @xmath1764 .
[ le : p ] the following statements hold . [
le : p : exists ] for every triple @xmath1757 there exists an extension @xmath1759 of the pair @xmath1758 , such that @xmath2 acts freely on @xmath1765 , where @xmath1766 denotes the fiber of @xmath1760 over @xmath41 .
[ le : p : iso ] let @xmath66 be a topological @xmath2-bundle over @xmath1 , @xmath168 a continuous @xmath2-equivariant map , and @xmath1759 an extension of @xmath1758 , such that @xmath2 acts freely on @xmath1767 .
furthermore , let @xmath1768 be another such tuple , and @xmath89 an isomorphism of topological @xmath2-bundles , such that @xmath1769
. then there exists a unique isomorphism of topological @xmath2-bundles @xmath1770 , satisfying @xmath1771 the proof of this lemma is postponed to the appendix ( page ) . for the definition of @xmath105 $ ]
we also need the following .
[ rmk : homology ] let @xmath2 be a topological group , @xmath586 a closed oriented topological ( hausdorff ) manifold of dimension @xmath322 , @xmath1772 a topological @xmath2-bundle , @xmath1581 a topological space , equipped with a continuous action of @xmath2 , and @xmath1773 a @xmath2-equivariant map .
then @xmath318 carries an equivariant homology class @xmath1774_g\in h_k^g(y,{\mathbb{z}})$ ] , defined as follows .
we choose a universal @xmath2-bundle @xmath108 and a continuous @xmath2-equivariant map @xmath1775 .. ] the map @xmath1776 descends to a map @xmath1777 .
we define @xmath1778_g\in h_k^g(y,{\mathbb{z}})=h_k\big((y{\times}{{\operatorname{eg}}})/g,{\mathbb{z}}\big)\ ] ] to be the push - forward of the fundamental class of @xmath586 , under the map @xmath1779 .
this class does not depend on the choice of @xmath1780 , nor on @xmath1781 in the following sense .
if @xmath1782 is another universal @xmath2-bundle , then the corresponding class @xmath1774_g'\in h_k\big((y{\times}{{\operatorname{eg}}}')/g,{\mathbb{z}}\big)$ ] is mapped to @xmath1774_g$ ] under the canonical isomorphism to @xmath1781.]@xmath1783 [ defi : homology ] we define the _ equivariant homology class @xmath105\in h_2^g(m,{\mathbb{z}})$ ] _ of an element @xmath1784 to be the equivariant homology class of @xmath1763 , where @xmath1759 is an extension as in lemma [ le : p ] , corresponding to any representative @xmath101 of @xmath50 . here
the equivalence relation @xmath1785 is defined as in section [ sec : fredholm ] .
it follows from lemma [ le : p ] that this class does not depend on the choices of @xmath1759 and @xmath98 .
[ rmk : w well - defined ] the condition @xmath204 is needed for @xmath105 $ ] to be well - defined , for every @xmath175 .
consider for example the case @xmath1786 and @xmath305 .
let @xmath156 .
we choose a smooth map @xmath311 such that @xmath1787 then the equivalence class @xmath50 of @xmath1788 lies in @xmath158 , for every @xmath313 .
however , there is no extension of @xmath1758 as in lemma [ le : p ] , since @xmath1789 diverges , as @xmath1790 .
therefore , @xmath105 $ ] is not well - defined .
@xmath271 [ rmk : homology vortex ]
let @xmath156 and @xmath1791 be a gauge equivalence class of triples @xmath101 of sobolev class . that each such triple consists of a @xmath2-bundle of class @xmath166 , and a connection @xmath79 and a @xmath2-equivariant map @xmath168 of class @xmath167 .
] assume that every representative of @xmath50 satisfies the vortex equations ( [ eq : bar dd j a u],[eq : f a mu ] ) , the energy of @xmath50 is finite , and the closure of its image compact .
as explained in remark [ rmk : geometry ] in section [ sec : remarks ] , we have @xmath175 , for every @xmath217 .
therefore , the equivariant homology class of @xmath50 is well - defined .
@xmath271 the contraction appearing in formula ( [ eq : ind ] ) has a concrete geometric meaning .
it can be interpreted as the maslov index of a certain loop of linear symplectic transformations , as follows .
let @xmath0 be a symplectic manifold and @xmath2 a connected compact lie group , acting on @xmath4 in a hamiltonian way , with momentum map @xmath7 .
assume that @xmath2 acts freely on @xmath6 .
let @xmath65 be a compact , connected , oriented topological surface with non - empty boundary .
we associate to this data a map @xmath1792 , called `` maslov index '' , as follows .
the domain of this map consists of the weak @xmath1793-homotopy classes of @xmath1793-admissible maps from @xmath1794 , and it takes values in the even integers . here
we call a continuous map @xmath1795 @xmath1793-admissible , iff for every connected component @xmath586 of the boundary @xmath1796 there exists a @xmath2-orbit which is contained in @xmath6 , and contains the set @xmath1797 .
we denote by @xmath1798 the set of such maps .
we call two maps @xmath1799 _ weakly @xmath1793-homotopic _ , iff they are homotopic through such maps.{\times}{\sigma}\to m$ ] , such that @xmath1800 , for @xmath1801 , and for every @xmath1802 $ ] we have @xmath1803 . ]
this defines an equivalence relation on @xmath1798 .
we call the corresponding equivalence classes _ weak @xmath1793-homotopy classes_. [ xpl : s 1 c n ] consider @xmath1804 , equipped with the standard symplectic form @xmath1805 , and the action of @xmath539 given by @xmath1806 with momentum map @xmath1807 we have @xmath1808 , and the @xmath588-orbits contained in the unit sphere are the fibers of the hopf fibration @xmath1809 .
consider the case in which @xmath65 is the unit disk @xmath1810 .
then for each integer @xmath1811 and each vector @xmath1812 the map @xmath1813 is @xmath1814-admissible .
furthermore , @xmath1815 and @xmath1816 are weakly @xmath1814-homotopic if and only if @xmath1817 . to see this ,
note that if @xmath1817 then a homotopy between the two maps is given by @xmath1103{\times}{\mathbb{d}}\ni(t , z)\mapsto z^dv(t)$ ] , where @xmath1818,s^{2n-1}\big)$ ] is any path joining @xmath897 and @xmath1819 .
conversely , assume that @xmath1815 and @xmath1816 are weakly @xmath1814-homotopic
. then by an elementary argument , there exists a _ smooth
_ weak @xmath1814-homotopy @xmath1820{\times}{\mathbb{d}}\to{\mathbb c}^n$ ] between these maps .
it follows that @xmath1821{\times}{\partial}{\mathbb{d}}}h^*{\omega}_0=d\pi+0,\ ] ] where in the last equality we used the fact that for every @xmath1822 , @xmath1823 is tangent to the characteristic distribution on @xmath1824 , is contained in a hopf circle , for every @xmath1802 $ ] . ] and therefore @xmath1825 vanishes on vectors tangent to @xmath1824 .
therefore , we have @xmath1817 .
this proves the claimed equivalence . @xmath271
the definition of the map @xmath1792 is based on the following maslov index of a regular symplectic transport over a curve .
let @xmath586 be an oriented , connected , closed topological curve , and @xmath1826 a symplectic vector space .
we denote by @xmath1827 the group of linear symplectic automorphisms of @xmath49 .
we define a _ regular symplectic transport over @xmath586 _ to be a continuous map @xmath1828 satisfying @xmath1829 we define the _ maslov index _ of such a map to be @xmath1830 here @xmath1831 is an arbitrary point , @xmath1832 denotes the salamon - zehnder map ( see ( * ? ? ?
* theorem 3.1 . ) ) , and @xmath1833 the degree of a map from @xmath586 to @xmath588 .
this definition does not depend on the choice of @xmath779 , by some homotopy argument .
let now @xmath1834 be as before , and @xmath1835 a weak @xmath1793-homotopy class . to define @xmath1836 , we choose a representative @xmath1795 of @xmath1835 , a symplectic vector space @xmath1837 of dimension @xmath1838 and a symplectic trivialization @xmath1839
. is connected , and the surface @xmath65 deformation retracts onto a wedge of circles .
here we use that @xmath65 is connected and has non - empty boundary .
] let @xmath586 be a connected component of @xmath1796 .
we define the map @xmath1840 here @xmath1841 denotes the unique element satisfying @xmath1842 , @xmath1843 denotes the induced action of @xmath1844 on @xmath1845 , and @xmath1846 .
the map @xmath1847 is a regular symplectic transport .
[ defi : maslov ] we define @xmath1848 where the sum runs over all connected components @xmath586 of @xmath1796 .
this definition does not depend on the choices of the symplectic vector space @xmath1837 and the trivialization @xmath1849 .
this follows from the fact that ( [ m si om mu a ] ) agrees with the maslov index associated with @xmath1835 and the coisotropic submanifold @xmath1850 , as defined in @xcite
. see ( * ? ? ?
* lemma 45 ) .
this maslov index is based on a definition by d. a. salamon and r. gaio ( for @xmath1851 ) , see @xcite .
@xmath271 [ xpl : u
d v ] consider the situation of example [ xpl : s 1 c n ] , with @xmath1815 defined as in ( [ eq : u ] ) .
this map carries the maslov index @xmath1852)=2dn,\ ] ] where @xmath208 $ ] denotes the @xmath1814-homotopy class .
this follows from a straight - forward calculation .
@xmath271 we now return to the setting of the beginning of this section .
we define @xmath65 to be the compact oriented topological surface obtained from @xmath1 by `` gluing a circle at @xmath41 '' to it .. ] the class @xmath50 carries a weak @xmath1793-homotopy class , whose definition relies on the following lemma . [
le : section ] let @xmath156 , @xmath204 , and @xmath1757 .
there exists a section @xmath553 of class @xmath167 , such that the map @xmath1853 continuously extends to @xmath65 .
furthermore , for every such section @xmath560 the corresponding extension @xmath1854 is @xmath1793-admissible .
the proof of this lemma is postponed to the appendix ( page ) .
let @xmath1855 [ defi : homotopy w ] we define the _ @xmath1793-homotopy class of @xmath50 _ to be the weak @xmath1793-homotopy class of the continuous extension @xmath1854 of the map @xmath1853 . here
@xmath101 is a representative of @xmath50 , and @xmath560 a section of @xmath66 as in the previous lemma . by lemma [ le : homotopic ] in appendix [ sec : proofs homology ] this homotopy class is independent of the choice of the section @xmath560 .
furthermore , it follows from a straight - forward argument that it is independent of the choice of the representative @xmath101 .
the contraction appearing in formula ( [ eq : ind ] ) can now be expressed as follows .
[ prop : chern maslov ] we have @xmath1856\big{\rangle}=m_{{\sigma},{\omega},\mu}\big(({\omega},\mu)\textrm{-homotopy class of } w\big).\ ] ] the proof of this result is postponed to the appendix ( page ) .
let @xmath537 , @xmath62 be the standard inner product on @xmath538 , the action of @xmath539 on @xmath540 be given by multiplication , with momentum map as in ( [ eq : mu c n ] ) with @xmath1857 , @xmath85 , equipped with the area form @xmath272 , and @xmath50 a gauge equivalence class of smooth finite energy vortices . by proposition [ prop : s 1 c image ] and remark [ rmk : homology vortex ]
the equivariant homology class @xmath105 $ ] is well - defined .
furthermore , we have @xmath1858\big{\rangle}=\deg(w),\ ] ] where the degree @xmath1859 is defined as in ( [ eq : deg w sum ] ) .
this follows from proposition [ prop : chern maslov ] and a straight - forward calculation of the maslov index of the @xmath1793-homotopy class of @xmath50 .
@xmath271 in this section theorem [ thm : fredholm ] is proved , based on a fredholm theorem for the augmented vertical differential and the existence of a bounded right inverse for @xmath219 , the formal adjoint of the infinitesimal action of the gauge group on pairs @xmath257 . in the present section
we always assume that @xmath1860 in this subsection we state the two results mentioned above and deduce theorem [ thm : fredholm ] from them . to formulate the first result ,
let @xmath156 , @xmath210 , @xmath294 be defined as in ( [ eq : bb p lam ] ) , and @xmath184 .
we denote @xmath1861 and by @xmath1862 the orthogonal projection onto @xmath1863 .
@xmath1864 induces an orthogonal projection @xmath1865 onto @xmath1866 .
recall that @xmath1867 denotes the bundle of one - forms on @xmath1868 with values in @xmath83 .
we write @xmath1869 note that @xmath1863 is in general not a subbundle of @xmath1845 , since the dimension of @xmath1870 may vary with @xmath74 . recall that @xmath182 denotes the space of @xmath167-sections of a vector bundle @xmath133 .
for @xmath1871 we define @xmath1872 where @xmath328 is defined as in ( [ eq : ze w p lam ] ) .
recall the definition ( [ eq : yy w ] ) of @xmath206 .
we denote by @xmath1873 the space of @xmath1874-sections of @xmath83 .
we define @xmath1875 recall the definition ( [ eq : l w * ] ) of the formal adjoint map for the infinitesimal action of the gauge group on pairs @xmath257 . restricting the domain and target ,
this becomes the operator @xmath1876 it follows from the fact @xmath1877 ( for every @xmath74 ) and compactness of @xmath1878 that this operator is well - defined and bounded .
we define the _ augmented vertical differential of the vortex equations over @xmath1 at @xmath98 _ to be the map @xmath1879 [ thm : fredholm aug ] let @xmath156 and @xmath1880 be real numbers , and @xmath1881 . then the following statements hold . [ thm : fredholm x w ] the normed spaces @xmath1882 and @xmath1873 are complete .
[ thm : fredholm dd w ] assume that @xmath1883 .
then the operator @xmath1884 ( as in ( [ eq : dd w p lam],[eq : dd w ze ] ) ) is fredholm of real index @xmath1885\big{\rangle},\ ] ] where @xmath1886 $ ] denotes the equivalence class of @xmath98 , and @xmath1887 $ ] denotes the equivariant homology class of @xmath1886 $ ] .
this theorem will be proved in section [ subsec : proof : fredholm aug ] , based on the existence of a suitable trivialization of @xmath220 in which the operator @xmath1884 becomes standard up to a compact perturbation .
the second ingredient of the proof of theorem [ thm : fredholm ] is the following .
[ thm : l w * r ] let @xmath156 , @xmath204 , and @xmath184 .
then the map @xmath1888 admits a bounded ( linear ) right inverse .
the proof of theorem [ thm : l w * r ] is postponed to section [ subsec : proof : thm : l w * r ] ( page ) .
it is based on the existence of a bounded right inverse for the operator @xmath1889 over a compact subset of @xmath1890 diffeomorphic to @xmath1891 ( proposition [ prop : right ] ) and the existence of a neighborhood @xmath1892 of @xmath6 , such that @xmath1893 recall the definition ( [ eq : m * ] ) of the subset @xmath1894 of @xmath4 .
theorem [ thm : fredholm ] can be reduced to theorems [ thm : fredholm aug ] and [ thm : l w * r ] , as follows :
let @xmath156 , @xmath204 , and @xmath175 .
* we prove statement ( [ thm : fredholm x w])*. we have @xmath1895 and the restriction of the norm @xmath1896 to @xmath1897 is equivalent to @xmath331 .
it suffices to prove that @xmath1898 and this inclusion is bounded .
hypothesis ( h ) implies that there exists @xmath993 such that @xmath1899 .
we have @xmath1900 it follows from lemma [ le : si ] in appendix [ sec : proofs homology ] that there exists @xmath567 such that @xmath1901 .
let @xmath1902 .
then @xmath1903 , and thus , using the last assertion of remark [ rmk : c ] in appendix [ sec : add ] , @xmath1904 hence @xmath1898 , and this inclusion is bounded .
this proves the claim .
* statement ( [ thm : fredholm x w ] ) * follows from theorem [ thm : fredholm aug]([thm : fredholm : x y ] ) and the claim .
* statement ( [ thm : fredholm : ddd ] ) * follows from theorem [ thm : fredholm aug]([thm : fredholm dd w ] ) , theorem [ thm : l w * r ] and lemma [ le : x y z ] ( appendix [ sec : add ] ) with @xmath1905 where @xmath1906 is defined to be the vector consisting of the first and second rows of @xmath1907 ( as in ( [ eq : dd w ze ] ) ) .
this proves theorem [ thm : fredholm ] .
this subsection contains the core of the proof of theorem [ thm : fredholm aug ] .
here we introduce the notion of a good complex trivialization , and state an existence result for such a trivialization ( proposition [ prop : triv ] ) .
furthermore , we formulate a result saying that every good trivialization transforms @xmath1908 into a compact perturbation of the direct sum of copies of @xmath226 and a certain matrix operator ( proposition [ prop : x x w ] ) .
the results of this subsection will be proved in subsection [ subsec : proofs ] .
we denote by @xmath277 and @xmath278 the standard coordinates in @xmath540 . for @xmath1909 and @xmath1910
we denote @xmath1911 we equip the bundle @xmath1867 with the ( fiberwise ) complex structure @xmath1912 defined by @xmath1913 .
furthermore , we denote @xmath1914 and for @xmath1915 we use the notation @xmath1916 for @xmath74 we write @xmath1917 for the complex linear extension of @xmath1918 . we define @xmath1919 note that in general , the union @xmath1920 of all the @xmath1921 s is not a smooth subbundle of @xmath1845 , since the dimension of @xmath1921 may depend on @xmath196 .
however , there exists an open neighborhood @xmath1892 of @xmath6 such that @xmath1922 is a subbundle of @xmath1923 .
let @xmath156 , @xmath1880 and @xmath184 .
we denote @xmath1924\big{\rangle}. \ ] ] for @xmath485 we define @xmath1925 consider a complex trivialization ( i.e , a bundle isomorphism descending to the identity on the base @xmath1 ) @xmath1926 [ defi : triv ] we call @xmath1849 _ good _ , if the following properties are satisfied . *
( splitting)*[defi : triv split ] for every @xmath485 we have @xmath1927 furthermore , there exists a number @xmath567 , a section @xmath560 of @xmath1928 , of class @xmath167 , and a point @xmath1929 , such that the following conditions are satisfied . for every @xmath222 we have @xmath1930 and for every @xmath222 and @xmath1931 , we have @xmath1932 [ defi : triv c ] there exists a number @xmath345 such that for every @xmath1933 @xmath1934 [ defi : triv na ] we have @xmath1935 .
the first ingredient of the proof of theorem [ thm : fredholm aug ] is the following result .
[ prop : triv ] if @xmath156 , @xmath1880 , and @xmath184 , then there exists a good complex trivialization of @xmath220 .
the proof of this proposition is postponed to subsection [ subsec : proofs ] ( page ) .
the next result shows that a good trivialization transforms @xmath1884 into a compact perturbation of some standard operator .
we denote @xmath1936 and @xmath1937 let @xmath1938 , @xmath1939 , @xmath148 , @xmath210 , @xmath1940 be an open subset , @xmath50 a real or complex vector space , and @xmath1941 a @xmath322-times weakly differentiable map .
we define @xmath1942,\\ { { \nonumber}}\vert u\vert_{w^{k , p}_{\lambda}({\omega},w)}&:=&\sum_{|{{\alpha}}|\leq k}\vert\langle\cdot\rangle^{\lambda}{\partial}^{{\alpha}}u\vert_{l^p({\omega},w)}\in[0,\infty],\\ \label{eq : l k p}l^{k , p}_{\lambda}({\omega},w)&:=&\big\{u\in w^{k , p}_{{\operatorname{loc}}}({\omega},w)\,|\,\vert u\vert_{l^{k , p}_{\lambda}({\omega},w)}<\infty\big\}\\ \label{eq : w k p } w^{k , p}_{\lambda}({\omega},w)&:=&\big\{u\in w^{k , p}_{{\operatorname{loc}}}({\omega},w)\,|\,\vert u\vert_{w^{k , p}_{\lambda}({\omega},w)}<\infty\big\}.\end{aligned}\ ] ] if @xmath1943 , @xmath440 , are normed vector spaces then we endow @xmath1944 with the norm @xmath1945 .
let @xmath1811 . if @xmath1946 then we choose @xmath1947)$ ] such that @xmath1948 for @xmath1949 and @xmath1950 for @xmath1951 . in the case
@xmath1952 we set @xmath1953 .
the isomorphism of lemma [ le : x d iso ] ( appendix [ sec : weighted ] ) induces norms on the vector spaces @xmath1954 we define @xmath1955 @xmath1956 for a complex vector space @xmath50 we denote by @xmath1957 ( @xmath1958 ) the operator @xmath1959 ( @xmath1960 ) acting on functions from @xmath1 to @xmath50 .
we denote by @xmath1961 the hermitian inner product on @xmath1962 ( complex anti - linear in its first argument ) extending @xmath62 .
furthermore , we denote by @xmath1963 the bundle of complex anti - linear one - forms on @xmath1 with values in @xmath189 , and define the isomorphisms @xmath1964 we are now ready to formulate the second ingredient of the proof of theorem [ thm : fredholm aug ] : [ prop : x x w ] let @xmath156 , @xmath1880 , @xmath1965 , and @xmath1966 be a good trivialization .
we define @xmath591 as in ( [ eq : d lan ] ) .
the following statements hold . [
prop : x x w iso ] the following maps are well - defined isomorphisms of normed spaces : @xmath1967 [ prop : x x w fredholm ] there exists a positive for every @xmath1968 . ]
@xmath1-linear map @xmath1969 such that the following operator is compact : @xmath1970 the proof of proposition [ prop : x x w ] is postponed to subsection [ subsec : proofs ] ( page ) .
it is based on some inequalities and compactness properties for weighted sobolev spaces and a hardy - type inequality ( propositions [ prop : lam d morrey ] and [ prop : hardy ] in appendix [ sec : weighted ] ) .
[ proof : fredholm aug ] we fix @xmath156 , @xmath1880 , and a triple @xmath1971 .
we prove * part ( [ thm : fredholm x w])*. the space ( [ eq : l k p ] ) is complete , see @xcite .
the same holds for the space ( [ eq : w k p ] ) by proposition [ prop : lam d morrey]([prop : lam w ] ) ( appendix [ sec : weighted ] ) . combining this with propositions [ prop : triv ] and [ prop : x x w]([prop :
x x w iso ] ) , part ( [ thm : fredholm x w ] ) follows .
* part ( [ thm : fredholm dd w ] ) * follows from propositions [ prop : triv ] and [ prop : x x w]([prop : x x w fredholm ] ) , corollary [ cor : d l l ] and proposition [ prop : d a b d ] ( appendix [ sec : weighted ] ) .
this proves theorem [ thm : fredholm aug ] .
[ rmk : fredholm aug ] an alternative approach to prove theorem [ thm : fredholm aug ] is to switch to `` logarithmic '' coordinates @xmath337 ( defined by @xmath1972 ) . in these coordinates and a suitable trivialization the operator @xmath1884 is of the form @xmath1973 .
hence one can try to apply the results of @xcite .
however , this is not possible , since @xmath1974 contains the operator @xmath1975 , which diverges for @xmath1976 . @xmath271 [ proof : triv ] let @xmath1977 and @xmath98 be as in the hypothesis .
we choose a section @xmath560 of @xmath1978 and a point @xmath1929 as in lemma [ le : si ] in appendix [ sec : proofs homology ] .
[ claim : u triv ] there exists an open @xmath2-invariant neighborhood @xmath1892 of @xmath1979 such that @xmath1922 is a smooth subbundle of @xmath1845 with the following property .
there exists a smooth complex trivialization @xmath1980 satisfying @xmath1981 by hypothesis ( h ) we have @xmath1982 ( where @xmath1894 is defined as in ( [ eq : m * ] ) ) .
we choose a @xmath2-invariant neighborhood @xmath1983 of @xmath1979 so small that @xmath1984 and @xmath1985 intersect transversely , for every @xmath1986 .
then @xmath1987 is a smooth subbundle of @xmath1988 .
furthermore , by the local slice theorem there exists a pair @xmath1989 , where @xmath1990 is a @xmath2-invariant neighborhood of @xmath1979 and @xmath1991 is a submanifold of dimension @xmath1992 that intersects @xmath361 transversely in exactly one point , for every @xmath1993 .
we choose a complex trivialization of @xmath1994 and extend it in a @xmath2-equivariant way , thus obtaining a trivialization @xmath1995 of @xmath1922 .
this proves claim [ claim : u triv ] .
we choose @xmath1627 and @xmath1995 as in claim [ claim : u triv ] .
it follows from lemma [ le : si ] that there exists @xmath1996 such that @xmath1997 , for @xmath223 , if @xmath222 .
we define @xmath1998 this is a complex trivialization of @xmath1999 ( of class @xmath167 ) .
[ claim : wt psi 1 ] @xmath2000 extends to a complex trivialization of @xmath189 .
we define @xmath2001 by @xmath2002 .
we choose a complex trivialization @xmath2003 of class @xmath167 . of the bundle .
an argument using local trivializations of class @xmath167 , shows that we may regularize @xmath2004 , so that it becomes of class @xmath167 .
] we define @xmath2005 for a continuous map @xmath2006 we denote by @xmath2007 its degree . [
claim : lan deg ] we have @xmath2008\big{\rangle}=\deg(f\circ\det\circ\phi ) . \ ] ] we define @xmath1760 to be the quotient of @xmath2009 under the equivalence relation generated by @xmath2010 , where @xmath288 is determined by @xmath2011 , for @xmath2012 , @xmath2013 .
furthermore , we define @xmath2014):=\left\{\begin{array}{ll}u(p),&\textrm{for } p\in p,\\ { { \widetilde{u}}}([\infty , g]):=g^{-1}x_\infty,&\textrm{for } g\in g.\end{array}\right.\ ] ] the statement of lemma [ le : si ] implies that this map is continuous and extends @xmath318 . the fiberwise pullback form @xmath2015 on @xmath1760 descends to a symplectic form @xmath2016 on the vector bundle @xmath2017 .
similarly , the structure @xmath42 induces an complex structure @xmath2018 on @xmath2019 .
the structures @xmath2016 and @xmath2018 are compatible , and therefore , we have @xmath2020 using lemma [ le : chern bundle ] in appendix [ sec : proofs homology ] , it follows that @xmath2021\big{\rangle}=\big{\langle}c_1\big(tm^{{{\widetilde{u}}}},{{\widetilde{j}}}\big),[s^2]\big{\rangle}. \ ] ] we define the local complex trivialization @xmath2022,\psi^u_{u\circ{\sigma}(z)}v_0+l^{\mathbb c}_{u\circ{\sigma}(z)}{{\alpha}}\big),&\textrm{if } z\in { \mathbb c}{\setminus}b_r,\\ g\cdot\big([\infty,{\mathbf{1}}],\psi^u_{x_\infty}v_0+l^{\mathbb c}_{x_\infty}{{\alpha}}\big),&\textrm{if } z=\infty .
\end{array}\right . \ ] ] recalling the definition ( [ eq : phi s 1 r ] ) , we have @xmath2023 therefore , @xmath2024 is the transition map between @xmath2025 and @xmath2026 .
it follows that @xmath2027\big{\rangle}=\deg(f\circ\det\circ\phi).\ ] ] combining this with ( [ eq : c 1 g ] ) , equality ( [ eq : c 1 g deg ] ) follows .
this proves claim [ claim : lan deg ] .
we denote @xmath2028\big{\rangle}$ ] . by claim [ claim : lan deg ] and lemma [ le : aut ] in appendix
[ sec : add ] the maps @xmath2024 and @xmath2029 are homotopic .
hence there exists a continuous map @xmath2030 such that @xmath2031 , if @xmath2032 , and @xmath2033 , if @xmath2034 .
we define @xmath2035 regularizing @xmath2036 on the ball @xmath2037 , we obtain the required extension of @xmath2000 , of class @xmath167 .
this proves claim [ claim : wt psi 1 ] .
we define the trivialization @xmath2038 [ claim : hhat psi 1 ] @xmath2039 extends to a complex trivialization of the bundle @xmath1867 over @xmath1 .
we denote by @xmath2040 and @xmath2041 the adjoint representations of @xmath2 on @xmath3 and @xmath1962 respectively .
we have @xmath2042 we choose a continuous section @xmath2043 of the restriction @xmath2044 .
we define @xmath2045 to be the unique map such that @xmath2046 , for every @xmath2034 .
since @xmath2047 , we have @xmath2048 hence lemma [ le : aut ] ( appendix [ sec : add ] ) implies that there exists a continuous map @xmath2049 satisfying @xmath2050 , for every @xmath2034 .
we define @xmath2051 to be the trivialization that equals @xmath2052 on @xmath2053 , and satisfies @xmath2054 where @xmath2055 , for every @xmath2056 , @xmath2057 .
regularizing @xmath2058 on the ball @xmath2037 , we obtain an extension of @xmath2039 , of class @xmath167 .
this proves claim [ claim : hhat psi 1 ] .
we choose extensions @xmath2036 and @xmath2058 of @xmath2059 and @xmath2052 as in claims [ claim : wt psi 1 ] and [ claim : hhat psi 1 ] , and we define @xmath2060 [ claim : psi good ] the map @xmath1849 is a good complex trivialization
. condition * ( [ defi : triv split ] ) * of definition [ defi : triv ] follows from the construction of @xmath1849 .
( the condition @xmath2061 follows from the statement of lemma [ le : si ] . ) to prove * ( [ defi : triv c ] ) * , note that for @xmath2062 and @xmath2063 , we have @xmath2064 here we used the fact @xmath2065 , for every @xmath74 . by our choice of @xmath1627
, @xmath2066 is a smooth subbundle of rank @xmath2067 .
it follows that @xmath2068 is a smooth subbundle of @xmath1923 of rank @xmath2069 .
hence @xmath2070 is injective , for every @xmath1993 .
since by assumption @xmath2071 is compact , the same holds for the set @xmath2072 .
it follows that there exists a constant @xmath345 such that @xmath2073 for every @xmath2074 , @xmath2075 , and @xmath2076 . combining this with equality ( [ eq : psi z z ] ) , condition ( [ defi : triv c ] ) follows .
we check condition * ( [ defi : triv na])*. let @xmath2077 we choose a point @xmath223 and a vector @xmath2078 such that @xmath2079 .
then @xmath2080 where @xmath2081 is defined as in ( [ eq : wt na a ] ) .
furthermore , for every smooth vector field @xmath586 on @xmath1627 we have @xmath2082 we define @xmath2083 where the maximum is taken over all @xmath2084 , @xmath2085 and @xmath2086 such that @xmath2087 , @xmath2088 .
furthermore , we define @xmath2089 it follows that @xmath2090 we now define @xmath2091 to be the unique equivariant maps such that @xmath2092 , @xmath2093 .
we have @xmath2094 $ ] , and similarly for @xmath2095 .
since @xmath2096 using ( [ eq : hhat psi infty ] ) , it follows that @xmath2097\big|.\ ] ] using the fact @xmath2098 and inequality ( [ eq : vert na a v wt psi ] ) , condition ( [ defi : triv na ] ) follows .
this proves claim [ claim : psi good ] and concludes the proof of proposition [ prop : triv ] .
[ proof : x x w ] let @xmath2099 and @xmath1849 be as in the hypothesis .
we choose @xmath1947)$ ] such that @xmath1948 for @xmath2100 and @xmath1950 for @xmath299 .
we fix @xmath2101 , @xmath560 and @xmath1979 as in definition [ defi : triv]([defi : triv split ] ) .
* we prove statement ( [ prop : x x w iso ] ) .
* for every @xmath2102 leibnitz rule implies that @xmath2103 [ claim : f x bdd ] the first map in ( [ eq : x y psi ] ) is well - defined and bounded .
proposition [ prop : lam d morrey]([prop : lam morrey ] ) in appendix [ sec : weighted ] and the fact @xmath1880 imply that there exists a constant @xmath2104 such that @xmath2105 we choose a constant @xmath2106 as in part ( [ defi : triv c ] ) of definition [ defi : triv ] . then by ( [ eq : c ] ) and ( [ eq : lan ran - d i d ze ] ) , we have @xmath2107 it follows from ( [ eq : psi infty x h ] ) and ( [ eq : psi 0 al be ] ) , the definition @xmath2108 and the compactness of @xmath1878 that there exists @xmath2109 such that , for every @xmath2110 , @xmath2111 where @xmath2112 . for @xmath136 we denote @xmath2113 we define @xmath2114 by condition ( [ defi : triv na ] ) of definition [ defi : triv ] we have @xmath2115 .
let @xmath2110 .
then by ( [ eq : na phi psi f ze ] ) we have @xmath2116 defining @xmath2117 we have , by proposition [ prop : lam d morrey]([prop : lam d iso ] ) , @xmath2118 combining this with ( [ eq : na a psi ze p lam ] ) and ( [ eq : lan ran - d i d ze ] ) , we get @xmath2119 by a direct calculation there exists a constant @xmath2120 such that @xmath2121 claim [ claim : f x bdd ] follows from this and ( [ eq : psi ze infty],[eq : d mu u v],[eq : na a psi ze ] ) . [
claim : f -1 x bdd ] the map @xmath2122 is well - defined and bounded .
we choose a neighborhood @xmath1892 of @xmath6 as in lemma [ le : u ] ( appendix [ sec : add ] ) , and define @xmath2123 as in ( [ eq : c inf ] ) , and @xmath2124 .
since @xmath2125 converges to @xmath1979 , uniformly in @xmath2126 , as @xmath2127 , there exists @xmath2128 such that @xmath1997 , for every @xmath2129
. then ( [ eq : psi infty x h],[eq : psi 0 al be ] ) and ( [ eq : c inf ] ) imply that @xmath2130 where @xmath2131 , for every @xmath2132 .
[ claim : d rho 0 p ] there exists a constant @xmath2133 such that for every @xmath2134 , we have @xmath2135 it follows from equality ( [ eq : na phi psi f ze ] ) and conditions ( [ defi : triv c ] ) and ( [ defi : triv na ] ) of definition [ defi : triv ] that there exist constants @xmath297 and @xmath2136 such that @xmath2137 on the other hand , leibnitz rule implies that @xmath2138 hence by a short calculation , using leibnitz rule again , it follows that there exists a constant @xmath2139 such that @xmath2140 combining this with ( [ eq : p - d i d f ] ) , claim [ claim : d rho 0 p ] follows .
let @xmath2134 .
we denote @xmath2141 by inequality ( [ eq : d rho 0 p ] ) the hypotheses of proposition [ prop : hardy ] in appendix [ sec : weighted ] with @xmath2142 and @xmath157 replaced by @xmath2143 are satisfied .
it follows that there exists @xmath2144 such that @xmath2145 , uniformly in @xmath2126 , as @xmath2127 , and @xmath2146 since @xmath1880 , we have @xmath2147 hence the convergence @xmath2148 and the estimate ( [ eq : vert al be vert ] ) imply that @xmath2149 .
we choose a constant @xmath345 as in part ( [ defi : triv c ] ) of definition [ defi : triv ] .
the convergence @xmath2150 and the first inequality in ( [ eq : c ] ) imply that @xmath2151 we define @xmath2152 proposition [ prop : lam d morrey]([prop : lam d iso ] ) in appendix [ sec : weighted ] and inequalities ( [ eq : wt ze ze infty ] ) and ( [ eq : d rho 0 p ] ) imply that there exists a constant @xmath2120 ( depending on @xmath2153 and @xmath1849 , but not on @xmath2154 ) such that @xmath2155 finally , by a straight - forward argument , there exists a constant @xmath2156 ( independent of @xmath2154 ) such that @xmath2157 combining this with ( [ eq : vert al be vert],[eq : v infty 1],[eq : v 1 v 2 v bar n ] ) , claim [ claim : f -1 x bdd ] follows .
claims [ claim : f x bdd ] and [ claim : f -1 x bdd ] imply that the first map in ( [ eq : x y psi ] ) is an isomorphism ( of normed vector spaces ) .
it follows from condition ( [ defi : triv c ] ) of definition [ defi : triv ] that the second map in ( [ eq : x y psi ] ) is an isomorphism .
this completes the proof of * ( [ prop : x x w iso])*. * we prove statement ( [ prop : x x w fredholm ] ) .
* recall that we have chosen @xmath2158 and @xmath1979 as in definition [ defi : triv]([defi : triv split ] ) .
we define @xmath2159 to be the complex linear extension of @xmath2160 . by our hypothesis ( h )
the lie group @xmath2 acts freely on @xmath6 .
it follows that @xmath2161 is injective .
therefore @xmath2162 is positive with respect to @xmath1961 . by ( [ eq : psi z c bar n ] ) and ( [ eq : psi z 0 ] ) there
exist complex trivializations @xmath2163 such that @xmath2164 .
we denote by @xmath2165 the canonical inclusion and projection .
we define @xmath2166 note that @xmath2167 and @xmath2168 .
we define @xmath2169 as in ( [ eq : f psi dd ] ) .
since @xmath2170 is finite dimensional , @xmath2171 is compact .
hence it suffices to prove that @xmath2172 is compact . to see this ,
we denote @xmath2173 and we define @xmath2174 ( for @xmath2175 ) and @xmath2176 by @xmath2177 here @xmath2178 , for @xmath2076 ( viewed as a constant section of @xmath2179 ) .
( recall also that @xmath2159 is the complex linear extension of @xmath2160 . )
a direct calculation shows that @xmath2180 for a subset @xmath1373 we denote by @xmath2181 its characteristic function .
it follows that @xmath2182 is of 0-th order .
since it vanishes outside @xmath2183 , it follows that this map is compact .
[ claim : s i j compact ] the operators @xmath2184 , @xmath2175 , and @xmath2185 are compact . to see that @xmath2186 is compact , note that leibnitz rule and holomorphicity of @xmath2187
imply that @xmath2188 for a normed vector space @xmath49 we denote by @xmath2189 the space of bounded continuous maps from @xmath1 to @xmath49 .
since @xmath204 , assertions ( [ prop : lam d iso ] ) and ( [ prop : lam morrey ] ) of proposition [ prop : lam d morrey ] imply that the map @xmath2190 is well - defined and compact . by definition [ defi : triv]([defi : triv na ] ) , the map @xmath2191 is bounded .
condition ( [ defi : triv c ] ) of definition [ defi : triv ] implies boundedness of the map @xmath2192 compactness of @xmath2186 follows . by the definition of @xmath294
, we have @xmath2193 .
this together with proposition [ prop : lam d morrey]([prop : lam d iso ] ) and ( [ prop : lam morrey ] ) and definition [ defi : triv]([defi : triv c ] ) implies that the map @xmath2185 is compact .
furthermore , it follows from definition [ defi : triv]([defi : triv split ] ) that @xmath2194 . to see that @xmath2195 is compact , we define @xmath2196 by setting @xmath2197 to be the complex linear extension of the map @xmath2198 since @xmath2125 converges to @xmath1979 , uniformly in @xmath44 , as @xmath2127 , the map @xmath2199 converges to 0 , uniformly in @xmath44 , as @xmath2200 . hence by proposition [ prop : lam d morrey]([prop : lam w cpt ] )
, the map @xmath2201 is compact .
definition [ defi : triv]([defi : triv split ] ) implies that @xmath2202 .
it follows that this operator is compact .
finally , proposition [ prop : lam d morrey]([prop : lam morrey ] ) and parts ( [ defi : triv na ] ) and ( [ defi : triv c ] ) of definition [ defi : triv ] imply that the map @xmath2203 is compact .
claim [ claim : s i j compact ] follows .
it follows that the operator @xmath2169 ( as in ( [ eq : f psi dd ] ) ) is compact .
this completes the proof of statement * ( [ prop : x x w fredholm ] ) * and hence of proposition [ prop : x x w ] . for the proof of this result
we need the following .
let @xmath1939 , @xmath444 , @xmath2204 , @xmath2 be a compact lie group , @xmath62 an invariant inner product on @xmath2205 , @xmath586 a manifold ( possibly with boundary ) , and @xmath1772 a @xmath2-bundle of class @xmath2206 .
we denote by @xmath2207 the adjoint bundle , and by @xmath2208 the affine space of connections on @xmath66 of class @xmath2209 , i.e. , of class @xmath2210 on every compact subset of @xmath586 .
if @xmath586 is compact then we abbreviate @xmath2211 .
let @xmath2212 be a riemannian metric on @xmath586 and @xmath2213 .
the connection @xmath79 induces a connection @xmath194 on the adjoint bundle @xmath2214 . for
every @xmath820 this connection and the levi - civita connection of @xmath2212 induce a connection @xmath2215 on the bundle @xmath2216 .
we abbreviate these connections by @xmath188 .
let @xmath2217 .
for a vector bundle @xmath133 over @xmath586 we denote by @xmath2218 the space of @xmath2219-sections of @xmath133 .
for @xmath2220 we define @xmath2221 where @xmath2222 denotes the pointwise norm induced by @xmath2212 and @xmath62 , and @xmath2223 denotes the @xmath161-norm of a function on @xmath586 , induced by @xmath2212 .
we denote by @xmath2224 the bundle of @xmath179-forms on @xmath586 with values in @xmath83 , and @xmath2225 we denote by @xmath2226 the formal adjoint of @xmath194 , with respect to the @xmath1141-metrics induced by @xmath2212 and @xmath62 . here
@xmath192 denotes the hodge star operator .
[ prop : right ] let @xmath2227 and @xmath62 be as above , @xmath2228 , and @xmath2229 a riemannian manifold .
assume that @xmath586 is diffeomorphic to @xmath2230 .
then the following statements hold . [
prop : right : exists ] for every @xmath2-bundle @xmath1772 of class @xmath867 and every connection one - form @xmath79 on @xmath66 of class @xmath332 there exists a bounded right inverse of the operator ( [ eq : d a * ] ) .
[ prop : right : eps c ] there exist constants @xmath296 and @xmath345 such that for every @xmath2-bundle @xmath1772 of class @xmath867 , and every @xmath2231 satisfying @xmath2232 , there exists a right inverse @xmath221 of the operator @xmath1889 ( as in ( [ eq : d a * ] ) ) , satisfying @xmath2233 we postpone the proof of proposition [ prop : right ] to the appendix ( page ) .
[ proof : thm : l w * r ] let @xmath1977 and @xmath366 be as in the hypothesis .
we construct a map @xmath2234 such that @xmath2235 is well - defined and equals @xmath2236 , and we show that @xmath221 restricts to a bounded map from @xmath1873 to @xmath2237 .
( see claim [ claim : ze x w ] below . )
it follows from hypothesis ( h ) that there exists @xmath993 such that @xmath1899 ( defined as in ( [ eq : m * ] ) ) .
it follows from lemma [ le : si ] in appendix [ sec : homology chern ] that there exists a number @xmath2238 such that @xmath2239 , for every @xmath2240 .
we choose constants @xmath2241 and @xmath2104 as in the second assertion of proposition [ prop : right ] ( corresponding to @xmath559 and @xmath297 , for @xmath2242 ) .
furthermore , we choose constants @xmath2133 and @xmath2243 as in lemma [ le : infty 1 p a ] in appendix [ sec : proof : prop : right ] ( corresponding to @xmath297 and @xmath559 ) .
we define @xmath2244 . by assumption
we have @xmath2245 .
hence there exists an integer @xmath2246 such that @xmath2247 we choose a smooth function @xmath2248\to[0,1]$ ] such that @xmath2249 on @xmath2250\cup[3/4,1]$ ] , @xmath2251 on @xmath2252 $ ] , and @xmath2253 and @xmath2254 , for all @xmath1802 $ ] .
we choose a bijection @xmath2255 we define @xmath2256 $ ] by @xmath2257 and @xmath2258 $ ] by @xmath2259 .
furthermore , for @xmath2260 we define @xmath2261,\quad\rho_i(s , t):=\rho(s-{\varphi}(i))\rho(t-\psi(i)).\ ] ] we choose a compact subset @xmath2262 ^ 2 $ ] diffeomorphic to @xmath1891 , such that @xmath2263 ^ 2{\subseteq}{{\operatorname{int}}}k_0,\ ] ] and we denote @xmath2264 . furthermore , we choose a compact subset @xmath2265 ^ 2 $ ] diffeomorphic to @xmath1891 , such that @xmath2266 ^ 2{\subseteq}{{\operatorname{int}}}k$ ] . for @xmath2260
we define @xmath2267 for @xmath2268 we define @xmath2269 by the first assertion of proposition [ prop : right ] there exists a bounded right inverse @xmath758 of @xmath418 .
we fix @xmath2260 .
since @xmath2270 , we have @xmath2271 , and by inequality ( [ eq : f a n ] ) , the right hand side is bounded by @xmath559 .
hence it follows from the statement of proposition [ prop : right ] that there exists a right inverse @xmath1289 of @xmath383 , satisfying @xmath2272 we define @xmath2273 each section @xmath2274 induces a section @xmath2275 . for every @xmath2276 we have @xmath2277 , and
therefore the map @xmath2278 is invertible .
we define @xmath2279 by @xmath2280 furthermore , we define the map ( [ eq : r ] ) by @xmath2281 the operator @xmath2235 is well - defined and equals @xmath2236 .
the statement of theorem [ thm : l w * r ] is now a consequence of the following .
we denote by @xmath1873 the space of @xmath1874-sections of @xmath83 .
[ claim : ze x w ] the map @xmath221 restricts to a bounded operator from @xmath1873 to @xmath2237 ( defined as in ( [ eq : x w p lam ] ) ) .
we choose a constant @xmath2282 so big that @xmath2283 for a weakly differentiable section @xmath2284 we denote @xmath2285 [ claim : d a * wt al ] there exists a constant @xmath2286 such that @xmath2287 let @xmath2288 .
we denote @xmath2289 and @xmath2290 .
since @xmath2291 , a straight - forward calculation shows that @xmath2292 fix @xmath485
. then @xmath2293 .
hence equality ( [ eq : d a * wt al ] ) implies that @xmath2294 combining this with ( [ eq : r i 1 p],[eq : sup c 3 ] ) , we obtain @xmath2295 by ( [ eq : d a * wt al ] ) , we have @xmath2296 using again ( [ eq : r i 1 p ] ) , claim [ claim : d a * wt al ] follows .
we choose @xmath2286 as in claim [ claim : d a * wt al ] .
the following will be used in the proofs of claims [ claim : hat r wt r ] and [ claim : na a p lam ] below .
let @xmath2288 .
we abbreviate @xmath2297 . by the fact @xmath2298 , lemma [ le : infty 1 p a ] in appendix [ sec : proof : prop : right ] ( twisted morrey s inequality , using ( [ eq : f a n ] ) ) , the fact @xmath2270 and claim [ claim : d a * wt al ] , we have @xmath2299 recall that we have chosen @xmath993 such that @xmath1899 .
we define @xmath2300 recall that @xmath2301 denotes the orthogonal projection onto @xmath1866 .
claim [ claim : ze x w ] is now a consequence of the following three claims .
[ claim : hat r wt r ] we have @xmath2302 let @xmath2288 be such that @xmath2303 we denote @xmath2297 .
inequality ( [ eq : wt v infty ] ) , remark [ rmk : c ] , and the assumption @xmath2304 imply that @xmath2305 where @xmath2123 is defined as in ( [ eq : c inf l x xi ] ) .
we fix @xmath2268 and denote @xmath2290 . for @xmath1458 lemma [ le : infty 1 p a ] in appendix [ sec : proof : prop : right ] , ( [ eq : r i 1 p ] ) , the fact @xmath2270 , and the assumption @xmath2304 imply that @xmath2306 furthermore , we have @xmath2307 where @xmath2308 an argument involving a finite cover of @xmath2309 by small enough balls and lemma [ le : infty 1 p a ] , implies that @xmath2310 .
it follows that @xmath2311 we define @xmath2312 the definition ( [ eq : wt r xi z ] ) , the second inequality in ( [ eq : l x * l x c ] ) ( remark [ rmk : c ] in appendix [ sec : add ] ) , the statement of claim [ claim : d a * wt al ] , and the assumption @xmath2304 imply that @xmath2313 by the last equality in ( [ eq : l x * l x c ] ) , we have @xmath2314 hence using again the second inequality in ( [ eq : l x * l x c ] ) and the statement of claim [ claim : d a * wt al ] , we obtain @xmath2315 for every @xmath485 there are at most four indices @xmath2268 for which @xmath2316
. therefore , we have @xmath2317 using ( [ eq : r i 1 p],[eq : sup c 3 ] ) and the assumption @xmath2304 , it follows that @xmath2318 combining this with ( [ eq : wt r xi],[eq : hhat r xi infty],[eq : vert d mu u],[eq : vert pr wt r xi ] ) , claim [ claim : hat r wt r ] follows . [ claim : na a p lam ] we have @xmath2319 let @xmath2288 .
we define @xmath2320 and @xmath2321 as in ( [ eq : xi rho v v ] ) in appendix [ sec : add ] . by lemma [ le : na a l u ] in the same appendix , we have @xmath2322 where @xmath2323 denotes the vector field on @xmath4 generated by an element @xmath2324 . using the second part of lemma [ le : na a l u ] ( with @xmath2325 ) , it follows that @xmath2326 we choose a constant @xmath297 so big that @xmath2327 we define @xmath2328 .
since @xmath2329 , equalities ( [ eq : na a l u eta],[eq : d a wt xi ] ) and remark [ rmk : c ] imply that @xmath2330 since @xmath2331 , claim [ claim : d a * wt al ] and ( [ eq : wt v infty ] ) now imply claim [ claim : na a p lam ] .
[ claim : na a al p lam ] we have @xmath2332 let @xmath2288 be such that @xmath2304 .
we write @xmath2290
. then we have @xmath2333
setting @xmath2334 it follows that @xmath2335 here in the second inequality we used the fact @xmath2336 , and ( [ eq : r i 1 p ] ) .
this proves claim [ claim : na a al p lam ] , and completes the proof of claim [ claim : ze x w ] and hence of theorem [ thm : l w * r ] .
in the appendix some additional results are recollected , which were used in the proofs of the main theorems of this memoir . as always , we denote @xmath1936 , by @xmath749 the open ball of radius @xmath570 , by @xmath2337 the circle of radius @xmath570 , centered at @xmath1193 , and by @xmath67 the affine space of smooth connection one - forms on a smooth principal bundle @xmath66 .
let @xmath349 be as in chapter [ chap : main ] , and @xmath63 a riemann surface , equipped with a compatible area form @xmath64 .
for @xmath884 and @xmath74 we denote by @xmath2338 the ( infinitesimal ) action of @xmath2339 at @xmath196 .
let @xmath156 , @xmath66 be a ( principal ) @xmath2-bundle over @xmath65 of class @xmath166 , @xmath79 a connection one - form on @xmath66 of class @xmath167 , and @xmath168 a @xmath2-equivariant map of class @xmath167 .
we denote @xmath95 , define the energy density @xmath174 as in ( [ eq : e w ] ) , and denote by @xmath2340 the energy of @xmath98 . [
prop : e int u om ] let @xmath95 be a smooth vortex over @xmath1 with finite energy , such that the closure of the image @xmath2341 is compact .
then there exists a smooth section @xmath553 , such that @xmath2342 this result is a consequence of ( * ? ? ?
* proposition 11.1 ) . for the convenience of the reader
we include a proof here .
we need the following .
we denote by @xmath572 the standard angular one - form on @xmath573 . over any circle centered at the origin . ]
[ le : int s r si a]let @xmath66 be a smooth @xmath2-bundle over @xmath1868 and @xmath1104 .
then there exists a smooth section @xmath2343 , such that @xmath2344 where the norms are with respect to the standard metric on @xmath1868 .
we choose @xmath560 such that @xmath2345 ( this means that @xmath2346 is in radial gauge on @xmath2347 .
such a section exists , since ( [ eq : si a x x ] ) corresponds to a family of ordinary differential equations , one for each direction @xmath2348 . )
we identify @xmath2349 with @xmath588 and define @xmath2350 where the subscript `` @xmath278 '' refers to the @xmath278-component of the one - form @xmath2351 .
it follows from ( [ eq : si a x x ] ) that the @xmath277-component of this form vanishes , and therefore , @xmath2352 using the estimate @xmath2353 it follows that @xmath2354{\times}s^1}\big|({\sigma}\circ{\varphi})^*f_a\big|_{\sigma}ds'dt+\int_{\{0\}{\times}s^1}\big|({\sigma}\circ{\varphi})^*a\big|_{\sigma}dt,\ ] ] for every @xmath2355 , where the subscript `` @xmath65 '' indicates that the norms are taken with respect to the standard metric on @xmath65 . inequality (
[ eq : int s 1 r | ] ) follows from this by a straight - forward calculation .
this proves lemma [ le : int s r si a ] .
let @xmath95 be as in the hypothesis and @xmath567 .
we choose a section @xmath553 as in lemma [ le : int s r si a ] . denoting by @xmath2356 the energy of @xmath98 over @xmath2183
, we have , by ( * ? ? ?
* proposition 3.1 ) , @xmath2357 let @xmath296 . by (
* corollary 1.4 ) there exists a constant @xmath2104 such that @xmath2358 combining this with the estimate @xmath2359 , and using ( [ eq : int s 1 r | ] ) , it follows that @xmath2360 combining this with the inequality @xmath579 and ( [ eq : sqrt e w z ] ) , it follows that @xmath2361 by choosing @xmath2362 and using ( [ eq : e w b r int ] ) , the convergence ( [ eq : int b r e ] ) follows .
this proves proposition [ prop : e int u om ] .
[ le : a priori ] let @xmath754 be a compact subset .
then there exists a constant @xmath1184 such that the following holds
. let @xmath136 , @xmath66 be a smooth @xmath2-bundle over @xmath2363 , @xmath156 , and @xmath257 a vortex on @xmath66 of class @xmath167 , such that @xmath2364 ( where @xmath2365 denotes the energy of @xmath98 over the ball @xmath2363 ) .
then we have @xmath2366 since @xmath2 is compact , we may assume w.l.o.g . that @xmath795 is @xmath2-invariant .
the result then follows from theorem [ thm : reg gauge bdd ] below , the calculation in step 1 of the proof of ( * ? ? ?
* proposition 11.1 . ) , and lemma [ le : heinz ] .
[ cor : quant ] if @xmath4 is equivariantly convex at @xmath41 , on p. ] then we have @xmath2371 where @xmath366 ranges over all vortices over @xmath1 with @xmath66 smooth and @xmath257 of class @xmath167 for some @xmath156 , such that @xmath2372 and @xmath2373 is compact .
the next lemma was used in the proofs of proposition [ prop : cpt bdd ] and lemma [ le : conv e ] ( section [ sec : comp ] ) .
it is a consequence of ( * ? ? ?
* lemma 9.1 ) .
let @xmath932 be a surface with an area form and a compatible complex structure .
[ le : mu u ] let @xmath2374 , @xmath2375 and @xmath754 be compact subsets , and @xmath156 . then there exist positive constants @xmath758 and @xmath2376 such that the following holds .
let @xmath2377 , @xmath66 be a smooth @xmath2-bundle over @xmath65 , and @xmath257 an @xmath221-vortex on @xmath66 of class @xmath167 , such that @xmath2378 then @xmath2379 where we view @xmath2380 as a function from @xmath65 to @xmath2381 . [ thm : reg gauge bdd ] assume that @xmath65 is compact .
let @xmath66 be a smooth @xmath2-bundle over @xmath65 , @xmath156 , and @xmath257 a vortex on @xmath66 of class @xmath332 .
then there exists a gauge transformation @xmath2382 such that @xmath2383 is smooth over @xmath2384 .
[ prop : reg gauge ] let @xmath2385 be a number , @xmath66 a smooth @xmath2-bundle over @xmath1 , @xmath156 , and @xmath888 an @xmath221-vortex on @xmath66 of class @xmath167 .
then there exists a gauge transformation @xmath905 on @xmath66 of class @xmath166 such that @xmath2383 is smooth . by theorem [ thm : reg gauge bdd ]
there exists a gauge transformation @xmath2389 such that @xmath2390 is smooth .
let @xmath1283 be an integer and assume by induction that there exist gauge transformations @xmath2391 , for @xmath2392 , such that ( [ eq : g j w ] ) holds for @xmath2392 , and ( [ eq : g j g j-1 ] ) holds for @xmath2393 .
we show that there exists a gauge transformation @xmath2394 such that @xmath2395 we choose a smooth function @xmath2396 such that @xmath2397 for @xmath2398 . by theorem [ thm : reg gauge bdd ]
there exists a gauge transformation @xmath2399 such that @xmath2400 is smooth over @xmath2401 .
we define @xmath2402 then @xmath2403 is of class @xmath867 over @xmath2404 , and ( [ eq : g ell+1 g ell ] ) is satisfied .
furthermore , @xmath2405 is smooth over @xmath2406 .
therefore , using smoothness of @xmath2400 over @xmath2404 , lemma [ le : g smooth]([le : g k+1 p ] ) below implies that @xmath2407 is smooth over @xmath2406 .
it follows that @xmath2408 is smooth over @xmath2404 .
this proves ( [ eq : g ell+1 ] ) , terminates the induction , and concludes the proof of the claim .
we choose a collection @xmath2409 as in the claim , and define @xmath905 to be the unique gauge transformation on @xmath66 that restricts to @xmath1664 over @xmath2410 .
this makes sense by ( [ eq : g j g j-1 ] ) . furthermore ,
( [ eq : g j w ] ) implies that @xmath2383 is smooth .
this proves proposition [ prop : reg gauge ] .
[ thm : cpt cpt ] let @xmath65 be a compact surface ( possibly with boundary ) , @xmath64 an area form , @xmath1381 a compatible complex structure on @xmath65 , @xmath66 a @xmath2-bundle over @xmath65 , @xmath754 a compact subset , @xmath933 , @xmath156 , and @xmath2411 an @xmath722-vortex on @xmath66 of class @xmath332 , for every @xmath132 .
assume that @xmath722 converges to some @xmath2412 , and @xmath2413 then there exist a smooth @xmath758-vortex @xmath2414 on @xmath2415 and gauge transformations @xmath912 on @xmath66 of class @xmath867 , such that @xmath763 converges to @xmath2414 , in @xmath471 on every compact subset of @xmath2384 .
this follows from a modified version of the proof of ( * ? ? ?
* theorem 3.2 ) : we use a version of uhlenbeck compactness for a compact base with boundary , see theorem [ thm : uhlenbeck compact ] below , and a version of the local slice theorem allowing for boundary , see ( * ? ? ?
* theorem 8.1 ) .
note that the proof carries over to the case in which @xmath2416 for some @xmath132 , or @xmath2417 .
[ prop : bounded ] assume that @xmath4 is equivariantly convex at @xmath41 . then there exists a @xmath2-invariant compact subset @xmath1421 such that the following holds . let @xmath156 , @xmath66 a @xmath2-bundle over @xmath1 , and @xmath257 a vortex on @xmath66 of class @xmath167 , such that @xmath578 and @xmath1878 is compact .
then we have @xmath2418 .
let @xmath66 be a @xmath2-bundle over @xmath1 . by an elementary argument
every smooth vortex on @xmath66 is smoothly gauge equivalent to a smooth vortex that is in radial gauge outside @xmath550 . using proposition [ prop : reg gauge ]
, it follows that every vortex on @xmath66 of class @xmath167 is gauge equivalent to a smooth vortex that is in radial gauge outside @xmath550 .
hence the statement of proposition [ prop : bounded ] follows from ( * ? ? ?
* proposition 11.1 ) .
the proof of proposition [ prop : en conc ] ( energy concentration near ends ) in section [ sec : soft ] was based on an isoperimetric inequality and an energy action identity for the invariant action functional ( theorem [ thm : isoperi ] and proposition [ prop : en act ] below ) .
building on work by d. a. salamon and r. gaio @xcite , we define this functional as follows .
we first review the usual symplectic action functional : let @xmath0 be a symplectic manifold without boundary .
we fix a riemannian metric @xmath1106 on @xmath4 , and denote by @xmath2419 , and @xmath2420 the distance function , the exponential map , the norm of a vector @xmath2421 , and the injectivity radii of a point @xmath74 and a subset @xmath1111 , respectively .
we define the symplectic action of a loop @xmath2422 of length @xmath2423 to be @xmath2424 here @xmath1810 denotes the ( closed ) unit disk , and @xmath2425 is any smooth map such that @xmath2426 let now @xmath2 be a compact connected lie group with lie algebra @xmath3 .
suppose that @xmath2 acts on @xmath4 in a hamiltonian way , with ( equivariant ) momentum map @xmath5 , and that @xmath1106 is @xmath2-invariant .
we denote by @xmath2429 the natural contraction .
let @xmath66 be a smooth @xmath2-bundle over @xmath588 and @xmath2430 .
we call @xmath2431 _ admissible _ iff there exists a section @xmath2432 such that @xmath2433 , and @xmath2434 for every @xmath2435 satisfying @xmath2436 . [
defi : action ] let @xmath2431 be an admissible pair , and @xmath79 be a connection on @xmath66 .
we define the _ invariant symplectic action of @xmath2437 _ to be @xmath2438 where @xmath2432 is a section as above .
to formulate the isoperimetric inequality , we need the following .
if @xmath586 is a manifold , @xmath66 a @xmath2-bundle over @xmath586 and @xmath362 , then we define @xmath2439 by @xmath2440 , where @xmath229 is any point in the fiber over @xmath2441 .
we define @xmath1894 as in ( [ eq : m * ] ) . for a loop @xmath2442
we denote by @xmath2443 its length w.r.t .
the riemannian metric on @xmath2444 induced by @xmath1106 .
furthermore , for each subset @xmath1111 we define @xmath2445 the first ingredient of the proof of proposition [ prop : en conc ] is the following .
[ thm : isoperi ] assume that there exists a @xmath2-invariant @xmath61-compatible almost complex structure @xmath42 such that @xmath2446 .
then for every compact subset @xmath2447 and every constant @xmath2448 there exists a constant @xmath993 with the following property .
let @xmath66 be a @xmath2-bundle over @xmath588 and @xmath2430 , such that @xmath2449 and @xmath2450 .
then @xmath2431 is admissible , and for every connection @xmath79 on @xmath66 we have @xmath2451 here we view @xmath2452 as a one - form on @xmath588 with values in the bundle @xmath2453 , and @xmath2454 as a section of the co - adjoint bundle @xmath2455 .
furthermore , @xmath588 is identified with @xmath2349 , and the norms are taken with respect to the standard metric on @xmath2349 , the metric @xmath1106 on @xmath4 , and the operator norm on @xmath81 induced by @xmath62 . was identified with @xmath2456 instead .
note also that hypothesis ( h ) is not needed for theorem [ thm : isoperi ] . ]
the second ingredient of the proof of proposition [ prop : en conc ] is the following . for @xmath2457
we denote by @xmath2458 the map given by @xmath2459 .
let @xmath2460 be manifolds , @xmath2461 , @xmath66 a @xmath2-bundle over @xmath586 , @xmath79 a connection on @xmath66 , and @xmath362 .
then the pullback triple @xmath2462 consists of a @xmath2-bundle @xmath2463 over @xmath2464 , a connection on @xmath2463 , and an equivariant map from @xmath2463 to @xmath4 .
[ prop : en act ] for every compact subset @xmath2447 there exists a constant @xmath993 with the following property .
let @xmath2465 be numbers , @xmath2466{\times}s^1 $ ] , @xmath64 an area form on @xmath65 , @xmath1381 a compatible complex structure , and @xmath888 a smooth vortex over @xmath65 , such that @xmath2467 and @xmath2468 , for every @xmath2469 $ ] .
then the pairs @xmath2470 are admissible , and @xmath2471 this section contains the proofs of lemmas [ le : p],[le : section ] and proposition [ prop : chern maslov ] , which were stated in section [ sec : homology chern ] . we also state and prove lemma [ le : homotopic ] , which was used in definition [ defi : homotopy w ] in that section .
let @xmath2472 and @xmath42 be as in chapter [ chap : main ] , @xmath85 , @xmath64 the standard area form @xmath272 , @xmath156 , and @xmath204 .
assume that hypothesis ( h ) holds .
lemma [ le : p ] was used in the definition of the equivariant homology class of an equivalence class of triples @xmath101 ( definition [ defi : homology ] ) .
its proof is based on the following result , which was also used in the proofs of theorem [ thm : fredholm ] ( section [ subsec : reform ] ) , [ thm : l w * r ] ( section [ subsec : proof : thm : l w * r ] ) and proposition [ prop : triv ] ( section [ subsec : proof : fredholm aug ] ) .
[ le : si ] for every @xmath1757 there exists a section @xmath560 of the restriction of the bundle @xmath66 to @xmath2473 , of class @xmath167 , and a point @xmath1929 , such that @xmath2474 converges to @xmath1979 , uniformly in @xmath2126 , as @xmath2127 , and @xmath2475 .
we define the function @xmath2477 .
it follows from the ad - invariance of @xmath62 that @xmath2478 since @xmath2071 is compact , we have @xmath2479 and @xmath2480 .
furthermore , @xmath2481 . combining this with ( [ eq : d mu u 2 ] ) , it follows that @xmath2482 .
therefore , by proposition [ prop : hardy ] in the next section ( hardy - type inequality , applied with @xmath318 replaced by @xmath162 and @xmath157 replaced by @xmath2143 ) the expression @xmath2199 converges to some number @xmath2483 , as @xmath2127 , uniformly in @xmath2126 . since @xmath2484
, it follows that @xmath2485 .
this proves claim [ claim : r u ] .
it follows from hypothesis ( h ) that there exists a @xmath993 such that @xmath1899 ( defined as in ( [ eq : m * ] ) ) .
we choose @xmath567 so big that @xmath2486 if @xmath2487 .
since @xmath2 is compact , the action of it on @xmath4 is proper .
hence the local slice theorem implies that @xmath2444 carries a unique manifold structure such that the canonical projection @xmath2488 is a submersion .
consider the map @xmath2489 defined by @xmath2490 , where @xmath223 is arbitrary .
we choose @xmath1939 and an embedding @xmath2493 .
furthermore , we choose a smooth function @xmath2494 that vanishes on @xmath2495 and equals 1 on @xmath2496 .
we define @xmath2497 to be the map given by @xmath2498 on @xmath2499 and by 0 on @xmath2495 .
it follows that @xmath2500 a short calculation shows that @xmath2501 , and therefore , @xmath2502 our assumption @xmath2503 implies that @xmath2504 .
furthermore , @xmath7 is proper by the hypothesis ( h ) , hence the set @xmath2505 is compact .
thus the same holds for the set @xmath2506 .
this set contains the image of @xmath949 .
it follows that @xmath2507 . combining this with ( [ eq : df],[eq : d iota bar u ] )
, we obtain the estimate @xmath2508 hence the hypotheses of proposition [ prop : hardy ] ( with @xmath157 replaced by @xmath2143 ) are satisfied .
it follows that the point @xmath2199 converges to some point @xmath2509 , uniformly in @xmath2126 , as @xmath2127 .
claim [ claim : bar u ] follows .
let @xmath2510 be as in claim [ claim : bar u ] .
we choose a local slice around @xmath2510 , i.e. , a pair @xmath2511 , where @xmath2512 is an open neighborhood of @xmath2510 , and @xmath2513 is a smooth map satisfying @xmath2514 .
then there exists a unique section @xmath2515 of @xmath2516 , of class @xmath167 , such that @xmath2517 . by the continuous homotopy lifting property of
@xmath66 we may extend this to a continuous section @xmath2518 of @xmath2519 .
regularizing @xmath2518 on @xmath2520 , we obtain a section @xmath560 of @xmath2519 , of class @xmath167 . suitably in local trivializations . ]
we define @xmath2521 .
it follows from claim [ claim : bar u ] that @xmath2125 converges to @xmath1979 , uniformly in @xmath44 , for @xmath2127 .
furthermore , @xmath2522 since @xmath2523 it follows that @xmath2475 .
this proves lemma [ le : si ] .
[ proof : le : p ] we prove * statement ( [ le : p : exists ] ) * : we denote by @xmath2524 the bundle projection , and identify @xmath608 .
we choose @xmath560 and @xmath1979 as in lemma [ le : si ] , and we define @xmath1760 to be the quotient of @xmath2009 under the equivalence relation generated by @xmath2525 , where @xmath288 is determined by @xmath2526 , for @xmath229 . we define @xmath2527 to be the canonical map from @xmath66 to @xmath1760 , and @xmath1763 to be the unique map satisfying @xmath2528):=g^{-1}x_\infty,\,\forall g\in g.\ ] ] the statement of lemma [ le : si ] implies that this map is continuous .
this proves ( [ le : p : exists ] ) .
we prove * statement ( [ le : p : iso])*. uniqueness of @xmath2529 follows from the condition @xmath2530 and continuity of @xmath2529 .
we prove existence : we define the map @xmath2531 as follows .
the map @xmath2532 descends to a continuous map @xmath2533 . recall that @xmath1894 denotes the set of points in @xmath4 where @xmath2 acts freely .
we denote @xmath2534 . since @xmath1894 is open , the set @xmath2535 is , as well .
since @xmath2 is compact , the canonical map @xmath2536 defines a smooth @xmath2-bundle .
we denote by @xmath2537 the projection map .
we define @xmath2538 furthermore , we define @xmath2539 in an analogous way , using @xmath2540 and @xmath2541 .
since @xmath2024 descends to the identity on @xmath1 , and @xmath1769 , the maps @xmath318 and @xmath2542 descend to the same map @xmath2543 .
since @xmath1764 , @xmath2544 , and @xmath2532 and @xmath2541 are continuous , it follows that @xmath2545 .
we claim that there exists a unique map @xmath1770 , satisfying @xmath2546 to see uniqueness of this map , note that the hypothesis @xmath2547 implies that @xmath2548 .
hence conditions ( [ eq : wt phi wt psi ] ) determine @xmath2529 on the whole of @xmath2540 .
existence of this map follows from the equality @xmath2549 on @xmath2550 , which follows from the assumptions @xmath2544 , @xmath1764 , and @xmath2551 .
the map @xmath2529 has the required properties .
this proves ( [ le : p : iso ] ) and completes the proof of lemma [ le : p ] .
we now prove lemma [ le : section ] ( p. ) , which was used in section [ sec : homology chern ] , in order to define the @xmath1793-homotopy class of an equivalence class @xmath50 of triples @xmath101 .
( see definition [ defi : homotopy w ] . )
[ proof : le : section ] to prove the first statement , we choose a section @xmath2552 of the restriction of @xmath66 to the disk @xmath2553 , of class @xmath166 . by lemma [ le : si ]
there exists a section @xmath2043 of the restriction of the bundle @xmath66 to @xmath2473 , of class @xmath167 , and a point @xmath1929 , such that @xmath2554 converges to @xmath1979 , uniformly in @xmath2126 , as @xmath2127 .
we define @xmath2555 to be the unique map satisfying @xmath2556 , on @xmath588 , and @xmath560 to be the continuous section of @xmath66 that agrees with @xmath2552 on @xmath550 , and satisfies @xmath2557 by regularizing @xmath560 , we may assume that it is of class @xmath167 .
this section satisfies the requirements of the first part of the lemma . to prove the second statement ,
let @xmath560 be a section of @xmath66 of class @xmath167 , such that the map @xmath1853 continuously extends to a map @xmath1854 .
it follows from claim [ claim : bar u ] in the proof of lemma [ le : si ] that there exists a point @xmath2558 , such that @xmath2559 , for every @xmath2560 .
the second statement follows from this .
this proves lemma [ le : section ] .
[ le : homotopic ] let @xmath156 , @xmath204 , @xmath1757 be a triple , and @xmath2561 sections of @xmath66 as in lemma [ le : section ] .
then the continuous extensions @xmath2562 of @xmath2563 are weakly @xmath1793-homotopic .
we define @xmath2567 there exists a continuous map @xmath2568{\times}{\mathbb c}\to g$ ] such that @xmath2569 to see this , observe that we may assume without loss of generality that @xmath2570 .
( here we use the assumption that @xmath2 is connected . )
we choose a continuous map @xmath2571{\times}{\mathbb c}\to{\mathbb c}$ ] such that @xmath2572 we define @xmath2573 .
this map satisfies ( [ eq : g 0 ] ) .
we now define @xmath2574 this map satisfies the conditions ( [ eq : h 0 ] ) .
this proves the claim .
it follows from hypothesis ( h ) that there exists a number @xmath567 such that @xmath2575 . by the claim ,
we may assume without loss of generality that @xmath2576 on @xmath2577 .
since @xmath2 is compact , the canonical projection @xmath2578 naturally defines a smooth @xmath2-bundle .
it follows that the map @xmath2579{\times}({\sigma}{\setminus}b_r)\ni(t , z)\mapsto\pi\circ{{\widetilde{u}}}(z)\in m^*/g\ ] ] has a continuous lift @xmath1820{\times}({\sigma}{\setminus}b_r)\to m^*$ ] that agrees with the map @xmath2580 on @xmath2581 , with the map @xmath2582 on @xmath2583 , and with the map @xmath2584 on @xmath1103{\times}s^1_r$ ] , where @xmath2585 denotes the circle of radius @xmath221 , around 0 .
the map @xmath2579{\times}{\sigma}\ni(t , z)\mapsto\left\{\begin{array}{ll } { { \widetilde{u}}}(z),&\textrm{if } z\in{{\overline{b_r}}},\\ h(t , z),&\textrm{otherwise , } \end{array}\right.\ ] ] is a weak @xmath1793-homotopy from @xmath2532 to @xmath2586 .
this proves lemma [ le : homotopic ] .
we now prove proposition [ prop : chern maslov ] .
we need the following .
let @xmath98 be a representative of @xmath50 .
let @xmath1759 be an extension as in lemma [ le : p ] ( p. ) .
then @xmath61 induces a fiberwise symplectic form @xmath2016 on the topological vector bundle @xmath2587 over @xmath19 .
we denote by @xmath2588 its first chern class .
[ rmk : u e g ] let @xmath2 be a topological group , @xmath586 and @xmath2464 topological spaces , @xmath1772 and @xmath2590 topological @xmath2-bundles , @xmath4 a topological @xmath2-space , @xmath133 a @xmath2-equivariant symplectic vector bundle over @xmath4 , and @xmath168 , @xmath2591 continuous @xmath2-equivariant maps .
we define @xmath2592 to be the map induced by @xmath318 and @xmath1780 . then the symplectic vector bundles @xmath2593 are isomorphic . here
we denote by @xmath2594 the symplectic vector bundle over @xmath586 obtained from the pullback bundle @xmath2595 as the quotient by the @xmath2-action , by @xmath2596 the natural symplectic vector bundle over @xmath2597 , and by @xmath2598 the induced symplectic vector bundle over @xmath2599 .
an isomorphism between the bundles in ( [ eq : u e g ] ) is given by the map @xmath2600 we choose a continuous @xmath2-equivariant map @xmath2601 .
we denote by @xmath2602 the map induced by @xmath2603 . by definition
, @xmath2604 is the first chern class of the vector bundle @xmath2605 equipped with the fiberwise symplectic form induced by @xmath61 .
furthermore , we have @xmath105=f_*[s^2]$ ] .
therefore , equality ( [ eq : c 1 g w ] ) follows from naturality of the first chern class under pullback by the map @xmath162 , and remark [ rmk : u e g ] , with @xmath1649 , @xmath1758 replaced by @xmath2606 , and @xmath2607 .
this proves lemma [ le : chern bundle ] .
[ rmk : c 1 m ] let @xmath65 be an oriented topological surface homeomorphic to the closed disk .
we denote by @xmath269 the surface obtained from @xmath65 by collapsing its boundary to a point is homeomorphic to @xmath19 . ] , and by @xmath2608 the canonical `` collapsing '' map .
let @xmath2609 be a symplectic vector bundle over @xmath269 , @xmath1837 a symplectic vector space of dimension the rank of @xmath133 , and @xmath2610 a symplectic trivialization .
then we have @xmath2611\big{\rangle}=m_{{\partial}{\sigma},{\omega}}\big({\partial}{\sigma}{\times}{\partial}{\sigma}\ni(z , z')\mapsto\psi_{z'}^{-1}\psi_z\in{{\operatorname{aut}}}{\omega}\big ) . \ ] ] here @xmath2612 $ ] denotes the fundamental class of @xmath269 and @xmath2613 the group of linear symplectic automorphisms of @xmath49 , and @xmath2614 is defined as in ( [ eq : m x om ] ) .
furthermore , we denote by @xmath2615 the canonical map , and we use the canonical identification @xmath2616 , for @xmath2560 . equality ( [ eq : c 1 m ] ) follows from an elementary argument ( e.g. an argument as in the proof of ( * ? ? ?
* theorem 2.69 ) . ) @xmath271 [ proof : prop : chern maslov ] we choose an extension @xmath1759 as in lemma [ le : p]([le : p : exists ] ) , such that @xmath2617 .
( it follows from the proof of lemma [ le : p]([le : p : exists ] ) that @xmath2532 may be chosen to satisfy this condition . ) by lemma [ le : chern bundle ] , equality ( [ eq : c 1 g w ] ) holds .
we denote by @xmath2537 the canonical projection , by @xmath65 the compact surface obtained from @xmath1 by `` gluing a circle at @xmath41 '' , and by @xmath2618 the map that is the identity on the interior @xmath2619 , and maps the boundary @xmath2620 to @xmath41 .
we choose a continuous map @xmath2621 , such that @xmath2622 .
we define @xmath2623 .
this map continuously extends the map @xmath2624 , where @xmath2625 .
hence , by definition , the @xmath1793-homotopy class of @xmath50 equals the @xmath1793-homotopy class of @xmath897 .
we choose a symplectic trivialization @xmath2626 .
for @xmath2627 we define @xmath1841 to be the unique element satisfying @xmath2628 it follows that @xmath2629 , and hence , using the definition of the maslov index , @xmath2630 the statement of proposition [ prop : chern maslov ] is now a consequence of the following claim .
we define the map @xmath2632 this is a continuous symplectic trivialization .
we denote by @xmath269 the surface obtained from @xmath65 by collapsing its boundary @xmath1796 to a point .
there is a canonical homeomorphism @xmath2633 , and the composition of @xmath162 with this map agrees with the collapsing map . therefore ,
applying remark [ rmk : c 1 m ] with @xmath2634 and @xmath2635 replaced by @xmath2636 , we have @xmath2637\big{\rangle}=m_{{\partial}{\sigma},{\omega}}\big({\partial}{\sigma}{\times}{\partial}{\sigma}\ni(z , z')\mapsto{{\widetilde{\psi}}}_{z'}^{-1}{{\widetilde{\psi}}}_z\in{{\operatorname{aut}}}{\omega}\big ) . \ ] ] equality ( [ eq : wt si g ] ) implies that @xmath2638 combining this with ( [ eq : c 1 m wt psi ] ) , the claim follows .
let @xmath1910 .
the following lemma was used in section [ subsec : proof : fredholm aug ] in order to define norms on the vector spaces @xmath2639 and @xmath2640 defined in ( [ eq : xxx p lam d xxx p lam ] ) . if @xmath1946 then let @xmath1947)$ ] be such that @xmath1948 for @xmath1949 and @xmath1950 for @xmath1951 . in the case
@xmath1952 we set @xmath1953 .
recall the definitions @xmath2641 the following proposition was used in the proofs of theorem [ thm : fredholm aug ] ( section [ subsec : proof : fredholm aug ] ) and proposition [ prop : x x w ] ( section [ subsec : proofs ] ) . for every normed vector space @xmath49 we denote by @xmath2645 the space of bounded continuous maps from @xmath1890 to @xmath49 .
we denote by @xmath2363 the ball of radius @xmath570 in @xmath1890 , and by @xmath2646 the complement of a subset @xmath2647 .
recall the definitions ( [ eq : l k p],[eq : w k p ] ) of the weighted sobolev spaces @xmath2648 and @xmath2649 .
[ proof of proposition [ prop : lam d morrey ] ] inequality ( [ eq : lam morrey ] ) follows from inequality ( 1.11 ) in theorem 1.2 in the paper @xcite by r. bartnik .
assume now that @xmath2652 .
then it follows from morrey s embedding theorem that there exists a canonical bounded inclusion @xmath2663 . in order to show that this inclusion is compact ,
let @xmath2664 be a sequence such that @xmath2665 by morrey s embedding theorem and the arzel - ascoli theorem on @xmath2666 ( for @xmath1661 ) and a diagonal subsequence argument , there exists a subsequence @xmath2667 of @xmath726 that converges to some map @xmath2668 , weakly in @xmath2669 , and strongly in @xmath2670 , for every @xmath1661 .
we choose a constant @xmath2136 as in the first part of ( [ prop : lam morrey ] ) . for every @xmath567 we have @xmath2672 hence @xmath2673 .
since @xmath2652 , by inequality ( [ eq : lam morrey ] ) , this implies @xmath2671 . to see the second statement , we choose a smooth function @xmath2674 $ ] such that @xmath2675 for @xmath2676 , @xmath2677 for @xmath2678 , and @xmath2679 .
let @xmath2101 and @xmath1661 .
we define @xmath2680 $ ] .
abbreviating @xmath2681 , we have @xmath2682 inequality ( [ eq : lam morrey ] ) and the fact @xmath2683 on @xmath2183 imply that @xmath2684 furthermore , using ( [ eq : c sup nu ] ) , we have @xmath2685 combining this with ( [ eq : vert v j ] ) and ( [ eq : vert v j rho r ] ) , and the fact @xmath2686 , it follows that @xmath2687 since @xmath2652 and @xmath2101 is arbitrary , it follows that @xmath2667 converges to @xmath318 in @xmath2654 .
this proves the claim and completes the proof of statement * ( [ prop : lam morrey])*. * proof of statement ( [ prop : lam w cpt ] ) : * let @xmath2657 be as in the hypothesis .
let @xmath2688 be a sequence such that @xmath2689 by the rellich - kondrashov compactness theorem on @xmath2666 ( for @xmath1661 ) and a diagonal subsequence argument there exists a subsequence @xmath2690 and a map @xmath2691 , such that @xmath2692 converges to @xmath897 , strongly in @xmath2693 , as @xmath1653 , for every compact subset @xmath2694 .
elementary arguments show that @xmath2695 and @xmath2692 converges to @xmath897 in @xmath2696 .
( for the latter we use the hypothesis that @xmath2658 , as @xmath2659 . )
this proves * ( [ prop : lam w cpt])*. by morrey s theorem there is a constant @xmath297 such that @xmath2710 for every @xmath136 , weakly differentiable function @xmath2711 , and @xmath2712 .
let @xmath2713 and @xmath2441 be as in the hypothesis of the claim .
let @xmath2714 be such that @xmath2715 .
for @xmath1311 we define @xmath2716 . furthermore , we set @xmath2717 and we choose points @xmath2718 such that @xmath2719 for @xmath2720 we have @xmath2721 .
hence it follows from the statement of morrey s theorem that @xmath2722 moreover , for @xmath2723 we have @xmath2724 , and hence analogously , @xmath2725 using the inequality @xmath2726 the claim follows .
let @xmath2697 be such that @xmath2698 .
it follows from the claim that there exists @xmath2727 such that @xmath2699 converges to @xmath2728 , as @xmath2127 , uniformly in @xmath2700 . to prove inequality ( [ eq : hardy y ] ) , we choose a smooth map @xmath2729 $ ] such that @xmath2730 for @xmath2731 , @xmath2732 for @xmath2733 and @xmath2734 .
we fix a number @xmath567 and define @xmath2735,\quad\rho_r(x):=\rho(|x|/r).\ ] ] we abbreviate @xmath2736 . using lemma [ le : hardy ] with @xmath318 replaced by @xmath2737 , we have @xmath2738 combining this with a calculation using leibnitz rule , it follows that @xmath2739 the above claim implies that @xmath2740 using the equalities @xmath2741 and @xmath2742 , it follows that @xmath2743 inequality ( [ eq : hardy y ] ) follows by inserting this into the right hand side of ( [ eq : vert v lam ] ) and sending @xmath221 to @xmath41 .
this proves proposition [ prop : hardy ] .
the next result will be used to prove corollary [ cor : d l l ] below , which was used in the proof of theorem [ thm : fredholm aug ] ( section [ subsec : proof : fredholm aug ] ) . for every @xmath1811
we define @xmath2744 and @xmath2745 to be the spaces of polynomials in @xmath485 and @xmath2746 of degree _ less than _ @xmath591 .
we have @xmath2747 . ]
we abbreviate @xmath2748 let @xmath586 be a normed vector space and @xmath1577 a closed subspace .
we denote by @xmath2749 the dual space of @xmath586 and equip @xmath2750 with the quotient norm .
[ thm : t fredholm ] let @xmath2754 be such that @xmath2755 , @xmath2756 , @xmath210 , and @xmath2757 a constant coefficient homogeneous elliptic linear operator of order @xmath1595 on @xmath1890 . if @xmath2758 then @xmath380 is fredholm .
[ rmk : x y ] let @xmath586 be a normed vector space and @xmath1577 a closed subspace . we equip @xmath2750 with the quotient norm .
the map @xmath2759 is well - defined and an isometric isomorphism .
this follows from a straight - forward argument .
we denote by @xmath245 the space of schwartz functions on @xmath1 and by @xmath2760 the space of temperate distributions . by @xmath2761
we denote the fourier transform , and by @xmath2762 the inverse transform .
we prove * statement ( [ prop : d l l : kernel ] ) . *
a calculation in polar coordinates shows that for every polynomial @xmath318 in @xmath364 we have @xmath2765 hence our assumption @xmath2766 implies that @xmath2767 .
therefore , statement ( [ prop : d l l : kernel ] ) is a consequence of the following claim .
let @xmath2769 .
then @xmath2770 ( as temperate distributions ) .
it follows that the support of @xmath2771 is either empty or consists of the point @xmath2772 .
hence the paley - wiener theorem implies that @xmath318 is real analytic in the variables @xmath277 and @xmath278 , where @xmath2773 , and there exists @xmath2774 such that @xmath2775 .
therefore , by liouville s theorem @xmath318 is a polynomial in the variable @xmath364 .
since by our assumption @xmath1880 , it follows from ( [ eq :
u l ] ) that @xmath2776 .
this proves claim [ claim : ker t p d ] . to prove * statement ( [ prop : d l l : coker ] ) * , we define @xmath2777 .
consider the isometric isomorphism @xmath2778 denoting by @xmath2779 the adjoint operator of @xmath380 , we have @xmath2780 where the derivatives are taken in the sense of distributions . for every polynomial @xmath318 in @xmath2746 we have @xmath2782 our assumption @xmath1880 and ( [ eq : u
l * ] ) imply that @xmath2783 .
furthermore , the inclusion @xmath2784 is proved analogously to the inclusion @xmath2768 , using @xmath2766 and ( [ eq : u l * ] ) .
this proves claim [ claim : t p ] .
it follows from claim [ claim : t p ] that the map @xmath2024 restricts to a @xmath1-linear isomorphism between @xmath2785 and @xmath2786 .
the composition of this map with the canonical isomorphism @xmath2787 described in remark [ rmk : x y ] , equals the map ( [ eq : bar p -d ] ) .
statement ( [ prop : d l l : coker ] ) follows .
this completes the proof of proposition [ prop : d l l ] .
let @xmath1811 , @xmath2642 , @xmath1883 , and @xmath2788 $ ] be a smooth function that vanishes on @xmath2789 and equals 1 on @xmath2790 .
we equip @xmath2791 with the norm induced by the isomorphism of lemma [ le : x d iso ] .
this norm is complete .
( see e.g. @xcite . )
the composition of the isomorphism of lemma [ le : x d iso ] with the above map is given by @xmath2794 the map @xmath380 is the composition of the canonical projection @xmath2795 with the operator @xmath2796 . using proposition [ prop : d l l ]
, it follows that @xmath380 is fredholm of real index @xmath2793 .
furthermore , @xmath1611 is compact , since it equals the composition of the canonical projection @xmath2797 ( which is compact ) with a bounded operator .
corollary [ cor : d l l ] follows .
the next result was used in the proof of theorem [ thm : fredholm aug ] ( fredholm property for the augmented vertical differential ) in section [ subsec : proof : fredholm aug ] .
let @xmath2798 be a finite dimensional hermitian vector space , @xmath2799 positive linear maps , @xmath210 and @xmath2642 .
we define @xmath2800 consider first the case @xmath2804 and @xmath2805 .
we define @xmath2806 the map @xmath2807 is well - defined . by caldern s theorem
this map extends uniquely to an isomorphism @xmath2808 ( see ( * ? ? ?
* theorem 1.2.3 . ) . )
note that @xmath2809 for every @xmath2810 .
it follows that the inverse of ( [ eq : l g u ] ) is given by @xmath2811 hence this is an isomorphism .
we define @xmath2820 by a short calculation we have @xmath2821 since the linear maps @xmath2822 are positive , proposition [ prop : calderon ] implies that @xmath2823 is an isomorphism .
we denote by @xmath2824 its inverse and define @xmath2825 then @xmath221 is bounded and @xmath2826 . by a short calculation
, we have @xmath2827 , for every schwartz function @xmath2828 .
this implies that @xmath2829 and therefore @xmath2830 .
since @xmath2831 is continuous and @xmath2832 is dense , it follows that @xmath2833 .
the claim follows .
consider now the * general case @xmath210*. the map @xmath2835 is an isometric isomorphism .
furthermore , by proposition [ prop : lam d morrey]([prop : lam w ] ) the map @xmath2836 is well - defined and an isomorphism .
we define @xmath2837 direct calculations show that @xmath2838 therefore , proposition [ prop : lam d morrey]([prop : lam w cpt ] ) implies that the operator @xmath1611 is compact .
since we proved that @xmath418 is an isomorphism , it follows that @xmath2801 is a fredholm map of index 0 .
this proves proposition [ prop : d a b d ] in the general case .
the main result of this section states that a principal bundle of sobolev class @xmath166 is sobolev isomorphic to a smooth bundle , if @xmath170 is large enough .
this will be used in the proofs of propositions [ prop : right ] , [ prop : uhlenbeck ] , and [ prop : right d a * d a ] in the next section .
let @xmath1939 be an integer , @xmath2839 a real number , @xmath586 a smooth manifold ( possibly with boundary ) of dimension @xmath54 , @xmath2 a compact lie group , and @xmath66 a @xmath2-bundle over @xmath586 of class @xmath166 .
be open subsets , and @xmath2840 and @xmath2841 local trivializations .
then the corresponding transition function @xmath2842 is bounded in @xmath867 on every compact subset @xmath2843 .
note here that @xmath795 may intersect the boundary @xmath2844 . ]
the proof of this result relies on the facts that there exists a smooth @xmath2-bundle that is `` @xmath725-close '' to @xmath66 , and that if two `` sobolev bundles '' are `` @xmath725-close '' then they are `` sobolev - isomorphic '' . in order to explain this , let @xmath1582 be an open cover of @xmath586 .
we call a collection of functions @xmath2846 _ compatible _ iff it satisfies @xmath2847 recall that a cover @xmath1582 of @xmath586 is called _ locally finite _ iff every point @xmath927 possesses a neighborhood which intersects only finitely many sets in @xmath1582 . by a _ refinement _ of the cover @xmath1582 we mean a cover @xmath2849 of @xmath586 , together with a map @xmath2850 , such that @xmath2851 , for every @xmath2852
. the first ingredient of the proof of theorem [ thm : reg bundle ] is the following . [ prop : smooth collection ] let @xmath2853 be a compatible collection , and @xmath50 a neighborhood of the diagonal in @xmath2854 . then there exists a locally finite refinement @xmath2850 consisting of open precompact sets , and a compatible collection of smooth maps @xmath2855 such that @xmath2856 this result says that there exists a smooth ech @xmath1243-cocycle , which is arbitrarily close in the uniform sense to a given ech @xmath1243-cocycle of sobolev class . condition ( [ eq : wt g g ] ) may look unconventional , but it is a natural way of stating the closeness condition . an alternative would be to introduce a riemannian metric on @xmath2 and formulate the condition in terms of the induced distance function on @xmath2 .
@xmath271 we may assume w.l.o.g .
that @xmath2857 , by considering the double @xmath2858 . if @xmath586 is compact then the statement follows from ( * ? ? ?
* theorem 2.1 ) , using the fact that @xmath867-maps are continuous , since @xmath2839 . in the general case
, it follows from an adaption of that proof : we first choose a locally finite refinement @xmath2850 by precompact sets , and define @xmath2859 .
then we follow the argument of the proof of ( * ? ? ?
* theorem 2.1 ) , for the cover @xmath2849 . as in that proof to be precompact , since the sets in @xmath2849 are precompact .
] this proves proposition [ prop : smooth collection ] .
[ le : k v ] there exists a neighborhood @xmath2860 of the diagonal @xmath2854 with the following property .
let @xmath1939 be an integer , @xmath2845 a real number , @xmath586 a smooth manifold of dimension @xmath54 , and @xmath1582 a locally finite cover of @xmath586 by precompact sets .
then there exists a refinement @xmath2850 such that the following holds .
let @xmath2861 be compatible collections of maps satisfying @xmath2862 then there exists a collection of maps @xmath2863 ( @xmath2852 ) , such that @xmath2864 [ rmk : p ] let @xmath1582 be an open cover of @xmath586 , and @xmath2865 ( @xmath2866 ) be a compatible collection of maps .
we define the set @xmath2867 where the equivalence relation @xmath232 is defined by @xmath2868 if the maps @xmath2865 are smooth , then this set naturally is a smooth @xmath2-bundle , and if the maps @xmath2865 are of sobolev class @xmath2219 with @xmath2869 , then it naturally is a @xmath2-bundle of class @xmath2219 . in either case ,
a system of local trivializations is given by @xmath2870 , \ ] ] where @xmath2871 $ ] denotes the equivalence class of @xmath2872 .
the map @xmath2865 is the transition map from @xmath2873 to @xmath2874 .
this means that @xmath2875 let @xmath2876 , and @xmath2848 and @xmath2877 be compatible collections of maps of class @xmath2219 .
then the bundles @xmath2878 and @xmath2879 are @xmath2219-isomorphic , if there exists a collection of maps @xmath2880 , satisfying the equation @xmath2881 defining @xmath2873 as in ( [ eq : phi u u g ] ) and @xmath2882 similarly , with @xmath2865 replaced by @xmath2883 , an isomorphism @xmath2884 is given by @xmath2885\mapsto(\psi_u)_x\big(k_u(x)(\phi_u)_x^{-1}(g)\big).\ ] ] ( it follows from the compatibility condition ( [ eq : g u u u ] ) and ( [ eq : k u ] ) that this map is well - defined , i.e. , the right hand side above does not depend on the choice of the representative @xmath2872 . )
@xmath271 we choose a cover @xmath1582 of @xmath586 , and a system of local trivializations @xmath2886 of class @xmath166 ( @xmath1584 ) . for @xmath2866
we denote by @xmath2842 the corresponding transition map , defined by @xmath2887 where for @xmath288 , @xmath2888 denotes right multiplication .
we choose a neighborhood @xmath2860 of the diagonal in @xmath2854 as in lemma [ le : k v ] , and a refinement @xmath2849 and a collection of smooth maps @xmath2889 ( @xmath2890 ) as in proposition [ prop : smooth collection ] . using ( [ eq : wt g g ] )
, we may apply lemma [ le : k v ] with @xmath2891 replaced by @xmath2892 , to conclude that there exists a refinement @xmath2893 , and a collection of maps @xmath2894 ( @xmath2895 ) , such that @xmath2896 therefore , the statement of theorem [ thm : reg bundle ] follows from remark [ rmk : p ] . in this section
we prove proposition [ prop : right ] ( section [ subsec : proof : thm : l w * r ] ) .
we need the following results .
let @xmath1939 , @xmath2 be a compact lie group with lie algebra @xmath3 , @xmath62 an invariant inner product on @xmath3 , and @xmath2229 a riemannian manifold ( possibly with boundary ) of dimension @xmath54 . recall the definitions ( [ eq : vert al vert],[eq : om i k p a],[eq : ga k p a ] ) of @xmath2897 .
if @xmath586 is compact , @xmath2839 , and @xmath66 is a @xmath2-bundle over @xmath586 of sobolev class @xmath867 , then we denote by @xmath2898 the affine space of connection one - forms on @xmath66 of class @xmath332 .
[ prop : uhlenbeck ] let @xmath2899 .
assume that @xmath586 is compact and diffeomorphic to the closed ball @xmath2900 .
then there exist constants @xmath296 and @xmath297 with the following property .
let @xmath274 and @xmath66 be @xmath2-bundles over @xmath586 of class @xmath867 , and @xmath2901 connections , such that @xmath724 is flat and @xmath2902 then there exists an isomorphism of @xmath2-bundles @xmath2903 , of class @xmath867 , such that @xmath2904 [ le : uhlenbeck trivial ] let @xmath2899 .
assume that @xmath2905 , equipped with the standard metric @xmath2906 .
we denote by @xmath274 the trivial @xmath2-bundle @xmath2907 , and by @xmath724 the trivial connection on this bundle .
then there exist constants @xmath296 and @xmath345 , such that for every connection @xmath2908 , satisfying @xmath2232 , there exists a gauge transformation @xmath905 on @xmath274 of class @xmath867 , such that @xmath2909 [ rmk : metric ] let @xmath2910 and @xmath2911 be a riemannian metric on @xmath586 .
assume that @xmath586 is compact .
then there exists a constant @xmath345 such that @xmath2912 for every @xmath2-bundle @xmath66 over @xmath586 , of class @xmath2913 , connection @xmath2914 , integer @xmath2217 , and every differential form @xmath420 on @xmath586 with values in @xmath83 , of class @xmath825 .
( here the degree of @xmath420 is arbitrary . )
this follows from an elementary argument , using induction over @xmath322 .
@xmath271 consider first the * case @xmath2915 * together with the standard metric .
we choose constants @xmath2916 as in lemma [ le : uhlenbeck trivial ] .
let @xmath2917 be as in the hypothesis .
we show that the required isomorphism @xmath2024 exists .
it follows from theorem [ thm : reg bundle ] that we may assume without loss of generality that @xmath274 and @xmath66 are smooth . since @xmath586 is smoothly retractible to a point , @xmath274 and @xmath66 are smoothly isomorphic to the trivial bundle over @xmath586
. hence we may assume without loss of generality that they are the trivial bundle .
since @xmath724 is flat , it follows from lemma [ le : uhlenbeck trivial ] that @xmath724 is @xmath867-gauge equivalent to the trivial connection on @xmath2907 .
therefore , the statement of proposition [ prop : uhlenbeck ] is a consequence of lemma [ le : uhlenbeck trivial ] .
the proof of proposition [ prop : right ] is based on the following .
recall from ( [ eq : d a * ] ) that @xmath2918 denotes the formal adjoint of the operator @xmath194 .
if @xmath2919 and @xmath2920 are normed vector spaces and @xmath2921 a bounded linear map then we denote by @xmath2922 the operator norm of @xmath380 .
[ prop : right d a * d a ] let @xmath2228 .
assume that @xmath586 is diffeomorphic to @xmath2230 , and @xmath2229 can be embedded ( as a riemannian manifold ) into @xmath1890 , together with the standard metric .
then there exist constants @xmath296 and @xmath345 , such that for every @xmath2-bundle @xmath1772 of class @xmath867 , and every connection @xmath2923 satisfying @xmath2232 , there exists a right inverse @xmath221 of the operator @xmath2924 with operator norm @xmath2925 bounded above by @xmath297 .
[ le : infty 1 p a ] let @xmath2228 .
assume that @xmath586 is diffeomorphic to @xmath2926 .
then there exist constants @xmath297 and @xmath296 such that for every @xmath2-bundle @xmath1772 of class @xmath867 , and @xmath2923 , the following holds .
if @xmath2232 then @xmath2927 we choose constants @xmath296 and @xmath2106 as in proposition [ prop : uhlenbeck ] .
let @xmath66 be a @xmath2-bundle over @xmath586 , of class @xmath867 , and @xmath2923 , such that @xmath2232 . by the statement of proposition
[ prop : uhlenbeck ] , there exists an isomorphism @xmath2903 of class @xmath867 , such that @xmath2932 let @xmath2933 and @xmath2934 .
we set @xmath2935|\,\big|\,\xi,\eta\in { \mathfrak g}:\,|\xi|\leq1,\,|\eta|\leq1\big\},\ ] ] where the norm @xmath2936 is with respect to @xmath62 .
a direct calculation shows that @xmath2937,\ ] ] where @xmath2938:{\mathfrak g}_p\otimes{\mathfrak g}_p\to{\mathfrak g}_p$ ] denotes the map induced by the lie bracket on @xmath3 . using the above claim
, it follows that @xmath2939 the above claim , inequality ( [ eq : phi a a 0 c 2 ] ) , and the assumption @xmath2232 imply the estimate @xmath2940 . combining this with ( [ eq : al infty ] ) , the statement of lemma [ le : infty 1 p a ] follows .
[ le : a 0 a ] let @xmath2228 and @xmath296 .
assume that @xmath586 is diffeomorphic to @xmath2926 .
then there exists a constant @xmath993 with the following property .
let @xmath1772 be a @xmath2-bundle of class @xmath867 , @xmath2901 , and @xmath2941 , such that @xmath724 is flat @xmath2942 then the following inequalities hold : @xmath2943 we have @xmath2944 . & \ ] ] it follows that there exists a constant @xmath345 such that @xmath2945 for every @xmath2-bundle @xmath1772 of class @xmath867 , @xmath2901 , and @xmath2941 . depends on the lie bracket on @xmath3 and the metric on @xmath586 , but not on the bundle @xmath66 .
] using lemma [ le : infty 1 p a ] and flatness of @xmath724 , it follows that there exists a constant @xmath993 such that inequality ( [ eq : d a * d a a 0 ] ) holds for every @xmath2946 as in the hypothesis .
[ le : right inverse ] let @xmath586 and @xmath1581 be banach spaces , and @xmath2949 and @xmath2950 bounded linear maps , such that @xmath2951 then there exists a right inverse @xmath221 of @xmath2952 , satisfying @xmath2953 .
[ rmk : x t ] let @xmath586 be a banach space , and @xmath2954 a linear map satisfying @xmath2955 .
then @xmath2956 is invertible , and @xmath2957 this follows from a standard argument , using the neumann series @xmath2958 .
@xmath271 [ claim : flat right ] let @xmath2642 .
there exists a constant @xmath345 such that for every @xmath2-bundle @xmath1772 of class @xmath867 , and every _ flat _
connection @xmath79 on @xmath66 of class @xmath332 , there exists a right inverse @xmath221 of the operator ( [ eq : d a * d a ] ) , satisfying @xmath2961 . by theorem
[ thm : reg bundle ] we may assume without loss of generality that @xmath66 is smooth . we choose a smooth flat connection @xmath724 on @xmath66 . by proposition [ prop : uhlenbeck ]
there exists an automorphism @xmath2024 of @xmath66 , of class @xmath867 , such that inequality ( [ eq : phi a a 0 ] ) holds .
since @xmath79 is flat , it follows that @xmath2962 , and thus @xmath2963 is smooth .
hence we may assume w.l.o.g .
that @xmath79 is smooth . for an open subset @xmath2964
we denote by @xmath2965 the compactly supported smooth real valued functions on @xmath1627 .
we define the operator @xmath2966 as follows .
we denote by @xmath2967 the fundamental solution for the laplace equation .
it is given by @xmath2968 where @xmath2969 denotes the gamma function .
( see e.g. @xcite , p. 22 . )
let @xmath2970 .
we define @xmath2971 to be the extension of @xmath162 by 0 outside @xmath550 .
we denote by @xmath192 convolution in @xmath1890 and define @xmath2972 note that @xmath2024 is locally integrable , hence the convolution is well - defined . furthermore , @xmath2973 is smooth , and @xmath2974 .
young s inequality states that @xmath2976 furthermore , the caldern - zygmund inequality states that there exists a constant @xmath297 such that for every @xmath2977 we have @xmath2978 .. ] the statement of claim [ claim : c tf ] follows from this .
we fix a constant @xmath297 as in claim [ claim : c tf ] . by this claim
the map @xmath2979 uniquely extends to a bounded linear map @xmath2980 since @xmath2974 , for every @xmath2970 , a density argument shows that @xmath2981 , for every @xmath2982 , i.e. , @xmath380 is a right inverse for @xmath2983 .
let now @xmath2 be a compact lie group , @xmath62 an invariant inner product on @xmath2984 , and @xmath2229 a riemannian manifold as in the hypothesis of proposition [ prop : right d a * d a ] .
without loss of generality we may assume that @xmath586 is a submanifold ( with boundary ) of @xmath1890 , and that @xmath2212 is the standard metric .
let @xmath2985 be a @xmath2-bundle , and @xmath1104 be a flat connection .
we fix a point @xmath2986 , and denote by @xmath2987 the @xmath79-horizontal section through @xmath2988 .
this is the unique smooth section of @xmath66 satisfying @xmath2989 and @xmath2990 .
( such a section exists , since @xmath79 is flat , and @xmath586 is diffeomorphic to @xmath2926 . ) for @xmath454 we define the map @xmath2991 this is an isometric isomorphism .
we define @xmath2992 .
it follows that @xmath2993 a straight - forward calculation shows that @xmath2994 it follows that @xmath221 is a right inverse for @xmath2995 .
this proves claim [ claim : flat right ] .
we choose constants @xmath2996 as in claim [ claim : flat right ] , @xmath1117 as in lemma [ le : a 0 a ] , corresponding to @xmath2997 , @xmath2998 as in proposition [ prop : uhlenbeck ] , and @xmath2999 as in lemma [ le : infty 1 p a ] .
we define @xmath3000 let @xmath1772 be a @xmath2-bundle of class @xmath867 , and @xmath2923 , such that @xmath2232 .
the statement of proposition [ prop : right d a * d a ] is a consequence of the following claim .
we choose a flat connection @xmath3003 on @xmath66 of class @xmath332 . that such a connection exists . ] since @xmath3004 , by the statement of proposition [ prop : uhlenbeck ] with @xmath3005 there exists an automorphism @xmath2024 of @xmath66 of class @xmath867 , such that @xmath3006 we define @xmath3007 . by the statement of claim [ claim : flat
right ] there exists a right inverse @xmath758 of the operator @xmath3008 satisfying @xmath3009 , where @xmath3010 denotes the operator - norm . ''
to indicate that the definition of this norm involves the connection @xmath724 . ]
the assumption @xmath3011 and ( [ eq : phi a wt a 0 ] ) imply that the condition ( [ eq : a a 0 de ] ) is satisfied .
therefore , by ( [ eq : d a * d a a 0 ] ) with `` @xmath559''@xmath3012 , we have @xmath3013 therefore , applying lemma [ le : right inverse ] , there exists a right inverse @xmath221 of the operator @xmath3014 satisfying @xmath3015 . combining this with inequality ( [ eq : xi 2 p a ] ) ( with `` @xmath559''@xmath3016 ) , it follows that @xmath3017 .
this proves claim [ claim : r d a * d a ] and completes the proof of proposition [ prop : right d a * d a ] .
consider first the * case @xmath3021 and @xmath1857 * , and let @xmath3022 .
defining @xmath3023 if @xmath3024 , and @xmath1193 , otherwise , fubini s theorem implies that @xmath3025{\times}[0,1]}\chi(t , x)|f(t)|\,dt\,dx=\int_0 ^ 1(1-t)|f(t)|dt<\infty.\ ] ] the statement of the lemma in the case @xmath3021 and @xmath1857 follows . for @xmath3021 and a general @xmath54
the proof is similar . for a general exponent @xmath170 ,
hlder s inequality implies that @xmath3026 hence in general , the statement of the lemma follows from what we have already proved , by considering the function @xmath3027 .
[ proof : prop : right ] we prove * statement ( [ prop : right : exists])*. + let @xmath3028 be as in the hypothesis .
we show that the operator @xmath1889 admits a bounded right inverse . by theorem [ thm : reg bundle ]
we may assume w.l.o.g .
that @xmath66 is smooth .
since by assumption , @xmath586 is diffeomorphic to @xmath2926 , we may assume without loss of generality that @xmath2915 .
the hodge @xmath192-operator induces an isomorphism between @xmath3029 and @xmath3030 . using the equality @xmath3031
, it follows that the operator @xmath3032 admits a bounded right inverse , if and only if the operator @xmath3033 does so .
since @xmath2915 is compact , using remark [ rmk : metric ] , it follows that the condition that @xmath1889 admits a bounded right inverse , is independent of the metric @xmath2212 .
hence we may assume without loss of generality that @xmath2212 is the standard metric on @xmath2230 .
we define @xmath3036 to be the subset consisting of all @xmath3037 such that @xmath3038 where @xmath3039 .
furthermore , we denote by @xmath3040 the @xmath79-parallel transport in @xmath3041-direction .
let @xmath3042 .
we denote @xmath3043 .
then @xmath3044 , where @xmath3045 is the unique path satisfying @xmath3046 let @xmath3047 .
we define @xmath3048 by the condition @xmath3049=\xi\circ \pi(p)$ ] , for @xmath229 , and @xmath3050 where @xmath3051 .
furthermore , we define the section @xmath3052 by the condition @xmath3053 $ ] , for every @xmath229 , and @xmath3054 this defines the operator ( [ eq : t ga ] ) .
claim [ claim : l w ] is a consequence of the following .
the section is of class @xmath3058 , since @xmath2339 is of this class , by morrey s embedding theorem .
we denote by @xmath3059 the standard basis of @xmath1890 .
we show that @xmath3060 let @xmath3061 .
we fix a point @xmath2988 in the fiber of @xmath66 over @xmath3062 , and define the path @xmath3063 . using the equality @xmath3064 , it follows from ( [ eq : wt eta ] ) that @xmath3065 it follows that @xmath3066 and therefore , @xmath3067 .
this proves ( [ eq : d a eta ] ) .
it follows that @xmath3068 is of class @xmath332 .
let now @xmath3069 .
we show that @xmath3070 .
we fix @xmath3061 and choose a smooth path @xmath3071 , such that @xmath3072 . using ( [ eq : wt eta ] ) with @xmath3073 and differentiating under the integral , we obtain @xmath3074 since @xmath3075 and @xmath3076 , it follows that @xmath3077 by lemma [ le : n p f ] with @xmath3078 , using the assumption @xmath3079 , it follows that @xmath3070 .
therefore , @xmath3055 lies in @xmath3080 . to prove equality ( [ eq : d a * eta ] ) , observe that @xmath3081 where in the second step we used ( [ eq : d a eta ] ) .
this proves claim [ claim : t ] and hence claim [ claim : l w ] .
we choose a map @xmath380 as in claim [ claim : l w ] , and a flat connection @xmath3082 . by proposition [ prop : right d
a * d a ] there exists a bounded right inverse @xmath758 of @xmath3083 statement ( [ prop : right : exists ] ) is now a consequence of the following .
a short calculation shows that @xmath3086 is of first or zeroth order .
hence it is bounded as a map from @xmath3087 to @xmath3056 .
furthermore , the equality @xmath3088 holds on smooth sections of @xmath83 .
this implies that @xmath221 is well - defined and bounded .
a short calculation shows that @xmath3085 .
this proves claim [ claim : r ] , and completes the proof of ( [ prop : right : exists ] ) . to prove * statement ( [ prop :
right : eps c ] ) * , we choose constants @xmath559 and @xmath297 as in proposition [ prop : right d a * d a ] .
let @xmath1772 be a @xmath2-bundle of class @xmath867 , and @xmath2923 , such that @xmath2232 . by the statement of proposition
[ prop : right d a * d a ] , there exists a right inverse @xmath221 of the operator @xmath3089 , satisfying @xmath2961 .
the operator @xmath3090 is a right inverse for @xmath1889 , satisfying @xmath3091 . here in the last inequality we used the fact @xmath3092 , where @xmath3093 this proves ( [ prop :
right : eps c ] ) , and completes the proof of proposition [ prop : right ] .
[ thm : uhlenbeck compact ] let @xmath1939 , @xmath2 be a compact lie group , @xmath586 a compact smooth riemannian @xmath54-manifold ( possibly with boundary ) , @xmath66 a smooth @xmath2-bundle over @xmath586 , @xmath2839 a number , and @xmath723 a sequence of connections on @xmath66 of class @xmath332 .
assume that @xmath3094 then passing to some subsequence there exist gauge transformations @xmath912 of class @xmath867 , such that @xmath767 converges weakly in @xmath332 .
[ prop : compactness delbar][prop : comp delbar ] let @xmath4 be a manifold without boundary , @xmath397 , @xmath156 numbers , @xmath42 an almost complex structure on @xmath4 of class @xmath518 , @xmath975 open subsets , and @xmath3095 a sequence of functions of class @xmath167 .
assume that @xmath3096 is of class @xmath2219 , for every @xmath508 , and that for every open subset @xmath3097 with compact closure the following holds .
if @xmath875 is so large that @xmath876 then @xmath3098 then there exists a subsequence of @xmath726 that converges weakly in @xmath3099 and in @xmath518 on every compact subset of @xmath3100 . [ le : g c k+1]let @xmath905 be a gauge transformation of class @xmath167 and @xmath79 a connection on @xmath66 of class @xmath518 , such that @xmath904 is of class @xmath518 . then @xmath905 is of class @xmath3102 .
[ le : g k+1 p ] assume that @xmath586 is compact ( possibly with boundary ) .
let @xmath1582 be a subset of the space of @xmath825-connections on @xmath66 that is bounded in @xmath825 .
then there exists a @xmath3099-bounded subset @xmath2849 of the set of @xmath3099-gauge transformations on @xmath66 , such that the following holds .
let @xmath3103 and @xmath905 be a @xmath332-gauge transformation , such that @xmath3104
. then @xmath3105 .
[ prop : ka 0 ] let @xmath1939 , @xmath2 be a compact lie group , @xmath66 a @xmath2-bundle over @xmath1890 , and @xmath3107 smooth flat connections on @xmath66 .
then there exists a smooth gauge transformation @xmath905 such that @xmath3108 .
, see also ( * ? ? ?
* corollary 3.7 ) . ] in the case @xmath1857 such a @xmath905 exists , since then the condition @xmath3108 can be viewed as an ordinary differential equation for @xmath905 .
let @xmath1939 and assume by induction that we have already proved the statement for @xmath54 .
let @xmath66 be a @xmath2-bundle over @xmath3109 , and @xmath3107 smooth flat connections on @xmath66 .
we define @xmath3110 by @xmath3111 . by the induction hypothesis
there exists a smooth gauge transformation @xmath3112 on @xmath3113 , such that @xmath3114 since @xmath66 is trivializable , there exists a smooth gauge transformation @xmath3115 on @xmath66 such that @xmath3116 .
let @xmath3061 .
we define @xmath3117 by @xmath3118 .
there exists a unique smooth gauge transformation @xmath3119 on @xmath3120 , such that @xmath3121 to see this , note that these conditions can be viewed as an ordinary differential equation for @xmath3119 with prescribed initial value .
since this solution depends smoothly on @xmath196 , there exists a unique smooth gauge transformation @xmath3122 on @xmath66 such that @xmath3123 , for every @xmath3061 .
the gauge transformation @xmath3124 on @xmath66 satisfies the equation @xmath3108 .
this follows from ( [ eq : g 0 iota a],[eq : iota x wt g 0 ] ) and flatness of @xmath79 and @xmath3125 .
this proves proposition [ prop : ka 0 ] .
the next result was used in the proofs of proposition [ prop : cpt bdd ] , remark [ rmk : bar j ] ( section [ sec : comp ] ) , and theorem [ thm : bubb ] .
let @xmath3126 be as in chapter [ chap : main ] .
we define the almost complex structure @xmath356 on @xmath116 as in ( [ eq : bar j ] ) .
the energy density of a map @xmath3127 is given by @xmath3128 where the norm is with respect to the metrics @xmath96 on @xmath65 and @xmath3129 on @xmath116 .
let @xmath66 be a smooth @xmath2-bundle over @xmath65 , @xmath79 a connection on @xmath66 , and @xmath168 an equivariant map .
we define @xmath3130 where the norm is taken with respect to the metrics @xmath96 on @xmath65 and @xmath97 on @xmath4 .
furthermore , we define @xmath3131 where @xmath229 is an arbitrary point in the fiber over @xmath364 .
[ prop : bar del j ] let @xmath66 be a smooth @xmath2-bundle over @xmath65 , @xmath156 , @xmath79 a @xmath167-connection on @xmath66 , and @xmath168 a @xmath2-equivariant map of class @xmath167 , such that @xmath890
. then we have @xmath3132 if @xmath257 also solves the equation @xmath946 then @xmath3133 [ le : weight ] let @xmath148 be a number , @xmath633 a finite tree , @xmath3134 vertices , @xmath3135 a function , and @xmath1184 a number .
assume that for every vertex @xmath396 we have @xmath3136 then @xmath3137 the next result was used in the proof of proposition [ prop : soft ] ( section [ sec : soft ] ) .
let @xmath925 be a metric space is allowed to attain the value @xmath41 .
] , @xmath2 a topological group , and @xmath3138 a continuous action by isometries . by @xmath3139
we denote the canonical projection .
the topology on @xmath586 , determined by @xmath591 , induces a topology on the quotient @xmath3140 .
[ le : subsubseq ] let @xmath586 be a topological space , @xmath927 , and @xmath3143 be a sequence .
then @xmath3144 converges to @xmath196 , as @xmath134 , if and only if for every subsequence @xmath1650 there exists a further subsequence @xmath1651 , such that @xmath3145 converges to @xmath196 , as @xmath1653 .
[ le : x g ] assume that @xmath586 is first - countable and that the action of every @xmath288 is a continuous self - map of @xmath586 .
let @xmath3146 , @xmath132 , be a sequence that converges to a point @xmath3147 , and @xmath196 be a representative of @xmath2441 .
then there exists a representative @xmath3144 of @xmath3148 , for each @xmath132 , such that @xmath3144 converges to @xmath196 .
next we explain the twisted connection @xmath188 , which appeared in the definition of the space @xmath176 , occurring in theorem [ thm : fredholm ] .
let @xmath3149 be a real ( smooth ) vector bundle .
we denote by @xmath3150 the affine space of ( smooth linear ) connections on @xmath133 .
let @xmath3151 .
let @xmath342 be a smooth manifold , and @xmath3152 be a smooth map .
we denote by @xmath3153 the pullback bundle .
the pullback connection @xmath3154 is uniquely determined by the equality @xmath3155 let @xmath2 be a lie group , @xmath2985 a ( right-)@xmath2-bundle , and @xmath3156 a @xmath2-equivariant vector bundle .
then the quotient @xmath3157 has a natural structure of a vector bundle over @xmath586 .
assume that @xmath2 acts on a manifold @xmath4 , and let @xmath3149 be a @xmath2-equivariant vector bundle .
we denote by @xmath3158 the space of @xmath2-invariant connections on @xmath133 .
we fix @xmath3159 , @xmath3160 , and @xmath3161 .
we define @xmath3162 for @xmath3163 , @xmath229 , and @xmath2078 .
we denote by @xmath3164 the quotient bundle @xmath3165 .
we define the connection @xmath3166 by @xmath3167 for @xmath3168 and @xmath3169 , where @xmath3170 is an arbitrary vector such that @xmath2079 , and @xmath3163 is the @xmath2-invariant section defined by @xmath3171 , for every @xmath229 .
this definition is independent of the choice of @xmath3172 , since the connection @xmath3173 is basic ( i.e. , @xmath2-invariant and horizontal ) . + the following lemma was mentioned in chapter [ chap : main ] .
let @xmath349 be as in that chapter .
let @xmath156 , @xmath210 , and @xmath165 be a @xmath2-bundle of class @xmath166 . recall the definition ( [ eq : bb p lam p ] ) of @xmath3174 .
we denote by @xmath3175 the group of gauge transformations on @xmath66 of class @xmath166 .
assume that @xmath204 .
let @xmath3176 and @xmath3177 be such that @xmath3178 .
let @xmath3179 .
we show that @xmath3180 .
it follows from hypothesis ( h ) that there exists @xmath993 such that @xmath3181 ( defined as in ( [ eq : m * ] ) ) .
furthermore , lemma [ le : si ] ( appendix [ sec : proofs homology ] ) implies that @xmath3182 , for every @xmath2986 , for which @xmath3183 is large enough .
we fix such a point @xmath2988 .
our hypothesis @xmath156 implies that @xmath66 is a @xmath512-bundle .
hence we may choose a path @xmath3184,p)$ ] such that @xmath3185 , for @xmath1801 .
consider the map @xmath3186\to g$ ] . by assumption , we have @xmath3187 . since @xmath3188 , it follows that @xmath3189 .
furthermore , the assumption @xmath3190 implies that @xmath3122 solves the ordinary differential equation @xmath3191 the hypothesis @xmath156 implies that the map @xmath3192\to{\mathfrak g}$ ] is continuous .
hence the equation ( [ eq : dot h ] ) is of the form @xmath3193 , where @xmath162 is continuous in @xmath278 and lipschitz continuous in @xmath3122 .
therefore , by the picard - lindelf theorem , we have @xmath3194 . in particular , we have @xmath3195 .
it follows that @xmath3196 .
this proves lemma [ le : free ] .
the following result was used in the proof of proposition [ prop : triv ] in section [ subsec : proofs ] .
we define the map @xmath2001 by @xmath2002 . for two topological spaces @xmath586 and @xmath1581 we denote by @xmath3211 the set of all continuous maps from @xmath586 to @xmath1581 , and by @xmath3212 $ ] the set of all ( free ) homotopy classes of such maps .
let @xmath49 be a finite dimensional complex vector space .
we denote by @xmath3213 the space of its ( complex ) endomorphisms of @xmath49 , by @xmath3214 the determinant map , and by @xmath3215 the group of automorphisms of @xmath49 .
we choose a hermitian inner product @xmath49 and denote by @xmath3219 the corresponding group of unitary automorphisms of @xmath49 .
the map @xmath3220 induces an isomorphism of fundamental groups , see e.g. ( * ? ? ?
* proposition 2.23 ) .
furthermore , the space @xmath3221 strongly deformation retracts onto @xmath3219 .
let @xmath3222 .
it follows that the map @xmath3223 descends to an isomorphism between the fundamental group @xmath3224 and @xmath3225 .
since this group is abelian , the map @xmath3226\ ] ] that forgets the base point @xmath3227 , is a bijection .
the statement of lemma [ le : aut ] follows from this .
the next lemma was used in the proof of theorem [ thm : l w * r ] ( section [ subsec : proof : thm : l w * r ] ) .
let @xmath586 and @xmath4 be manifolds , @xmath2 a lie group with lie algebra @xmath3 , @xmath62 an invariant inner product on @xmath3 , @xmath1106 a @xmath2-invariant riemannian metric on @xmath4 , and @xmath147 its levi - civita connection . for @xmath884 we denote by @xmath3228 the vector field on @xmath4 generated by @xmath2339 .
we define the tensor @xmath3229 by @xmath3230 a short calculation shows that @xmath3231 is skew - symmetric .
this two - form was introduced in @xcite .
the next lemma corresponds to ( * ? ? ?
* proposition 7.1.3(a , b ) ) .
let @xmath1772 be a @xmath2-bundle , @xmath1104 , @xmath3232 , @xmath3233 , and @xmath3234 .
we define the connection @xmath188 on @xmath3235 as on page .
let @xmath1753 and @xmath42 be as in chapter [ chap : main ] , and @xmath3237 .
the following remark was used in the proofs of theorems [ thm : fredholm ] ( section [ subsec : reform ] ) and [ thm : l w * r ] ( section [ subsec : proof : thm : l w * r ] ) . recall the definition ( [ eq : m * ] ) of @xmath3238 , and that @xmath1862 denotes the orthogonal projection onto @xmath1863 .
[ rmk : c]let @xmath2447 be compact .
we define @xmath3239 then @xmath2374 .
let @xmath883 .
then @xmath3240 is invertible , and @xmath3241 where the @xmath2936 s denote operator norms .
furthermore , @xmath3242 , for every @xmath3243 .
these assertions follow from short calculations .
assume that hypothesis ( h ) holds .
the following lemma was used in the proof of proposition [ prop : x x w ] in section [ subsec : proofs ] . for @xmath74
we denote by @xmath3244 the complex linear extension of the infinitesimal action .
it follows from hypothesis ( h ) that there exists @xmath1114 such that @xmath3246 .
we define @xmath3247|\,\big|\,\xi,\eta\in{\mathfrak g}:\,|\xi|\leq1,\,|\eta|\leq 1\big\},&\\ { { \nonumber}}&c_0:=\inf\left\{\displaystyle\frac{|l_x\xi|}{|\xi|}\,\bigg|\,x\in \mu^{-1}(\bar b_{{\delta}_0}),\,0\neq\xi\in{\mathfrak g}\right\}.&\end{aligned}\ ] ] since the action of @xmath2 on @xmath1894 is free , it follows that @xmath75 is injective , for @xmath3248 .
furthermore , by hypothesis ( h ) the set @xmath3249 is compact .
it follows that @xmath3250 .
we choose a positive number @xmath3251 , and we define @xmath3252 .
let @xmath1993 and @xmath3253
. then @xmath3254+l_x^*l_x\eta.\ ] ] using the last assertion in ( [ eq : l x * l x c ] ) , we have @xmath3255.\ ] ] by the first assertion in ( [ eq : l x * l x c ] ) , we have @xmath3256 . combining this with ( [ eq : d mu x l x],[eq : pr x l x ] ) and the second assertion in ( [ eq : l x * l x c ] ) , we obtain @xmath3257 inequality ( [ eq : c inf ] ) follows now from our choice of @xmath1117 .
this proves the claim and completes the proof of lemma [ le : u ] .
d. r. adams , l. i. hedberg , _
function spaces and potential theory _ , grundlehren der mathematischen wissenschaften ( fundamental principles of mathematical sciences ) , * 314 * , springer - verlag , berlin , 1996 , xii+366 pp . k. cieliebak , a. r. gaio and i. mundet i riera , d. a. salamon , _ the symplectic vortex equations and invariants of hamiltonian group actions _ , journal of symplectic geometry * 1 * ( 2002 ) , * no . 3 * , 543645 .
v. l. ginzburg and l. d. landau , _ on the theory of superconductors _ , zh . eksp .
jetp ) * 20 * ( 1950 ) , 10641082 .
english translation in : l. d. landau , _ collected papers _ , oxford , pergamon press , 1965 , p. 546 .
o. kavian , _ introduction la thorie des points critiques et applications aux problmes elliptiques _ ( introduction to critical point theory and applications to elliptic problems ) , mathmatiques & applications ( berlin ) , * 13 * , springer , 1993 , viii+325 pp .
m. kontsevich , _ enumeration of rational curves via torus actions _ , 335368 in r. dijkgraaf , c. faber , g. van der geer , eds . , _ the moduli space of curves _ , progress in mathematics * 129 * , birkhuser boston , 1995 . | consider a hamiltonian action of a compact connected lie group on a symplectic manifold @xmath0 .
conjecturally , under suitable assumptions there exists a morphism of cohomological field theories from the equivariant gromov - witten theory of @xmath0 to the gromov - witten theory of the symplectic quotient .
the morphism should be a deformation of the kirwan map .
the idea , due to d. a. salamon , is to define such a deformation by counting gauge equivalence classes of symplectic vortices over the complex plane @xmath1 .
the present memoir is part of a project whose goal is to make this definition rigorous .
its main results deal with the symplectically aspherical case .
the first one states that every sequence of equivalence classes of vortices over the plane has a subsequence that converges to a new type of genus zero stable map , provided that the energies of the vortices are uniformly bounded .
such a stable map consists of equivalence classes of vortices over the plane and holomorphic spheres in the symplectic quotient .
the second main result is that the vertical differential of the vortex equations over the plane ( at the level of gauge equivalence ) is a fredholm operator of a specified index .
potentially the quantum kirwan map can be used to compute the quantum cohomology of symplectic quotients .
[ multiblock footnote omitted ] |
among active galactic nuclei ( agns ) , blazars are a subclass characterized by the emission of strong non - thermal radiation across the entire electromagnetic spectrum , and in particular intense and variable @xmath3-ray emission above 100 mev ( see @xcite ) .
the typical observational properties include irregular , rapid and often very large variability , apparent super - luminal motion in the radio emission of the jet , flat radio spectrum , high and variable polarization at radio and optical frequencies .
these features are interpreted as the result of the emission of electromagnetic radiation from a relativistic jet that is viewed closely aligned to the line of sight @xcite and @xcite .
blazars emit across several decades of energy , from radio to tev energy bands , and thus they are the perfect candidates for simultaneous observations at different wavelengths .
multi - wavelength studies of variable @xmath3-ray blazars have been carried out since the beginning of the 1990s , thanks to the egret instrument onboard compton gamma - ray observatory ( cgro ) , providing evidence that the spectral energy distributions ( seds ) of the blazars are typically double humped with the first peak occurring in the ir / optical band in the so - called @xmath4 @xmath5 ( including flat spectrum radio quasars , fsrqs , and low - energy peaked bl lacs , lbls ) and in uv / x - rays in the so - called @xmath6 @xmath5 ( including high - energy peaked bl lacs , hbls ) .
this first peak is interpreted as synchrotron radiation from high - energy electrons in a relativistic jet .
the sed second component , peaking at mev
gev energies in @xmath4 @xmath5 and at tev energies in @xmath6 @xmath5 , is commonly interpreted as inverse compton scattering of seed photons by highly relativistic electrons @xcite , although other models have been proposed ( see e.g. @xcite for a recent review ) .
3c 279 is the best example of multi epoch studies at different frequencies performed by egret during the period 19912000 @xcite .
nevertheless , only a few object were detected on a timescales of about two weeks in the @xmath3-ray band , and simultaneously monitored at different energies in order to obtain a wide multi - frequency coverage . since its launch
, the agile satellite @xcite detected several flaring blazars , and thanks to the fast quick - look analysis procedure , extensive multi - wavelength campaign were organized for many of them . +
agile is an italian space agency ( asi ) mission successfully launched on 23 april 2007 and capable of observing cosmic sources simultaneously in x - ray ( 1860 kev ) and gamma - ray ( 30 mev30 gev ) bands with the coded - mask hard x - ray imager ( superagile ) and the gamma - ray imaging detector ( grid ) , respectively .
gamma - ray observations of blazars are a key scientific project of the agile satellite and in the last two years agile detected several blazars during high @xmath3-ray emission .
however , up to now a few sources were detected more than once during very high activity state : s5 0716 + 714 , pks 1510089 and 3c 454.3 .
we note that we detected at least one object for each blazar category : 3c 454.3 ( fsrq ) , pks 0537441 ( lbl ) , s5 0716 + 714 ( intermediate bl lac , ibl ) and mrk 421 ( hbl ) .
moreover , the @xmath3-ray activity timescales goes from a few days ( e.g. s5 0716 + 714 ) to several weeks ( e.g. pks 1510089 and 3c 454.3 ) and the flux variability for e @xmath7 100 mev could be negligible ( e.g. 3c 279 ) or extremely high ( e.g. 3c 454.3 and pks 1510089 ) . in the following sections we will present the most interesting results on multi - wavelength observations of some of the sources detected by agile .
3c 454.3 is the blazar which exhibited the most variable activity in the @xmath3-ray sky in the last two years . in the period
july 2007january 2009 the agile satellite monitored intensively 3c 454.3 together with @xmath0 , gasp - webt , rem , mitsume , @xmath1 , rxte , @xmath2 and integral observatories , yielding the longest multi - wavelength coverage of this @xmath3-ray quasar so far @xcite .
it is clear that this source is playing the same role for agile as 3c 279 had for egret .
100 mev in galactic coordinates integrated over the agile oberving period 2430 july 2007 , showing the detection of 3c 454.3 during flaring activity.,width=288 ] agile detected 3c 454.3 for the first time during a dedicated target of opportunity ( too ) activated immediately after an extremely bright optical flare in mid - july 2007 ( see fig .
2 ) @xcite . during a 6-day observation ,
the average @xmath3-ray flux was @xmath8 = ( 280 @xmath9 40 ) @xmath10 10@xmath11 photons @xmath12 s@xmath13 , more than a factor two higher than the maximum value reported by egret .
moreover , the peak flux on a daily timescale was of the order of ( 400 @xmath9 100 ) @xmath10 10@xmath11 photons @xmath12 s@xmath13 . since this detection
, this source become the most luminous object in the agile sky .
subsequently , a multi - wavelength campaign on 3c 454.3 was organized during november 2007 as reported in @xcite .
we monitored the source for about 18 days , obtaining simultaneous observations also from radio to hard x - ray energy bands .
figure 1 ( left panel ) shows the optical ( webt and rem ) , soft x - rays ( @xmath1/xrt ) , hard x - rays ( integral / ibis ) and @xmath3-rays ( agile / grid ) light curves .
the @xmath3-ray flux appears to be variable on a short timescale ( 24 - 48 hours ) , while the soft and hard x - ray energy bands show a less pronunced variability .
the average @xmath3-ray flux over the entire campaign is @xmath8 = ( 170 @xmath9 13 ) @xmath10 10@xmath11 photons @xmath12 s@xmath13 .
the @xmath14-band optical flux appears extremely variable , with a brigthening of several tenths of magnitude in a few hours .
a correlation analysis between optical and @xmath3-ray flux variations is consistent with no time - lags .
the average @xmath3-ray photon index during the entire campaign is substantially harder than that reported in the past by egret ( @xmath15 = 1.73 @xmath9 0.16 vs. @xmath16 = 2.22 @xmath9 0.06 ) .
figure 1 ( right panel ) shows the sed , in which the dominant emission mechanism over 100 mev seems to be the inverse compton scattering of relativistic electrons in the jet on the external photons from the broad line region ( blr ) . in december 2007 , the activity of the source continued to be high and agile collected other 15 days of monitoring on 3c 454.3 , detecting the source with a flux of the order of 250 @xmath10 10@xmath11 photons @xmath12 s@xmath13 .
therefore , we triggered a second multi - wavelength campaign with @xmath0 , rem , webt , mitsume , @xmath1 and @xmath2 .
the hard @xmath3-ray spectrum and the high compton dominance observed by agile suggested the contribution of external compton of seed photons from a hot corona .
results of this campaign will appear in @xcite . during may
june 2008 , agile performed the most intensive and long campaign on 3c 454.3 , resulting in an almost uninterrupted 50-day long monitoring , collecting also data with webt , rem , @xmath1 and rxte .
this long observation showed the highly variable nature of 3c 454.3 , both on short and long time scales .
the multi - wavelength study of the source will be presented in a forthcoming paper @xcite .
ph @xmath12 s@xmath13 for e @xmath7 100 mev .
the downward arrows represent 2-@xmath17 upper limits . in the bottom panel , the @xmath14-band optical light curve as observed by gasp - webt .
the shaded regions indicate the two high - activity periods in @xmath3-rays.,width=312 ]
-ray observation in september 2007 ( green triangles ) .
historical data over the electromagnetic spectrum relative to a ground state of the source and egret non - simultaneous data are represented with blue squares .
red dots represent historical data simultaneous with a high x - ray state.,width=302 ] the intermediate bl lac object s5 0716 + 714 was observed by agile during two different periods : 423 september and 23 october 1 november 2007 , as discussed in @xcite . in mid - september
the source showed an high @xmath3-ray activity with an average flux of f@xmath18 = ( 97 @xmath9 15)@xmath1010@xmath11 photons @xmath12 s@xmath13 and a peak flux of f@xmath18 = ( 193 @xmath9 42)@xmath1010@xmath11 photons @xmath12 s@xmath13 .
the flaring activity appeared to be very fast with an increase of the @xmath3-ray flux by a factor of four in three days ( see fig .
this source was detected by egret in a low / intermediate @xmath3-ray level ( f@xmath18 = ( 2040 ) @xmath1010@xmath11 photons @xmath12 s@xmath13 ) , therefore the flux detected by agile is the highest flux detected by this object and one of the most high flux observed by a bl lac object .
an almost simultaneous gasp - webt optical campaign started after the agile detection and the resulting sed is consistent with a two - components synchrotron self compton ( ssc ) model ( fig .
recently @xcite has estimated the redshift of the source ( z = 0.31 @xmath9 0.08 ) and this allowed us to calculate the total power transported in the jet , which results extremely high and at limit of the maximum power generated by a spinning black hole of 10@xmath19 m@xmath20 @xcite . during october 2007 , agile detected the source for the second time , as reported in @xcite , at a flux about a factor 2 lower than september one with no significant variability .
instead , simultaneously @xmath1 observed strong variability in soft x - ray , moderate variability at optical / uv and approximately constant hard x - ray flux .
also this behaviour is compatible with the presence of two different ssc components in the sed .
3c 273 was the first extragalactic source detected simultaneously by the grid and superagile detectors during a pre - planned campaign over three weeks between mid - december 2007 and january 2008 with also simultaneous rem , @xmath1 , rxte and integral coverage . during this campaign , whose
results are reported in detail in @xcite , the average flux in the 2060 kev energy band is ( 23.9 @xmath9 1.2 ) mcrab , whereas the source was detected by the grid only in the second week , with an average flux of f@xmath18 = ( 33 @xmath9 11)@xmath1010@xmath11 photons @xmath12 s@xmath13 .
the comparison of the light curves seems to indicate no optical variability during the whole campaign and a possible anti - correlation between the @xmath3-ray emission and the soft and hard x - rays .
the sed is consistent with a leptonic model where the soft x - ray emission is produced by the combination of ssc and ec models , while the hard x - ray and @xmath3-ray emission is due to external compton scattering by thermal photons of the disk .
the spectral variability between the first and the second week is consistent with an acceleration episode of the electron population responsible for the synchrotron emission , leading to a shift in the inverse compton peak toward higher energies ( see fig .
/uvot and rem data .
the grey data refers to @xmath1/xrt observations , performed in the third week .
the model of the first week is reported as a dot - dashed line for comparison.,width=312 ]
on 2008 june 8 , veritas announced the detection of a tev flare from w comae @xcite .
about 24 hours later , agile was already pointing towards the source and detected it @xcite .
this source belongs to an agile cycle-1 guest investigator ( e. pian ) , and the results of a multi - wavelength campaign involving @xmath1 , agile and veritas will appear in a forthcoming paper , see @xcite in these proceedings for more details . during the too observation towards w comae ,
agile detected also the hbl object mrk 421 both with superagile ( see @xcite ) and grid ( preliminary results are reported in @xcite ) .
a multi - wavelength campaign with simultaneous mev
gev ( agile ) and tev ( magic and veritas ) data , together with webt , rxte and @xmath1 observations was organized and the results are presented in @xcite and discussed also in these proceedings @xcite .
agile , during its first two years of sky monitoring , demonstrated the importance of its wide field of view ( @xmath212.5 sr ) in detecting transient sources at high off - axis angle . moreover , its unique combination of a @xmath3-ray detector with an hard x - ray detector allowed us to study in detail the high energy part of the blazar seds .
the synergy between the agile wide field of view , its fast response to external triggers and the availability of network of ground - based telescope , allowed us to obtain a multi - wavelength coverage for almost all the detected sources .
up to now , agile detected already known @xmath3-ray emitting blazars .
this evidence , combined with the early @xmath22-lat results , which show a predominant fraction of already known @xmath3-ray emitters among its flaring sources , poses possible constraints on the emitting properties of the blazars .
future agile and @xmath22-lat observations will be fundamental to unveil the real nature of the blazars .
the agile mission is funded by the italian space agency ( asi ) with scientific and programmatic participation by the italian institute of astrophysics ( inaf ) and the italian institute of nuclear physics ( infn ) .
r. c. hartman , d. l. bertsch , s. d. , bloom , et al .
, _ apjs _ , 123 , 79 ( 1999 ) r. d. blandford @xmath23 m. rees , _ in bl lac objects _
a. m. wolfe , pittsburgh press ( 1978 ) c. m. urry @xmath23 p. padovani , _ publications of the astron .
soc . of the pacific_,107 , 803 ( 1995 ) m. ulrich , l. maraschi , c. m. megan , _ ara@xmath23a _ , 35 , 445 ( 1997 ) m. bttcher , _ astrophysics and space science _ , 309 , 95 ( 2007 ) r. c. hartman , m. bttcher , g. aldering , et al . ,
_ apj _ , 553 , 683 ( 2001 ) m. tavani , et al . ,
_ astronomy and astrophysics _ , in press , ( 2009 ) s. vercellone , f. dammando , v. vittorini , et al .
, _ in preparation
_ s. vercellone , a. w. chen , a. giuliani , et al . ,
_ apjl _ , 676 , 13 ( 2008 ) s. vercellone , a. w. chen , v. vittorini , et al .
, _ apj _ , 690 , 1018 ( 2009 ) i. donnarumma , g. pucella , v. vittorini , et al . , _ to be submitted to apj _ ( 2009 ) a. w. chen , f. dammando , m. villata , et al . ,
_ astronomy and astrophysics _ , 489 , l37 , ( 2008 ) k. nilsson , t. pursimo , a. sillanp , et al . ,
_ astronomy and astrophysics _ , 487 , l29 , ( 2008 ) v. vittorini , m. tavani , a. paggi , et al .
, _ submitted to apj _ ( 2009 ) p. giommi , s. colafrancesco , s. cutini , et al .
, _ astronomy and astrophysics _ , 487 , l49 ( 2008 ) l. pacciani , i. donnarumma , v. vittorini , et al . , _ astronomy and astrophysics _ , 494 , 49 ( 2009 ) s. swordy , _ the astronomer s telegram _ , 1565 , 1 ( 2008 ) f. verrecchia , et al . , _ the astronomer s telegram _ , 1582 , 1 ( 2008 ) g. maier , e. pian , et al .
_ these proceedings _ ( 2009 ) e. costa , et al . , _ the astronomer s telegram _ , 1574 , 1 ( 2008 ) c. pittori , et al . , _ the astronomer s telegram _ , 1583 , 1 ( 2008 ) i. donnarumma , v. vittorini , s. vercellone , et al .
, _ apj _ , 691 , l13 ( 2009 ) r. wagner , et al . , _ these proceedings _ ( 2009 ) | during the first two years of observation , agile detected several blazars at high significance : 3c 279 , 3c 454.3 , pks 1510089 , s5 0716 + 714 , 3c 273 , w comae , mrk 421 and pks 0537441 .
we obtained long - term coverage of 3c 454.3 , for a total of more than three months at energies above 100 mev .
3c 273 was the first blazar detected simultaneously by the agile gamma - ray detector and by its hard x - ray monitor .
s5 0716 + 714 , an intermediate bl lac object , exhibited a very fast and intense gamma - ray transient during an optical high - state phase , challenging the current theoretical models for energy extraction from a rotating black hole , while w comae and mkn 421 were detected during an agile target of opportunity ( too ) repointing , and were simultaneously detected by cherenkov telescopes .
thanks to the rapid dissemination of our alerts , we were able to obtain multi - wavelength too data from other observatories such as @xmath0 , @xmath1 , rxte , @xmath2 , integral , magic , veritas , as well as radio - to - optical coverage by means of the webt consortium and rem , allowing us to study truly simultaneous spectral energy distributions from the radio to the tev energy band .
gamma - ray astronomy , active galactic nuclei , blazars |
we can learn a lot on the functioning of the human brain in health and disease when we consider it as a large - scale complex network , whose properties can be analyzed using graph theoretical analysis @xcite . with the advent of miscellaneous and noninvasive mri techniques , this _
connectome _ has been mainly characterized by either structural or functional connectivity .
structural connectivity is commonly based on white matter tracts quantified by diffusion tractography @xcite ; functional connectivity relies on the other hand on statistical dependencies such as temporal correlation @xcite .
an important addition to this framework can come from effective connectivity analysis @xcite , in which the flow of information between even remote brain regions is inferred by the parameters of a predictive dynamical model .
+ for some techniques , such as dynamic causal modelling ( dcm ) and structural equation modelling @xcite , these models are built and validated from specific anatomical and physiological hypotheses .
other techniques such as granger causality analysis ( gca ) @xcite , are on the other hand data - driven and rely purely on statistical prediction and temporal precedence . while powerful and widely applicable , this last approach could suffer from two main limitations when applied to blood - oxygenation level - dependent ( bold)-functional mri ( fmri ) data : confounding effect of hemodynamic response function ( hrf ) and conditioning to a large number of variables in presence of short time series .
early interpretation of fmri based directed connectivity by gca always assumed homogeneous hemodynamic processes over the brain ; several studies have pointed out that this is indeed not the case and that we are faced with variable hrf latency across physiological processes and distinct brain regions @xcite . recently , a number of studies have addressed this issue proposing to model the hrf according to several recipes @xcite . as well , a recent study has proposed that it would still feasible to infer connectivity at bold level , under the assumption that granger causality is theoretically invariant under filtering @xcite and that the hrf can be considered as a filter .
it is still unclear whether and how specific effects related to hrf disturb the inference of temporal precedence .
in addition a simulated or experimental ground truth is difficult to obtain , though some studies on simulated fmri data have tried to reveal the relationship between neural - level and bold - level causal influence @xcite .
a considerable help to obtain the hrf for deconvolution could come from multimodal imaging where the high temporal resolution of eeg is combined to the high spatial resolution of fmri , but this experimental approach is still far from being optimal and widely applicable .
hrf has been studied almost since the early days of fmri @xcite . for task - related fmri
, neural population dynamics can be captured by modeling signal dynamics with explicit exogenous inputs @xcite , i.e. deconvolution according to the explicit task design is possible in this case @xcite . for resting - state fmri on the other hand
, the absence of explicit inputs makes this task more difficult , unless relying on some specific prior physiological hypothesis @xcite . in order to overcome these issues and to allow a more general approach , here we present a simple and novel blind - deconvolution technique for bold - fmri signal .
+ coming to the second limitation , in order to distinguish among direct and mediated influences in multivariate datasets it is necessary to condition the analysis to other variables .
a bivariate analysis would indeed lead to the detection of many false positives . in presence of a large number of variable and short time series
, a fully multivariate conditioning could lead to computational problems due to the overfitting . furthermore
, conceptual issues would arise in presence of redundant variables @xcite . in this paper
we thus apply partial conditioning for granger causality ( pcgc ) to a limited subset of variables , as recently proposed @xcite for reconstructing the bold and deconvolved bold level effective connectivity network ( ecn ) and compare them .
hemodynamic deconvolution of bold signal is performed as described in @xcite . under the assumption that the transformation from neural activation to bold response can be modeled as a linear and time invariant system ,
measured fmri data @xmath0 can be seen as the result of the convolution of neural states @xmath1 with a hrf @xmath2 : @xmath3 where @xmath4 is the time and @xmath5 denotes convolution .
@xmath6 is the noise in the measurement , which we assume to be white . since the right side of the above equation includes three unobservable quantities , in order to solve the equation for @xmath2 we need to substitute @xmath1 with a hypothetical model of the neural activation for @xmath1 .
here we employ a simple on - off model of activation to model @xmath1 : @xmath7 where @xmath8 is the delta function .
this allows to fit the hrf @xmath2 according to @xmath9 using a canonical hrf ( two gamma functions ) and two derivatives ( multivariate taylor expansion : temporal derivative and dispersion derivative ) @xcite , as is common in most fmri studies .
+ once calculated @xmath2 , we can obtain an approximation @xmath10 of the neural signal from the observed data using a wiener filter @xmath11 let @xmath12 , @xmath13 , @xmath14 , and @xmath15 be the fourier transforms of @xmath2 , @xmath0 , @xmath16 , and @xmath17 , respectively .
then @xmath18 where @xmath19 denotes complex conjugate .
the estimation @xmath10 of the neural states @xmath1 is then given by @xmath20 where @xmath21 is the inverse fourier transform operator .
+ for task - related fmri , the stimulus function provides the prior expectations about neural activity and a generative model whose inversion corresponds to deconvolution ; this is in principle not the case for resting - state fmri .
nonetheless there is increasing evidence of specific events and neural states that govern the dynamics of the brain at rest @xcite .
furthermore , tagliazucchi et al . proposed that these events are reflected by relatively large amplitude bold signal peaks and thus that such fluctuations could encode relevant information from resting - state fmri recordings @xcite .
inspired by their work , we consider resting - state fmri as _ spontaneous event - related _ , and we propose to extract the hrf from those pseudo - events . after doing this
, we can employ the deconvolution model in the same way as described above .
it is known that the bold response is much slower than the neural activation that is presumed to drive it .
consequently , the peak of the bold signal lags behind the peak of neural activation ( i.e. by @xmath22 points ) .
so here we assume that these events are generated from @xmath9 .
+ glover pointed out that the noise spectrum in task - related fmri can be obtained from time series measurements in nonactivated cortical regions @xcite ; here we extend the model to cope with resting - state fmri for which there is no explicit activation . in this study we assumed covariance of noise @xmath23 equal to @xmath24 $ ] .
+ in order to obtain a value for @xmath22 , we search all integer values in the interval @xmath25 $ ] , where @xmath26 is an arbitrary maximum value , choosing the one for which the noise error covariance is smallest as the onset . by this method
we can perform deconvolution on all bold signals , requiring no information on timing or a priori spatial information of events ; furthermore , the time series could be the average of time series over a region of interest with any scale , or series extracted by independent or principal component analysis . a flow chart for bold signal deconvolution
is shown in fig.[fig1 ] .
+ this is the pseudo - code for our procedure . + ] here we employ a methodology proposed in @xcite which allows to compute granger causality conditioned to a limited number of variables in the framework of information theory .
the idea is that conditioning on a small number of variables , chosen as the most informative for the candidate driver variable , is sufficient to remove indirect interactions for sparse connectivity patterns . + we consider @xmath27 covariance - stationary variables @xmath28 , denoting the state vectors as : @xmath29 @xmath30 being the model order .
let @xmath31 be the mean squared error prediction of @xmath32 on the basis of the vectors @xmath33 .
the partially conditioned granger causality index @xmath34 is defined as follows : @xmath35 + where @xmath36 is a set of the @xmath37 variables , in @xmath38 , which are most informative for @xmath39 .
we adopt the following approximate strategy for @xmath40 : given the previous @xmath41 , the set @xmath42 is obtained adding the variable with greatest information gain .
this is repeated until @xmath37 variables are selected .
a method for establishing a ground truth for fmri data has not reached a general consensus . recently
a benchmark dataset , netsim @xcite has attracted a lot of attention .
previous studies have shown that lag - based methods perform very poorly on these datasets ; it is anyway worthy to mention that these data are simulated under the dcm framework , contain no reciprocal connections and only gaussian noise , limiting their universality as ground truth .
here we analyzed the largest of these datasets , consisting of 50 nodes .
after deconvolution the sensitivity improved significantly , increasing from 20@xmath43 to 30@xmath43 .
also the specificity improved from 88@xmath43 to 94@xmath43 .
this does not render gc the method of choice for these data , for which we also have to point out that neural events and noise are not distinguishable , but gives nonetheless an indicative result for the usefulness of deconvolution of the bold signal . in order to investigate the role of repetition time ( tr ) on the deconvolution procedure and on the effective network reconstruction
, our analyses were performed on a resting - state fmri dataset which has been publicly released in the 1000 functional connectomes project ] .
all participants had no history of neurological and psychiatric disorders and all gave the informed consent approved by local institutional review board . during the scanning participants
were instructed to keep their eyes closed , not to think of anything in particular , and to avoid falling asleep . +
two data sets with different tr ( tr=1.4s and tr=0.645s ) were acquired on siemens 3 t trio tim scanners using developed multiplexed echo planar imaging @xcite .
as specified in detail below , two resting - state fmri data are included in the protocol - a tr=0.645s ( 3 mm isotropic voxels , 10 minutes ) to provide optimal temporal resolution and tr=1.4s ( 2 mm isotropic voxels , 10 minutes ) to provide optimal spatial resolution .
the third data set , acquired on a 4 t scanner , contains standard resting - state fmri acquisitions with a longer tr ( tr=3 , 4 mm isotropic voxels , 5 minutes ) . for more detail on subject and data information
, please see website@xmath44 .
preprocessing of resting - state images was performed using the statistical parametric mapping software ( spm8 , http://www.fil.ion.ucl.ac.uk/spm ) .
the preprocessing included slice - timing correction relative to middle axial slice for the temporal difference in acquisition among different slices , head motion correction , spatial normalization into the montreal neurological institute stereotaxic space , resampling to 3-mm isotropic voxels .
8(9 ) subjects were excluded from the dataset with tr=0.645s ( tr=1.4s ) because either translation or rotation exceeded @xmath45 mm or @xmath46 , resulting in 16(tr=0.645s ) and 15(tr=1.4s ) subjects each one scanned in two sessions which were used in the analysis ) .
one subject whose data were too noisy was excluded from the tr=3 dataset , resulting in 10 subjects used in the analysis . in order to avoid introducing artificial local spatial correlations between voxels , no spatial smoothing was applied for further analysis , as previously suggested @xcite .
the functional images were segmented into 90 regions of interest ( roi ) using automated anatomical labeling ( aal ) template as reported in previous studies . for each subject ,
the representative time series of each roi was obtained by averaging the fmri time series across all voxels in the roi @xcite .
several procedures were used to remove possible spurious variances from the data through linear regression .
these were i ) six head motion parameters obtained in the realigning step , ii ) signal from a region in cerebrospinal fluid , iii ) signal from a region centered in the white matter , iv ) global signal averaged over the whole brain .
the bold time series were deconvolved into neural state signal using the above mentioned approach .
the topological properties of the effective connectivity network were defined on the basis of a @xmath47 binary directed graph @xmath48 , consisting of nodes and directed edges : @xmath49 where @xmath50 refers to the directed edge from roi @xmath51 to roi @xmath52 in the graph .
@xmath53 indicates the threshold . in a directed graph @xmath50
is not necessarily equal to @xmath54 . considering that the graph we focused on is directed ,
all topological properties were calculated on incoming and outgoing matrix , respectively .
graph theoretical analyses were carried out on the effective connectivity network using the brain connectivity toolbox @xcite . as previous studies suggested that the brain networks of each subject normally differ in both the number and weighting of the edges @xcite , we applied a matching strategy to characterize the properties of effective connectivity network .
both the global and local network efficiencies have a propensity for being higher with greater numbers of edges in the graph @xcite . modifying the sparsity values ( number of edges ) of the adjacency matrix also altered the graph s structure .
as a consequence it was suggested that the graphs to be compared must have ( a ) the same number of nodes and ( b ) the same number of edges @xcite .
the cost was defined as the ratio of the number of existing edges divided by the maximum possible number of edges in a network .
since there is currently no formal consensus regarding selection of cost thresholds , here we selected a range of 0.05 to 0.14 with step = 0.01 for subsequent network analyses .
the lower bound was chosen as the one yielding a sparse graph with mean degree @xmath55 ( total number of edges @xmath56 where @xmath57 ) .
the upper threshold corresponded to the smallest significant value of granger causality ( f - test with @xmath58 ) across all subjects ) . for effective connectivity network at each cost threshold
, we calculated both overall topological properties and nodal characteristics @xcite .
the overall topological properties included i ) small - worldness ( @xmath59 ) , related to normalized clustering coefficient ( @xmath60 ) and normalized characteristic path length ( @xmath61 ) ; ii ) network efficiency , divided in local efficiency ( @xmath62 ) and global efficiency ( @xmath63 ) .
the nodal characteristics included i ) the nodal degree , that quantifies the extent to which a node is relevant to the graph , and ii ) the nodal efficiency , that quantifies the importance of the nodes for the communication within the network @xcite .
furthermore we calculated the area under the curve ( auc ) across all cost values for the above mentioned network properties .
this quantity represents a summarized scalar for topological characterization of brain networks independent of single cost threshold selection .
we tested the proposed deconvolution method on resting - state fmri data ; following the procedure summarized in the box , firstly we set a maximum time lag from a given threshold crossing , and obtain an optimal value for this lag , denoted with @xmath22 . the histograms for @xmath22 , reported in fig.[fig2 ] show a maximum around @xmath64s , which is consistent with a previous study according to which the latency delay is @xmath65s in gray matter @xcite .
it is worth to mention that the lower tr could allow a more accurate estimation of the lag . + .
( @xmath19:without regression of global signal ) [ fig2 ] ] to assess the effect of deconvolution , we compared the shape of voxel based hrf over the whole brain using different trs .
we focused on three parameters : response height , time - to - peak , and full - width at half - max ( fwhm ) as potential measures of response magnitude , latency , and duration . using principal component analysis we determined the average intersubject variability of hrf maps .
we found that the first component of hrf accounted for @xmath66(response height ) , @xmath67(time to peak ) and @xmath68(fwhm ) of the variance .
furthermore , the spatial distribution is very similar to the mean group map .
the mean group results are plotted in fig.[fig3 ] . the response height ,
time to peak and fwhm of hrfs differ across brain regions , as a consequence of multiple factors including neural activity differences , global magnetic susceptibilities , vascular differences , baseline cerebral blood flow , slice timing differences etc .
@xcite . these patterns are remarkably similar across subjects and trs , reflecting the robustness of the proposed approach .
+ : without regression of global signal)[fig3 ] ] as another indicator of the stability of the proposed joint of deconvolution and pcgc approach we tested the variance of causality matrix across all subjects .
we calculated the variance of the granger causality matrix obtained both on the bold and deconvolved bold level signal .
firstly , we converted the matrix to z - scores , then we calculated the variance of each matrix element , finally summing up the all these values into an overall variance index .
the variance of granger causality matrix obtained from the deconvolved signal is much lower than the one of the bold level matrix for all tr values ( fig.[fig6 ] ) . also , pcgc method kept the variance lower than full conditioned gc method .
this result was confirmed testing a network at another scale using 1024 nodes ( fig.[fig6 ] , the native aal segmentation was parcellated into 1024 micro regions of interest of approximately identical size across both hemispheres @xcite ; in this case we could not test full conditioned gc due to small number of samples ) . ] as shown in previous studies , several sources of spurious variance should be removed by regression : motion artifacts , white matter and ventricular time courses .
still , the effects of regression against the global signal , calculated by averaging across all voxels within a whole brain mask , are debated . in order to evaluate this effect on our data we calculated spatial correlation between the the group mean image of hrf(response height , time to peak , fwhm ) with and without regression of global signal in the preprocessing step , obtaining high pearson correlation between them : r=0.97(response height ) , 0.90(time to peak ) , 0.88(fwhm ) .
we can thus conclude that regression against global signal still preserved the spatial distribution .
when trying to reconstruct effective connectivity networks , we are faced with the problem of coping with a large number of variables , when the application of multivariate granger causality may be questionable or even unfeasible , whilst bivariate granger causality would detect also indirect interactions .
conditioning on a large number of variables requires an high number of samples in order to get reliable results .
reducing the number of variables that one has to condition over would thus provide better results for small data - sets . in the general formulation of granger causality
, one has no way to choose this reduced set of variables ; on the other hand , in the framework of information theory , it is possible to individuate the most informative variables one by one .
+ the optimal model order @xmath30 ( order of the autoregressive model in granger causality , embedding dimension in transfer entropy ) for deconvolved bold and bold signal was determined by leave - one - out cross - validation , and was found to be 3 for tr=0.645s , 2 for tr=1.4s and 1 for tr=3s . under the gaussian assumption , we constructed effective connectivity network using pcgc method .
we firstly have to determine the number of variables upon which conditioning . to do this we look at how much uncertainty we eliminate adding an extra variable , letting the number of conditioning variables included @xmath37 vary from 1 to 20 .
this uncertainty can be expressed in terms of the information that we gain adding an extra variable . in fig.[fig4 ] , we plot the information gain as a function of @xmath37 ; as expected , both this quantity and its increment decrease monotonically with @xmath37 . + ) ,
when the @xmath69-th variable is included , is plotted versus @xmath37 .
the information gain is averaged over all the variables .
[ fig4 ] ] we can observe that the knee of the curves occurs when six variables are considered .
this happens also when we consider different brain prior templates with 17 or 160 nodes ( results not reported here ) .
this could be connected to the fact that the average number of modules which explain the equal - time correlations of resting brain is close to six @xcite , therefore picking one variable from each module is sufficient to have most of the information , about a given channel , that can be obtained from the remaining channels , and this independently on the number of nodes .
the effect of deconvolution , for the information gain , is to qualitatively raise the curve for tr=0.645s , and to lower them for tr=1.4s .
this trend ( not statistically significant ) might be the result of two competing effects , the fact the deconvolution may remove spurious correlations and/or restore genuine correlations obscured by noise .
+ synthesizing the knee of the curves , sensitivity and specificity , we consider @xmath70 as the most appropriate number of most informative variables to include in the conditioning procedure .
+ the global topological properties of brain ecn at deconvolved bold and bold level rely on the choice of thresholds .
we used multiple cost thresholds and the auc to evaluate the stability of the topological organization ( table 1 ) .
an higher number of differences between the two networks was found with a ( relatively ) longer tr ( tr=1.4s ) .
specifically , the auc of small - worldness ( @xmath59 ) , normalized clustering coefficient ( @xmath60 ) , clustering coefficient(@xmath71 ) and local efficiency(@xmath62 ) displayed the most significant differences , similar to what emerged with tr=0.645s .
for the data set with shorter tr we found significant differences in the characteristic path length and global efficiency of the outgoing network , whereas the most relevant differences were found for the incoming network with the longer tr .
.comparison of auc between deconvolved bold and bold .
@xmath70 , y : @xmath72 , fdr corrected ; n : otherwise . [ cols="^,^,^,^,^,^,^ " , ]
+ comparing the two networks on nodal degree , nodal global efficiency and nodal local efficiency revealed modifications in deconvolved bold and bold level ( fig.[fig5 ] ) .
the patterns of nodal degree modifications resembled to those of nodal global efficiency in incoming network in both tr=0.645s and tr=1.4s fmri data sets .
in addition , more brain regions showed modified nodal degree and ( global / local ) efficiency in tr=0.645s rather than tr=1.4s data .
, fdr corrected , ( @xmath70 ) .
blue indicate negative values , red positive values .
the point size is proportional to the absolute z value.[fig5 ] ]
in this study we proposed a novel methodology to achieve deconvolution in resting state data using spontaneous pseudo events , and to apply partially conditioned granger causality to the analysis of fmri data . in our opinion this joint approach is the most convenient to infer effective connectivity with granger causality from resting state fmri data .
in the absence of a well defined ground truth , and in the light of the still active and unresolved debate on the usefulness of hrf deconvolution granger causality based connectivity , we limit ourselves to validate the stability of the proposed method and indicate a possible path for the continuation of this debate , quantifying and comparing the overall topological properties of large - scale ecns on deconvolved bold - level versus bold - level signals , investigating also the effect of different time resolutions ( tr=0.645s and tr=1.4s ) .
+ previous discussions on evaluating effective connectivity from fmri data reached the conclusion that it is better to use state - space model for inferring causality on hidden neural states @xcite .
a pioneering eeg - fmri study provided the first experimental substantiation of the theoretical possibility to improve interregional coupling estimation from hidden neural states of fmri @xcite .
though promising @xcite , these implications are still limited by the fact that multimodal recording is invasive and not applicable to healthy controls . as a consequence ,
data - driven methods for substantiating the confounding variability of haemodynamics have been developed .
the two available types of state space models in estimation of hrf @xcite : the generic ( linear canonical / spline hrf ) @xcite and biophysically informed models ( dcm nonlinear hrf)@xcite .
generic models are widely applicable but lack specific biophysical constraints @xcite , while biophysically informed models are constrained by the hypothesis itself @xcite .
a recently proposed , biophysically informed bind deconvolution approach based on the state - of - the - art cubature kalman filtering could be a useful tool for resting - state fmri @xcite . in the present study ,
however , we use a simpler approach which employs the generic linear canonical hrf for deconvolution .
it is worth to point out that the significant differences between bold- and deconvolved bold - level effective connectivity found in complex network measures can not absolutely exclude the misestimation of hrf .
furthermore hrf latency effect does not always critically affect the evaluation of mutual influence , so ecns on bold and deconvolved bold level could have important consistencies @xcite .
+ findings from brain connectivity studies have now demonstrated that the human brain network exhibits robust small - world topological properties , not only in the anatomical connectivity ( reconstructed by diffusion tractography ) @xcite and functional connectivity network @xcite , but also in effective connectivity network @xcite .
the current results also suggested that the ecns obtained from bold and deconvolved data , with shorter and longer tr , have prominent small - world attributes , which would thus be confirmed as a general signature of robust organization of complex brain networks .
small - worldness indicates indeed an optimal balance between segregated and integrated organization to process the information @xcite .
for relatively longer tr we found significant differences between bold and deconvolved ecns .
although an explanation based on precise neurobiological mechanisms is still not evident , we can suggest that the bold effect results from a more complex sequence of effects linking neuronal activity , vascular changes and mri signal @xcite .
hemodynamic delay , and hence the correct onset of the events is indeed hard to capture with a long tr @xcite .
+ in complex networks organization , the normalized clustering coefficient and the clustering coefficient are two key measures .
they quantify the extent of local cliquishness or of local efficiency of information transfer of a network @xcite , reflecting the local properties of network topologies . for longer tr , we observed significant differences between the two level ecns .
thus the short - scale or local - scale network properties are indeed affected by deconvolution . moreover , the normalized characteristic path length and the characteristic path length quantify global efficiency or the capability for parallel information propagation of a network @xcite .
these two measurements along with global efficiency are mainly associated with long - range connections ensuring effective interactions or rapid transfers of information @xcite .
it is widely accepted that long - range axonal connectivity being an important indicator of the functional - anatomical organization of the human cortex @xcite .
this study reported no differences in long - range network organization .
+ it is known that resting - state functional connectivity studies using either seed functional connectivity or independent component analysis benefit from higher sampling rates to adequately sample undesirable respiration and cardiac effects @xcite , while for event - related fmri , faster sampling could allow for a better characterization of the hemodynamic response .
the same applies to gca .
the previous simulations showed that accuracy of granger causality depends on volume tr , faster sampling interval increased the detection capacity of gca of fmri data to neural causality @xcite . in this paper
, we focus on resting - state fmri data with tr=0.645s and 1.4s to maximally escape information loss due to low sampling . considering the limitation of acquisition sequence
, the conventional fast tr data acquisition brings to the loss of the fine spatial resolution @xcite .
+ other methodological considerations are worth to be mentioned .
the first one concerns data preprocessing . as a general idea
spatial smoothing can reduce the noise and increase signal - to - noise ratio , therefore improving the accuracy of detecting of neural event @xcite .
here we do not include this step .
as we used aal template , spatial smoothing would blur the boundary among these regions , which may affect the gc inference .
temporal filtering is frequently a necessary step for functional connectivity analysis of resting - state fmri data . in line with previous studies that considered a low model order in gca @xcite
, we did not performed low - pass filtering .
+ secondly , graph theoretic approach is one of the most powerful and flexible approaches to investigate functional and structural brain connectome ; still some controversies remain , concerning the definition of network nodes and edges @xcite .
different node definitions by prior anatomic brain templates @xcite or node scales @xcite could produce different results . in future works , more brain templates and more node scales comparison for effective connectivity network should be explored .
h. chen was supported by the natural science foundation of china ( no . 61125304 and no .
61035006 ) .
wu gratefully acknowledges the financial support from china scholarship council(2011607033 ) .
53 natexlab#1#1url # 1`#1`urlprefix[2][]#2 [ 2]#2 , , , , , , . .
, , , . . ,
, , , . . ,
, , , . . ,
, , , , , , , . . ,
, , , . . ,
, , , , , , , , , . . ,
. , . . , . , , ,
. . , . , , , ,
. . , . , , ,
. . , . , , , , , , ,
. . , . , , , , ,
. . , . , , , , . , . , , ,
. . , . , , , , ,
. . , . , , , ,
. . , . , , , , , , ,
. . , . , , ,
volume . .
, , , , , . . ,
, , , , . . ,
, , , . . ,
, , , , , , , , , . . ,
. , . . , . , , , ,
. . , . , , ,
. . , . , , , , ,
. . , . , , , , ,
, in press ( doi : 10.1002/hbm.21513 ) .
, , , , , , , . . ,
, , , . . ,
, , , . . ,
, , , , . . ,
, , , , , , . . ,
, , , , , , , , . . ,
, , , , . .
, , , , . . ,
, , , , , , , , , . . ,
, , , , , , , , , , . . ,
, , , . . ,
, , , , , , , . , .
, , , , , , , , , , et al . , . | a great improvement to the insight on brain function that we can get from fmri data can come from effective connectivity analysis , in which the flow of information between even remote brain regions is inferred by the parameters of a predictive dynamical model . as opposed to biologically inspired models , some techniques as granger causality ( gc ) are purely data - driven and rely on statistical prediction and temporal precedence .
while powerful and widely applicable , this approach could suffer from two main limitations when applied to bold fmri data : confounding effect of hemodynamic response function ( hrf ) and conditioning to a large number of variables in presence of short time series .
for task - related fmri , neural population dynamics can be captured by modeling signal dynamics with explicit exogenous inputs ; for resting - state fmri on the other hand , the absence of explicit inputs makes this task more difficult , unless relying on some specific prior physiological hypothesis . in order to overcome these issues and to allow a more general approach , here we present a simple and novel blind - deconvolution technique for bold - fmri signal . in a recent study
it has been proposed that relevant information in resting - state fmri can be obtained by inspecting the discrete events resulting in relatively large amplitude bold signal peaks .
following this idea , we consider resting fmri as spontaneous event - related , we individuate point processes corresponding to signal fluctuations with a given signature , extract a region - specific hrf and use it in deconvolution , after following an alignment procedure .
coming to the second limitation , a fully multivariate conditioning with short and noisy data leads to computational problems due to overfitting .
furthermore , conceptual issues arise in presence of redundancy .
we thus apply partial conditioning to a limited subset of variables in the framework of information theory , as recently proposed . mixing these two improvements
we compare the differences between bold and deconvolved bold level effective networks and draw some conclusions .
bold signal , deconvolution , effective connectivity , granger causality |
gamma - ray emission from solar flares is generally believed to occur predominantly in compact high - density regions near the photospheric footpoints of magnetic field lines .
observations of gamma - ray emission from flares whose host active regions are located behind the visible solar disk pose interesting questions regarding the acceleration sites and mechanism , the transport and interaction points of the accelerated particles during these rare events .
three behind - the - limb ( btl ) flares with emission up to 100 mev were observed during solar cycles 21 and 22 . the first occulted solar flare
whose active region ( ar ) was estimated to be 15@xmath2 behind the western limb was observed on 1989 september 29 by the gamma - ray spectrometer ( grs ) on - board the _ solar maximum mission ( smm)_. @xcite reported intense gamma - ray line emission in the 18 mev range and a strong 2.223 mev neutron capture line from this flare . given the strength of the line emission it was concluded that a spatially extended component was required in order to explain the observations .
the second , detected by phebus on _ granat _
@xcite on 1991 june 1 , had intense gamma - ray line emission in the 18 mev range but no neutron capture line , indicating that the emission was of coronal origin .
the third occulted flare , detected by phebus , batse and egret on 1991 june 30 , was an electron - dominated flare with no detectable line emission but with significant emission up to almost 100 mev .
report that the spectral properties of this flare were similar to those of the flares occurring on the visible disk .
although there was some speculation and some scenarios were put forth @xcite , no definite explanations were found on how and where the particles responsible for these emissions were accelerated and where the gamma - rays were produced .
hard x - ray ( hxr ) emissions from the loop - top of flares originating from ars located just beyond the limb is often observed by _ rhessi _
these are referred to as partially occulted flares since only the emission from the loop footpoints are occulted .
an unusual such flare was reported by @xcite where the flare originated 40@xmath2 behind the limb .
this is very similar to the 2014 september 1 flare discussed in this paper .
the _ fermi_-large area telescope ( lat ) @xcite observations have doubled the number of occulted flares detected and provided the first detections of emission in the gev range from such rare events .
fermi_-lat observations sample flares from active regions originating from behind both the eastern and western limbs , are all associated with very fast coronal massive ejections ( cmes ) and strong solar energetic particle ( sep ) events . in this paper
we present the observations of the first three btl flares detected by _
fermi_-lat together with those from complementary instruments such as _ stereo
_ , _ rhessi _ , konus-_wind _ @xcite ,
_ fermi_-gbm ( gbm ; * ? ? ?
* ) as well as solar observatories in the radio solar telescope network ( rstn ) .
in [ sec : btf ] we present the multiwavelength features including the light curves and localization of the various emissions .
then in 3 we describe a more detailed analysis of the spectral data and compare the properties of the btl flares with three on - disk flares observed by the _ fermi_-lat . in
4 we discuss the associated sep observations and in 5 we present a brief summary and interpretation of these results .
we have collected all relevant observations of the three btl flares detected by the _ fermi_-lat . in what follows we present these observations .
a first analysis of the sol2013 - 10 - 11 flare based on pass7_rep data was presented in @xcite .
we have re - analyzed all three of the flares using pass 8 data .
the pass 8 data benefit from an improved point - spread function , effective area , and energy reach . _
sol2013 - 10 - 11 : _ on 2013 october 11 ( oct13 ) between 07:05:51 and 07:10:51 ut _
stereo_-b detected a solar flare with an ar located at n21e103 .
_ goes_detected an m1.5 class flare starting at around the same time as _ stereo_-b . however , based on the _
stereo_-b 195 emission and the method described in @xcite we estimate that the _ goes_class of this flare if the active region had not been occulted would have been m4.9 with an uncertainty within a factor of three .
this method utilizes the pre - flare background subtracted , full - disk integrated euv intensity , as shown in figure [ fig : sol20131011_lc]b .
a fast cme was observed by lasco with a reported first appearance by the white light coronagraph c2 ( imaging from 2 - 6 solar radii ) at 07:24:10 ut and a linear speed of 1200 km s@xmath3 . _
fermi_coverage started at 07:08:00 ut and continued for more than 30 minutes .
fermi_-lat detected @xmath4100 mev emission for @xmath530 min with the maximum of the flux occurring between 07:20:0007:25:00 ut .
the _ fermi_-gbm detection of hxrs began a few minutes before the _
fermi_-lat and peaked earlier ( @xmath507:10 above 50 kev ) .
_ rhessi_coverage was from 07:08:0007:16:40 ut , overlapping with _
fermi_for 9 min .
konus-_wind _ , located at lagrangian point l1 , detected emission in the 20 - 78 kev and 78 - 310 kev energy bands simultaneously with _
rhessi_and _ fermi_-gbm .
this flare was also observed in radio by the radio solar telescope network ( rstn ) and the nobeyama radio polarimeter @xcite at frequencies up to 9 ghz ( see section [ sec : radio ] ) .
the light curves ( lcs ) for the _ goes _ , _
stereo_-b , _ rhessi _ , _
fermi_-gbm , konus-_wind_and _ fermi_-lat are shown in figure [ fig : sol20131011_lc ] .
the microwave ( mw ) emission is compared with the hxr light curve from konus-_wind_in figure [ fig : konusandradiosol2013 - 10 - 11 ] .
+ _ sol2014 - 01 - 06 : _ on 2014 january 6 ( jan14 ) between 07:35:46 and 07:45:46 ut a solar flare erupted from an ar located at s8w110 . both _
stereo_spacecrafts had a full view of the ar and detected a large filament eruption from the ar at approximately 07:50:00 ut .
the tip of this filament was seen from the visible solar disk by _
there is a hint of detection by _ goes _ of a smaller - than - c - class flare .
however , the peak rate of 2.5@xmath610@xmath7 photons s@xmath3 detected by _ stereo_-b in its 195 channel indicates that the flare would have been classified as a _ goes _
x3.5 if it had not been occulted @xcite .
lasco detected a halo cme with a first c2 appearance at 08:00:05 ut with a linear speed of 1400 km s@xmath3 .
upon exiting the south atlantic anomaly ( saa ) at 07:55:00 ut both instruments on - board _ fermi_detected emission associated with this flare .
fermi_-lat detected @xmath4100 mev emission for approximately 20 minutes ( with no evidence of temporally extended emission after 08:15:00 ut . ) , and the _ fermi_-gbm detected emission in the tens of kev range .
_ rhessi_detected emission starting at @xmath508:18:00 ut ( upon exiting spacecraft night ) also in the tens of kev energy range for over 40 minutes .
konus-_wind_had a full view of the flare and detected emission only in the softest energy band , 2078 kev , starting at 07:43:00 ut .
radio data above 1.3 ghz from rstn for this flare did not indicate a detection .
the _ goes _ , _ rhessi _ , konus-_wind _ , _ fermi_-gbm and _ fermi_-lat light curves for this flare are shown in figure [ fig : sol20140106_lc ] .
.,scaledwidth=48.0% ] _ sol2014 - 09 - 01 : _ on 2014 september 1 ( sep14 ) between 10:55:56 and 11:00:56 ut a bright solar flare occurred in an ar located at n14e126 .
there was no _ goes_signal but _
stereo_-b had an unblocked view of the entire flare and detected a maximum rate of 1.7@xmath610@xmath8 photons s@xmath3 in its 195 channel , indicating an un - occulted _
goes_x2.4 class @xcite .
lasco detected a halo cme with first c2 appearance at 11:12:05 ut with a linear speed of 1900 km s@xmath3 .
a type ii radio burst with an estimated velocity of 2079 km s@xmath3 was reported by noaa space weather alerts in association with this flare .
_ sdo_/aia reported a coronal wave from this ar starting during 10:45:3512:21:35 ut .
this wave was seen to propagate along the limb and over onto the visible disk .
11:3512:20 ut . the increase in rate seen in _ rhessi_and _
fermi_-gbm at 11:38 ut is due to a small on - disk flare.,scaledwidth=50.0% ] the _ fermi_-lat detected emission from this flare for @xmath52 hours , peaking between 11:10:0011:15:00 ut .
the gbm detected emission up to a few mev in temporal coincidence with the _ fermi_-lat emission in both the bgo and nai detectors .
_ rhessi_was in the saa from 10:55:00 to 11:11:00 ; upon exiting the saa it detected emission up to 30 kev .
konus-_wind_detected emission in all three energy bands : 2078 kev , 78310 kev and 3101180 kev in temporal coincidence with _
a significant radio flux at frequencies up to 16 ghz was detected by the san vito station of the rstn simultaneously with the hxr emission peak detected by both _
fermi_-gbm and konus-_wind_. figure [ fig : sol20140901lightcurve ] shows the light curves from goes , _ stereo _ , _ rhessi _ , _
fermi_-gbm , konus-_wind_and _ fermi_-lat . the light curves of the microwave emission ( mw ) are compared with the hxr light curve from konus-_wind_in figure [ fig : konusandradiosol2014 - 09 - 01 ] .
the mw intensity distribution with frequency has a maximum at about 1 ghz ( see section [ sec : radio ] ) .
thus we show the time profile of only this one frequency for this flare . as evident from figure [ fig : konusandradiosol2014 - 09 - 01 ]
there is good agreement between these light curves .
+ we present composite images of the sun as seen by _
sdo_and , whenever available , the position of hxr emission based on _
rhessi_and @xmath0 mev gamma - ray emission centroid based on _
fermi_-lat data .
we used the fits world coordinate system software package @xcite to co - register the locations of the flares between _ stereo _ and _ sdo_images .
we applied the clean imaging algorithm @xcite to _ rhessi_data using the detectors 39 to reconstruct the x - ray images .
the centroid for the @xmath4100 mev gamma - ray emission is determined using the ` gtfindsrc ` tool , which performs a likelihood analysis of the average position in the time integrated data set .
figure [ fig : sol20131011loc ] displays the composite image for the oct13 flare showing the _ stereo_-b 195 image of the ar located about 10@xmath2 behind the limb , the _
sdo_/aia 193 emission peaking above the limb , a contour of the _ rhessi_image and results from re - analysis of the flare with pass 8 data .
the new emission centroid is located in heliocentric coordinates [ -880@xmath9 , 290@xmath9 ] , about 200@xmath9 closer to the _
rhessi_centroid , and with a 68% error radius of 190@xmath9 , which is @xmath520% smaller than the value we reported in @xcite using pass7_rep data .
in addition , in the pass 8 data set , the total number of @xmath41 gev events measured from this flare increased from 4 to 7 .
the highest - energy photon detected from this flare was 3.4 gev and the arrival was 07:19:00 ut . for the jan14 flare the _
fermi_-lat photon statistics were not sufficient to provide an emission localization error circle smaller than 0.5@xmath2 . however , we can still conclude that the emission detected by the _ fermi_-lat was consistent with the position of the sun . in figure [
fig : sol20140106loc ] we show the _ stereo_-a and _ sdo_images of this event at two different times .
the top panels of figure [ fig : sol20140106loc ] present _
sdo_171 ( left ) and _ stereo_-a 195 ( right ) at 07:55:46 ut and show the filament eruption , while the bottom panels show the _
sdo_193 ( left ) and _ stereo_-a 195 ( right ) at 08:25:46 ut with the _ rhessi_6 - 12 and 25 - 50 kev contours of this flare .
fermi_-lat @xmath4100 mev emission centroid of sep14 is located at heliocentric coordinates [ -720@xmath11,610@xmath11 ] with a 68% error radius of 100@xmath11 .
_ rhessi_imaging shows a 612 kev source located above the visible limb slightly offset from the _ fermi_-lat centroid , both shown in figure [ fig : sol20140901loc ] . if the _ rhessi_source is the loop - top of the behind - the - limb flare then the minimum height needed for this source to be visible from @xmath540@xmath13 behind - the - limb would be @xmath510@xmath14 cm .
hxr loop - top emission from a flare located @xmath540@xmath13 behind the limb has been detected before by _
_ @xcite .
fermi_-lat measured 17 photons with energies @xmath41 gev ; 15 of these ( including a 3.5 gev photon with an arrival time of 11:16:01 ut ) arrived during the first 20 minutes of
_ fermi_-lat detection .
we performed an unbinned likelihood analysis of the _ fermi_-lat data with the ` gtlike ` program distributed with the _ fermi_`sciencetools ` .
we selected pass 8 source class events from a 10@xmath2 circular region centered on the sun and within 100from the local zenith ( to reduce contamination from the earth limb ) .
we fit three models to the _ fermi_-lat gamma - ray spectral data .
the first two , a pure power law ( pl ) and a power - law with an exponential cut - off ( plexp ) are phenomenological functions that may describe bremsstrahlung emission from relativistic electrons .
the third model uses templates based on a detailed study of the gamma rays produced from decay of pions originating from accelerated protons with an isotropic pitch angle distribution in a thick - target model ( updated from * ? ? ?
we rely on the likelihood ratio test and the associated test statistic ts @xcite to estimate the significance of the detection .
the ts is defined as twice the increment of the logarithm of the likelihood obtained by fitting the data with the source and background model components simultaneously . because the null hypothesis ( i.e. the model without an additional source ) is the same for the pl and plexp models , the increment of the ts ( @xmath15ts = ts@xmath16-ts@xmath17 ) is equivalent to the corresponding difference of maximum likelihoods computed between the two models .
note that the significance in @xmath18 can be roughly approximated as @xmath19 for 2 degrees of freedom .
ccccccc[ht ] + 07:10:0007:15:00 & 20 & 3 & @[email protected] & - & [email protected] & - + 07:15:0007:18:00 & 487 & 43 & @[email protected] & 130@xmath2130 & 17@xmath211 & [email protected] + 07:18:0007:20:00 & 846 & 77 & @[email protected] & 112@xmath2121 & 49@xmath212 & [email protected] + 07:20:0007:22:00 & 776 & 76 & @[email protected] & 128@xmath2123 & 46@xmath212 & [email protected] + 07:22:0007:24:00 & 677 & 58 & @[email protected] & 153@xmath2131 & 29@xmath212 & [email protected] + 07:24:0007:26:00 & 345 & 25 & @[email protected] & 208@xmath2161 & 21@xmath212 & [email protected] + 07:26:0007:28:00 & 219 & 34 & [email protected] & 86@xmath2122 & 16@xmath211 & [email protected] + 07:28:0007:40:00 & 283 & 29 & [email protected] & 86@xmath2126 & 7@xmath211 & [email protected] + + + 07:55 08:15 & 67 & 20 & @[email protected] & - & [email protected] & - + + + + 11:02:0011:04:00 & 42 & 9 & @[email protected] & - & 11@xmath213 & - + 11:04:0011:06:00 & 321 & 41 & @[email protected] & 117@xmath2127 & 117@xmath2110 & [email protected] + 11:06:0011:08:00 & 1070 & 120 & @[email protected] & 116@xmath2115 & 360@xmath2117 & [email protected] + 11:08:0011:10:00 & 1549 & 126 & @[email protected] & 158@xmath2120 & 477@xmath2119 & [email protected] + 11:10:0011:12:00 & 2549 & 115 & @[email protected] & 194@xmath2128 & 565@xmath2121 & [email protected] + 11:12:0011:14:00 & 6394 & 157 & @[email protected] & 167@xmath2110 & 522@xmath2120 & [email protected] + 11:14:0011:16:00 & 2430 & 159 & @[email protected] & 148@xmath2117 & 540@xmath2122 & [email protected] + 11:16:0011:18:00 & 1069 & 32 & @[email protected] & 488@xmath2190 & 465@xmath2123 & [email protected] + 11:18:0011:20:00 & 1047 & 49 & @[email protected] & 182@xmath2137 & 396@xmath2127 & [email protected] + 12:29:0012:34:00 & 56 & 12 & @[email protected] & - & 3@xmath211 & - + 12:34:0012:39:00 & 133 & 3 & @[email protected] & - & 3@xmath211 & - + 12:39:0012:44:00 & 130 & 5 & @[email protected] & - & 2@xmath211 & - + 12:44:0012:49:00 & 97 & 11 & @[email protected] & - & 2@xmath211 & - + 12:49:0012:54:00 & 84 & 3 & @[email protected] & - & 2@xmath211 & - + [ tab : spectralanalysis ] in table [ tab : spectralanalysis ] we list the ts@xmath17 , @xmath22 , @xmath23 the photon index for the best - fit model ( pl when @xmath24 or plexp when @xmath25 ) and plexp cut - off energy . for several intervals @xmath2625 , indicating that plexp provides a significantly better fit than pl . for these intervals
we fit the pion - decay models to the data to determine the best proton spectral index following the same procedure described in @xcite .
in particular , we performed a series of fits with the pion - decay template models calculated for a range of proton spectral indices .
we then fit the resulting profile of the log - likelihood with a parabolic function of the proton index .
the minimum gives the most likely index for the pion - decay model .
note that the ts values for plexp and pion - decay fits can not be directly compared @xcite because they are not nested models .
however the plexp approximates the shape of the pion - decay spectrum ; thus we expect the pion - decay models to provide a similarly acceptable fit .
the main contributions to the systematic uncertainties are uncertainties in the effective area ; in the energy range 100 mev to @xmath5100 gev these are of the order of @xmath215%.this uncertainty applies directly to the flux values and from our previous studies on lat - detected solar flares @xcite we find that the systematic uncertainties on the cut - off energy and photon index are also of the order of @xmath215% .
ccccccccc[ht ] 2013 - 10 - 11 & m4.9 & 1200 & n21e103 & 30 & 49@xmath212 & 1.5@xmath610@xmath27 & [email protected] & 9.8@xmath610@xmath28 + 2014 - 01 - 06 & x3.5 & 1400 & s8w110 & 20 & [email protected] & 4.2@xmath610@xmath29 & [email protected]@xmath30 & 3.5@xmath610@xmath27 + 2014 - 09 - 01 & x2.4 & 2000 & n14e126 & 113 & 565@xmath2114 & 1.4@xmath610@xmath28 & [email protected] & 7.0@xmath610@xmath31 + + 2011 - 03 - 07 & m3.7 & 2125 & n30w48 & 798 & 3@xmath211 & 5.1@xmath610@xmath27 & [email protected] & 3.6@xmath610@xmath31 + 2011 - 06 - 07 & m2.5 & 1255 & s22w53 & 38 & 3@xmath211 & 3.2@xmath610@xmath32 & 5.0 @xmath210.3 & 2.5@xmath610@xmath28 + 2012 - 03 - 07i@xmath33 & x5.4 & 2684 & n16e30 & 45 & 417@xmath2113 & 3.9@xmath610@xmath28 & 3.90@xmath21 0.02 & 2.1@xmath610@xmath34 + 2012 - 03 - 07e@xmath35 & x5.4 & 2684 & n16e30 & 1068 & 97@xmath212 & 1.4@xmath610@xmath31 & [email protected] & 9.0@xmath610@xmath34 + [ tab : btlquantities ] lc[ht ] + + + power - law ( 1 ) index & [email protected] + proton index & [email protected] + + power - law ( 1 ) index & [email protected] + power - law ( 2 ) index & @[email protected] + cutoff energy ( mev ) & [email protected] + proton index & [email protected] + + + + + power - law ( 1 ) index & [email protected] + cutoff energy ( mev ) & 90@xmath217 + proton index & [email protected] + + power - law ( 1 ) index & [email protected] + power - law ( 2 ) index & [email protected] + cutoff energy ( mev ) & [email protected] + proton index & [email protected] + [ tab : spectralfits ] we compare the characteristics of the @xmath0 mev emission associated with the three btl flares with three disk flares with similar _ goes_classifications and temporally extended emissions and are described in @xcite and @xcite .
in addition to spectral parameters we also compare the total @xmath0 mev emitted energies and the total energy released by protons with energy @xmath36 mev needed to produce the detected gamma - ray emission , as based on the templates of ( updated from * ? ? ?
we present these numbers along with the date , estimated _
goes_class , cme speed , ar position and the _ fermi_-lat detection duration in table [ tab : btlquantities ] .
the proton indexes are very similar whereas the on - disk flares appear to have more energy ; this is most likely because we observed the on - disk flares over longer time scales .
peak fluxes and the total energy released by protons with e@xmath4500 mev for sep14 and the impulsive phase of sol2012 - 03 - 07 are comparable .
figure [ fig : ondisk_btl_protonindex ] shows the proton index as a function of time for the oct13 and sep14 btl flares .
we fit a constant ( red dashed line ) and a first - degree polynomial ( blue dashed line ) to the data . in the case of the constant fit we list the best - fit value in the upper left corner , whereas for the straight - line fit we list the value of the slope .
the temporal variation over tens of minutes is not sufficient to conclude whether a softening or hardening is present for these btl flares . in @xcite
we found that for the on - disk flare sol2012 - 03 - 07 the spectrum softened with a time scale of a few hours .
2.5@xmath18 ) .
the parameters of the fits are listed in table [ tab : spectralfits].,scaledwidth=49.0% ] 10 mev and a pion - decay model to describe the _ fermi_-lat spectrum .
the neutron capture line ( at 2.223 mev , highlighted by the red vertical dashed line ) is not statistically significant ( @xmath52@xmath18 ) and neither is an additional power law at low energy @xmath52@xmath18 .
the fit parameters are listed in table [ tab : spectralfits].,scaledwidth=49.0% ] as described above we have hxr data from three instruments : _ rhessi _ , _ fermi_-gbm and konus-_wind_. konus-_wind_was in waiting mode during the jan14 and sep14 flares ; therefore these events were detected in only three broad energy channels , namely @xmath52078 , 78310 and 3101180 kev with 2.944 s time resolution .
_ rhessi_provides only a limited coverage of these flares . for the oct13 flare
there is some overlap with _
fermi_-gbm during the rise of the impulsive phase . as shown in @xcite the hxr spectra of _ rhessi_and
fermi_-gbm agree well .
_ fermi_-gbm was in the saa and _ rhessi_was in spacecraft night during the impulsive phase of the jan14 flare and detected the flare during the decay phase in only two low - energy channels .
thus , we do not have sufficient data for a spectral fit .
for the sep14 flare _ rhessi_gives information only on the decay phase of the flares . in the following
we show results from analysis of _
fermi_-gbm data for the oct13 and sep14 flares . for both oct13 and sep14 we performed a combined _ fermi_-gbm / lat fit using the ` xspec ` package @xcite .
the spectral fits were done by minimizing the ` pgstat ` . `
pgstat`is a profile likelihood statistic that takes into account poisson error on the total count spectrum and gaussian error on the background . in order to obtain the background - subtracted spectra of the _ fermi_-gbm data for sep14 we used both _ fermi_-gbm - nai and _ fermi_-gbm - bgo spectra accumulated before the flare ( from 10:54 ut to 10:57 ut ) , and after the flare ( from 11:42 - 11:50 ut ) . for oct13 , because there was a minor on - disk flare whose onset time was 7:01 ut , we used the procedure described in @xcite , which consists of using the background estimation tool developed in @xcite with an additional 5% systematic error . for the lat
, we follow the procedure used in @xcite and @xcite that consists of deriving the background spectrum directly from the model of the background used in the standard lat likelihood analysis ( using first ` gtlike ` and then ` gtbkg ` ) , and obtaining the response of the lat using ` gtrspgen ` ( all of these tools are available in the _ fermi_`sciencetools ` ) .
the combined spectral energy distribution ( sed ) derived from the _ fermi_-gbm and _ fermi_-lat data extending from 30 kev to 10 gev for these flares is shown in figures [ fig : sedsol20131011 ] and [ fig : sedsol20140901 ] , respectively , and the parameters for the spectral fit for both solar flares are listed in table [ tab : spectralfits ] .
the oct13 flare had a very weak signal in the bgo , and the best - fit model ( m@xmath37 ) consisted of a single power law and the pion - decay templates to describe the bremsstrahlung and @xmath460 mev emission detected by the _
fermi_-gbm and _ fermi_-lat , respectively . for the sep14 flare ,
the best - fit model ( m@xmath37 ) consisted of a single power law with an exponential cut off at high energies to describe the bremsstrahlung emission , and the pion - decay templates to describe the @xmath460 mev emission detected by the _
fermi_-lat , similar to what was done for the 12 june 2010 _
fermi_flare @xcite .
we also tested the statistical significance of an extra power law with a high - energy cut off ( model m@xmath38 ) . to this end
we performed monte carlo simulations ( using xspec ` fakeit ` ) by generating a spectrum described by the best fit model ( m@xmath37 ) .
we then re - optimized the parameters of m@xmath37 by fitting it to the simulated data .
we finally compared the improvement of ` pgstat`when fitting the simulated data with the model m@xmath38 . the improvement of ` pgstat`for the oct13 and sep14 flares is @xmath39408 and @xmath40109 , respectively . despite these high values of @xmath41 , dedicated monte carlo simulations
indicate that they correspond to significances of 2.5 and 2.0 @xmath18 , respectively .
we then checked whether the addition of the 2.223 mev line ( on top of model m@xmath37 ) significantly improved the fit . for the oct13 flare
the improvement is negligible ( @xmath41@xmath400 ) , while for the sep14 the improvement is ( @xmath41@xmath4025 ) , corresponding to a significance of 2.0 @xmath18 , estimated from monte carlo simulations .
we tested whether adding nuclear de - excitation narrow lines and continua provided an improvement to the fit for both flares and found that it was not significant .
the nuclear line and continua templates used here are based on a detailed study of the nuclear gamma - ray production from accelerated ion interactions with elements found in the solar atmosphere @xcite .
, found by fitting the optically thin part of the spectra for each flare , is also shown.,scaledwidth=46.0% ] the radio spectra obtained at the peaks of the oct13 and sep14 events are shown in figure [ fig : radiospecoct13 ] .
the spectrum was obtained by using the rstn 1-s resolution fluxes .
the spectra were background subtracted and the fluxes were integrated within the selected time intervals for each frequency .
for each spectrum we detected the frequency where the maximum flux was observed .
the spectra of the oct13 and sep14 flares are similar to those expected from the gyro - synchrotron ( gs ) mechanism , with peaks that separate the optically thin and thick parts .
the values for the microwave photon spectral index , @xmath42 , for both flares are shown in figure [ fig : radiospecoct13 ] .
unfortunately , the rstn data for the time period of the jan14 flare are not sufficient to obtain radio spectra .
for the goes sep protons associated with jan14 .
the y - axis gives the time in seconds since 2014 - 01 - 06 07:40 ut ( assumed start of the flare as detected by _
stereo_-b ) . assuming scatter - free propagation for the first arriving particles .
the red dashed line represents the fit using equation [ eq : seponsetimes ] , obtaining the pathlength @xmath43 au and @xmath4407:55@xmath2100:05 ut.,scaledwidth=48.0% ] interpretation of the origin of seps is more complicated than studying photons , since they move with different velocities along curved magnetic field lines and are detected only when the field lines are connected to the instrument .
moreover , they may be scattered by turbulence with energy - dependent mean free path .
goes_proton fluxes did not show a significant increase for the oct13 event ( which occurred behind the eastern limb ) .
however , _
stereo_-b , positioned behind the limb and with better magnetic connectivity , detected an increase of sep proton intensity starting roughly 2 hours after the detection of an electromagnetic signal from this flare . whereas the sep14 flare , which was located roughly 40@xmath2 behind the eastern limb ,
was associated with an increase in the proton fluxes detected both by _ stereo_-b ( which had a front view of the flare ) and _ goes_(starting roughly 9 hours after the flare ) .
the jan14 flare , originating in an ar 20@xmath2 behind the western limb , produced no significant increase in the proton flux at _ stereo _ , had better magnetic connectivity to the earth and was associated with a very strong sep event with neutrons detected on ground by the south pole neutron monitors .
the arrival time @xmath45 @xcite at the earth of a particle with velocity @xmath46 released from the sun at time @xmath47 , is : @xmath48 where @xmath49 is the distance traveled by the sep from the acceleration region to the earth . assuming scatter - free propagation for the first - arriving particles
, @xmath49 will be independent of particle energy or velocity and will be equal to the path length along a parker spiral magnetic field line from the sun to the earth ( usually assumed to be 1.2 au ) .
if all the particles are accelerated at the same place and time , then the intercept of the line fitting @xmath45 to @xmath50 would give @xmath47 , the time at which seps are released and is known as the solar particle release ( spr ) time .
the slope of this line , @xmath51 will be of the order of 600 s. since we do not have high energy resolution sep data we obtain the sep onset times using two methods depending on the energy range of the particles . for the goes
seps with energies @xmath52200 mev we apply a median filter to the intensity profile and define the onset time as when the profile reached 5% of the maximum . for the higher - energy seps we apply a fast fourier transform ( fft ) filter to the intensity profile and evaluate the time when the second derivative is maximum as the onset time . to estimate of the uncertainty on the onset time we performed a scan over a series of values for the median and fft filter windows and take the difference in onset times as the error .
the onset times as a function of @xmath53 for the _ goes_sep protons associated with jan14 are shown in figure [ fig : sol20140106onsettimes ] .
fitting the data to equation [ eq : seponsetimes ] , we estimate the spr time to be @xmath5407:55 @xmath21 0:05 ut and path length of @xmath55 a.u .
the value for @xmath47 is consistent with the estimated spr time , 07:47 ut , reported by @xcite but is somewhat later than the start of the flare activity at the sun . since _
fermi_was in the saa and _ rhessi_was in orbital night , the most accurate time of the initiation of the flare was obtained from konus-_wind_and _ stereo_data .
konus-_wind_detected emission in the 2178 kev energy range starting at 07:43 ut , implying an emission time at the sun of 07:35 ut , roughly 12 minutes prior to the spr time . assuming that the cme was released at this latter time the acceleration of seps started when the cme was at a height of 1.5 @xmath561400 km s@xmath57 .
unfortunately the high - energy sep data were not sufficient to estimate the spr time for the other two btl flares occurred on oct13 and sep14 .
we have presented the analysis of _
fermi_-lat data from three btl solar flares , whose location is determined by _ stereo_observations .
we also presented analyses of data from _ rhessi _ , _
fermi_-gbm , konus-_wind_in the hxr range , _
sdo_in euv , rstn in radio wavelengths , some sep observations , goes and south pole neutron monitors .
we found several results : 1 .
the hxr lcs measured by the three instruments ( in overlapping energy ranges ) are in good agreement with each other and with the high - frequency radio lcs .
the _ fermi_-lat emission , in general , commences several minutes after the hxrs , peaks at a later time and lasts longer .
we are able to obtain _
rhessi_images in soft hxrs for flares just over the limb , indicating a loop - top emission from loops with heights above the photosphere ranging from @xmath58 to @xmath59 cm .
_ sdo_detected uv emission is detected for oct13 , and possibly jan14 , but not the sep14 flare .
jan14 flare was too weak to permit localization of the gamma - ray source but localizations of the _ fermi_-lat emission with pass 8 data for the other two flares situates the centroid locations near the solar limb and about 50@xmath9 ( oct13 ) and 300@xmath9 ( sep14 ) from the _ rhessi_centroid .
the lcs ( available only for oct13 and sep14 flares ) show good agreement between the hxr spectra measured by _
fermi_-gbm , _
rhessi_(whenever contemporaneous data are available ) and konus-_wind_below a few 100 kev .
fermi_-lat spectra can be fitted with a relatively hard plexp phenomenological model ( possibly due to relativistic electron bremsstrahlung ) or to a pion decay model with proton indexes around 4 .
we note that production of photons of about @xmath53 gev detected here requires @xmath60 gev protons or @xmath61 gev electrons .
these energies are greater than any other reported for solar particles .
we compare several spectral properties of @xmath0 mev radiation from the btl flares with those of three on - disk flares previously analyzed . as evident from table [ tab : btlquantities ] and figure [ fig : ondisk_btl_protonindex ] , in general , these flares have similar characteristics . similar _
goes_class flares have similar peak photon fluxes , surprisingly , regardless of limb occultation . the higher total gamma - ray energy of on - disk flares is most likely results from their having been observed over longer timescales .
thus , the underlying acceleration and emission processes are most likely similar , but the transport paths of the radiating particles ( presumably protons ) from the acceleration to the emission site must be different .
that is , for btl flares , protons must travel greater distances to land on the front side of the sun to produce the detected gamma - rays . 5 .
we find that the proton index does not vary significantly during the 30-minute observation time windows for the btl flares .
this is similar to that found for the on - disk flares for the same time frames , but some on - disk flares are detected over longer time intervals ( up to 20 hours ) during which the proton index increases gradually . since the spectral information available for the jan14 flare
is limited we can not constrain the acceleration , transport or radiation processes .
however this flare is remarkable because the ar being located beyond the western limb provided good magnetic connection with the earth and as a result it was associated with a very strong sep event and increased the neutron flux in the south pole neutron monitor .
as presented in figure [ fig : sol20140106onsettimes ] , the acceleration time at the sun for the sep protons is 07:55 @xmath21 0:05 ut which is in agreement with the spr time of 07:47 @xmath21 0:08 ut reported by @xcite . unfortunately , the _ fermi_-lat was in the saa at the spr time and exited at 07:55 ut
. therefore we can only provide an upper limit on the time of acceleration of gamma - ray producing particles ( presumably protons ) at the sun .
however , konus-_wind_detected emission in the 2078 kev band starting at 07:43 ut implying that the electrons were accelerated at the sun at 07:35 ut , roughly 12 minutes prior to the spr .
due to the location of the ar the other two flares were not magnetically well connected to the earth . for the sep14 flare the _
goes_proton fluxes begin to increase roughly 9 hours after the flare start time .
_ stereo_-b observed the flare in its entirety but provides proton fluxes only in three energy bands , which is not sufficient to estimate the acceleration time at the sun as we did for jan14 .
thus , we do not have reliable sep data to test whether the onset times of the sep and gamma - rays coincide , however we can compare the height of the cme at the time of the gamma - ray onset with the cme height at spr found for jan14 . based on the 2@xmath62 order fit of height as a function of time ( from the lasco cme online catalog ) we estimated that the cme was at @xmath52.5at the onset time of the _ fermi_-lat emission .
we found a similar value for oct13 .
these heights are significantly smaller than the predicted values reported in @xcite for the cme heights at spr as a function of source longitudes from solar cycle-23 . to test whether these cme heights are compatible with the gamma - ray emission detected from the btl flares
, it would require detailed simulations on particle transport in the solar magnetic field .
this study is beyond the scope of this paper and will be addressed in a future work .
the _ rhessi_images show that the 20 to 50 kev hxrs , due to electron bremsstrahlung , are emitted from @xmath4030@xmath9 ( @xmath63 cm ) and @xmath4050@xmath9 ( @xmath64 cm ) sources from top of loops extending to heights of @xmath65 and @xmath66 cm above the photosphere for oct13 and sep14 flares , respectively .
the column depth required for a particle of mass @xmath67 and kinetic energy @xmath68 ( in units of @xmath69 ) , or lorentz factor @xmath70 , to lose all its energy via coulomb interactions is @xmath71 where @xmath72 is the classical electron radius and @xmath73 is the coulomb logarithm . from equation [ eq : couldepth ]
we find that the column depth required to stop the @xmath74 kev electrons that produce the hxrs detected by _
rhessi_is @xmath75@xmath76 . assuming a density of @xmath77 @xmath78 , the sizes of the loop - top sources imply a column depth of @xmath79 and @xmath80 @xmath76 for the two hxr sources , respectively .
this means that the energy loss time @xmath81 is more than 10 times longer than the crossing time @xmath82 .
these values , together with the results reported in @xcite suggest that the hxr emission from these two btl flares are thin target sources . if these loop - top sources were to be produced in a thick - target , which can come about in two ways under extreme conditions not likely to be the case here .
the first is in the strong diffusion limit , i.e. , if the electrons are trapped because the scattering time , @xmath83 , is smaller than the crossing time , @xmath84 .
note that because @xmath85 , the scattering can not be due to coulomb interactions but it could be due to turbulence .
the second possibility is in the weak diffusion limit @xmath86 , where trapping can occur if the field lines converge strongly . in this case
we need @xmath87 , the log of the ratio of magnetic field strengths at the middle and ends of the magnetic bottle . ]
( with proportionality constant @xmath88 increasing with increasing field convergence @xcite ) .
these are somewhat extreme conditions requiring either a very high density ( therefore increasing the scattering , presumably turbulence ) or a somewhat strong field convergence and a clean " loop with low density and low level of turbulence . based on the positions of the _ fermi_-lat @xmath4100 mev emission centroids alone we can not exclude that the gamma rays come from the loop - top source . in order to investigate this possibility we rely on equation [ eq : couldepth ] to find the column depth required to stop @xmath89 mev protons ( that produce @xmath0 mev photons ) and find @xmath90 @xmath76 .
this is much larger than the loop - top column depth @xmath91 @xmath76 so that , in the absence of trapping , we are again dealing with a thin target loop - top source and a small energy loss compared to what would be the case for emission from a thick target photospheric source .
consequently , we would need much larger energies of accelerated protons than those given in table [ tab : btlquantities ] assuming a thick target scenario .
the condition for proton trapping in the loop - top ( or any coronal trap region , for that matter ) is also more extreme than that described above for electrons and we would need escape times @xmath92 times longer than the crossing time . _
thus , the emission detected by the _ fermi_-lat is most likely due to decay of pions produced by energetic protons interacting in a thick - target photospheric source . _
as shown in @xcite and @xcite in a thin target scenario the relation between the hxr spectral index , @xmath93 , and the number index of the energy spectrum of the emitting electrons index , which in non relativistic limit is @xmath94 used commonly . in the relativistic regime
these indexes are equal and as stated in equation [ eq : deltagamma ] @xmath95 .
] through bremsstrahlung processes , @xmath96 , is @xmath97 for optically thin synchrotron radiation , the relation between @xmath96 and the microwave photon spectral index , @xmath42 , can be describe by the following relations for the relativistic ( e.g. * ? ? ?
* ) and semi - relativistic cases we will use these relations in the following for the interpretation of the detected emission from the two brightest behind - the - limb flares reported in this work .
the spectra of the two flares in the 30 kev to 10 mev energy range are somewhat different so we discuss them separately .
_ oct13 : _ as already shown in @xcite , both _
fermi_-gbm detect hxr spectra for oct13 flare that are relatively steep ( best - fit value of power - law photon index @xmath99 ) .
however , as shown in figure [ fig : sedsol20131011 ] there is a tentative indication that the spectrum may harden into a nearly flat part in the 110 mev range requiring either a similar hardening of the spectrum of the accelerated electrons , as e.g. described in @xcite , or a contribution from de - excitation nuclear lines .
we found that the inclusion of a 2.223 mev broad neutron capture line and the nuclear de - excitation lines did not improve the fit significantly and that the best - fit model for the emission observed by _
fermi_-gbm is a single power law . as described in section [ sec : position ] the thin target is most likely , in which case following equation [ eq : deltagamma ] for the non - relativistic case ( @xmath52100 kev ) the index of the electron number density @xmath100 .
the radio spectrum observed by rstn , shown in figure [ fig : radiospecoct13 ] , indicates microwave flux of 60 solar flux units ( @xmath101 jy ) that peaks at frequency @xmath102 ghz and declines as @xmath103 with the index @xmath104 above this frequency .
based on the relations described in equation [ eq : synchr ] we find that the required power - law index for the number density of electrons to be @xmath105 or @xmath106 in the relativistic or semi - relativistic regime , respectively .
these values are steeper than the index @xmath100 required in the non - relativistic hxr range , indicating a steepening of the electron spectrum above 1 mev .
_ we therefore conclude that the gbm flux in the energy range 300 kev - 10 mev is probably not due to the accelerated electrons and no extra component is expected .
_ _ sep14 : _ the best - fit model for the combined hxr _
fermi_-gbm gamma - ray emission from the sep14 flare is a single power law with an exponential decay with photon index [email protected] and a cut - off energy of 90@xmath217 mev .
this is an extremely hard spectrum for a solar flare and if it is produced by a thin - target electron - bremsstrahlung model as described above , it would require an electron number spectrum with index @xmath107 in the non - relativistic range steepening to an index @xmath108 in the relativistic range
. however , as noted above inclusion of the 2.223 mev line improves the fit slightly , indicating a possible contribution from nuclear de - excitation lines so the steepening of the electron spectrum could be larger .
the radio spectrum of this flare , shown in figure [ fig : radiospecoct13 ] is also harder .
the emission detected by the san vito station has an index @xmath109 above 1 ghz .
again assuming optically thin synchrotron emission would require semi - relativistic electrons with electron power - law index @xmath110 indicating a slightly larger steepening at high energies . if the magnetic field is lower , which could be the case because of the larger loop
, the steepening would be even larger ( @xmath111 ) .
_ thus , for the sep14 flare a power - law spectrum , steepening at higher energies , is required to explain the hxr and microwave observations from the loop - top region .
this plus a pion decay component provides an acceptable fit to the fermi data from 30 kev to several gev .
_ the above interpretation of the hxr and microwave emissions indicates that the electrons responsible for these emissions are most likely accelerated in the reconnection region above the loop - top source either by turbulence @xcite or in merging islands manifested in the particle - in - cell simulations @xcite .
the electrons are trapped in the loop - top region long enough ( longer than the crossing time of a fraction of second ) to produce detectable bremsstrahlung and synchrotron radiation , but most of their energy is lost in thick - target footpoints hidden from near earth instruments because of high optical depths @xcite .
this scenario can also explain the hxr emission from a btl flare reported by @xcite .
the steepening of the electron spectrum in the semi - relativistic regime is due to a decrease in the acceleration rate and not to transport or energy loss effects . if the _
fermi_-lat emission is from a thick - target photospheric site , then this emission site is located on the visible part of the disk .
it can not come from the occulted ar .
the fact that accelerated protons reach the on - disk emission site provides strong evidence that the acceleration site is the cme environment , as suggested by @xcite and @xcite
. they would diffuse across a wide range of magnetic fields , some connected to the ar behind the limb and some to the visible side of the sun .
these ions will most likely come from the down - stream region of the cme shock while seps escape the upstream region .
this can account for the differences in their spectral and temporal characteristics .
it is then possible that the discrepancy between the spectral characteristics in 110 mev for the gbm and radio observations can be explained by lower - energy 1100 mev protons or electrons precipitating to the visible disk side from the down - stream region and producing the 110 mev nuclear de - excitation line or bremsstrahlung emission below the chromosphere . these and other possibilities , for example , production of the _ fermi_-lat emission by relativistic electrons either in the cme shock , in reconnection regions on current sheets behind the cme or trapped in a large loop with strong convergence , will be discussed in future papers .
the @xmath112 lat collaboration acknowledges support from a number of agencies and institutes for both development and the operation of the lat as well as scientific data analysis .
these include nasa and doe in the united states , cea / irfu and in2p3/cnrs in france , asi and infn in italy , mext , kek , and jaxa in japan , and the k. a. wallenberg foundation , the swedish research council and the national space board in sweden . additional support from inaf in italy and cnes in france for science analysis during the operations phase is also gratefully acknowledged .
w.l . was supported by nasa hgi grant nnx16af78 g and lws grant nnx14aj49 g .
v.p , w.l and f.r.d.c are supported by nasa grants nnx14ag03 g , nnx13af79 g and nnx12ao70 g .
p and l.k .
thanks the rfbr grant 15 - 02 - 03717 we thank nariaka nitta and meng jin for friutful discussions on seps and magnetic connectivity . , g. , mcbreen , s. , connaughton , v. , & briggs , m. 2012 , in society of photo - optical instrumentation engineers ( spie ) conference series , vol .
8443 , society of photo - optical instrumentation engineers ( spie ) conference series , 3 | we report on the _ fermi_-lat detection of high - energy emission from the behind - the - limb solar flares that occurred on 2013 october 11 , 2014 january 6 and 2014 september 1 .
the _
fermi_-lat observations are associated with flares from active regions originating behind both the eastern and western limbs , as determined by _
stereo_. all three flares are associated with very fast coronal mass ejections ( cmes ) and strong solar energetic particle events .
we present updated localizations of the @xmath0 mev photon emission , hard x - ray ( hxr ) and euv images and broad - band spectra from 10 kev to 10 gev as well as microwave spectra .
we also provide a comparison of the behind - the - limb flares detected by the _
fermi_-lat with three on - disk flares and present a study of some of the significant quantities of these flares as an attempt to better understand the acceleration mechanisms at work during these occulted flares .
we interpret the hxr emission to be due to electron bremsstrahlung from a _ coronal _ thin - target loop - top with the accelerated electron spectra steepening at semirelativistic energies .
the @xmath0 mev gamma - rays are best described by a pion decay model resulting from interaction of protons ( and other ions ) in a thick target photospheric source .
the protons are believed to have been accelerated ( to energies @xmath1 gev ) in the cme environment and precipitate down to the photosphere from the down - stream side of the cme shock and landed on the front side of the sun , away from the original flare site and the hxr emission . |
the united nations organization initiative `` atoms for peace '' presented by usa president dwight d. eisenhower in december 1953 was the first step in peaceful use of nuclear technology . today , many countries of the world develop or begin to create their strong nuclear programs
there are more than 440 nuclear power plants ( npp ) operating in 30 countries , more than 400 ships with nuclear reactors used as propulsion systems .
about 300 research reactors operate in 50 countries .
this kind of reactors produces radioisotopes for medical diagnostics and therapy of cancer , neutron sources for research and training .
approximately 55 nuclear power plants are under construction and 110 ones are planned .
the republic of belarus is a newcomer country in nuclear energy .
there is no experience of the nuclear power plant construction but large scientific potential in the field of atomic and nuclear physics , radiochemistry and radiation chemistry .
hence the development of nuclear knowledge portal is one of the first step for the scenario of nuclear knowledge management .
the objectives of the portal are considered as the preservation and enhancing of nuclear knowledge , assisting all the participants of national nuclear energy system development in accumulation of the international experience and competence needed for the effective and safe use of nuclear energy as well as popularization of nuclear knowledge for schoolchildren and the general public .
since the beginning of the xxi century the international atomic energy agency ( iaea ) pays big attention to the nuclear knowledge management ( nkm ) @xcite@xcite . nuclear knowledge ( nk ) is the base stem of appropriate research and development as well as industrial applications of nuclear technologies and includes energy and non - energy applications .
knowledge management ( km ) @xcite@xcite is an integrated , systematic approach to identifying , acquiring , transforming , developing , disseminating , using , sharing and preserving knowledge , relevant to achieving specified objectives .
basic km concept by iaea is depicted as a pyramid . on its foundation
there is data .
data is presented as unorganized and unprocessed facts , a set of discrete facts about events . over data
there is the information as aggregation of data that makes decision making easier .
knowledge is the highest level of information .
there is wisdom and enlightenment on the top of km pyramid .
approximate percentage of subject area of nuclear knowledge by the iaea is as following : * nuclear physics 11% , * nuclear materials 9% , * engineering and instrumentation 9% , * elementary particle physics 16% , * atomic , molecular and condensed matter physics 10% , * life sciences 18% , * chemistry 4% , * nuclear power and safety 6% , * nuclear fuel cycle and radioactive waste 3% , * fusion research and technology 7% , * environmental and earth sciences 3% , * isotopes 1% , * non - nuclear energy 1% , * economic , legal and social fields 2% .
the strategy of the iaea in nkm is as following : it is extremely important that the educational process involves the enterprises of nuclear industry @xcite .
great attention is paid to the development of national , regional and international educational networks and portals .
the iaea developed detailed recommendations for the creation of portals with the formulation of purposes and principles of their design .
taxonomy is the main km concept @xcite .
taxonomy ( from greek _ taxis _ meaning arrangement or division and _ nomos _ meaning law ) is a hierarchical structure in which a body of information or knowledge is categorized , allowing an understanding of its various parts relations with each other .
taxonomies are used to organize information in systems , thereby helping users to find it .
the stages of taxonomy developing are : * determination of taxonomy requirements ; * identification of its concepts ( where is the content and what do the users think ) ; * developing draft taxonomy ; * its review by users ; * refining taxonomy ; * adaptation of taxonomy to the content ; * taxonomy management and maintenance .
each developed country , forming its own nuclear industry , should independently create , establish and maintain the nuclear knowledge portal , integrated into the global nuclear knowledge management industry .
such portals will allow to manage information resources , knowledge and competencies of the nuclear industry of belarus , as well as to preserve , maintain and develop the knowledge at the level that provides a safe , sustainable and efficient development of belarusian nuclear industry . nowadays in belarus
there are several websites of selected organizations and institutions that are not related to the united portal , providing separate information on the subject far from completeness .
creating of a full - fledged portal of nuclear knowledge is the multistage process . as the first step it is proposed to create educational and research portal of nuclear knowledge .
it will not be a portal of npp , which should be developed separately .
prospective participants of educational and research portal of nuclear knowledge are : ministry of education of the republic of belarus , belarusian state university ( bsu ) , research institute for nuclear problems of bsu , universities , training specialists for nuclear power plant , department for nuclear and radiation safety of the ministry for emergency situations of the republic of belarus ( gosatomnadzor ) , the joint institute for power and nuclear research sosny .
all works should be executed by the monitor of the iaea .
the first proposals for the development of educational and research portals of nuclear knowledge , which in the long term could be developed into a full - fledged national portal , are published in @xcite@xcite .
the portal will be developed on the basis of iaea experience and methodological support .
the development and support of the portal requires permanent governmental funding . for its development
it is necessary to build infrastructure and to have the availability of a critical mass of basic science to support practical applications .
the portal is a system that integrates all available ( in the country and abroad ) openly accessible information resources ( applications , databases , analytical systems , etc . ) , which allow the developers and users to interact with each other .
the portal should provide users with secure access to information and virtual channels of communication , e.g. they can work together on documents from geographically spaced locations ; access to all information resources of the portal through a single web - based mode with a strong collaborative personalization ( right of access to certain resources : data , services , applications , documents ) .
the mission of the portal consists in formation of favorable information , socio - cultural , business and educational environment for the sustainable development of nuclear industry in belarus .
portal objectives are the following : acceleration of search and access to necessary data and information ; creation of new knowledge ; promotion of participation in research , education and training programs in the nuclear industry .
it has to be an integration tool , an access tool for information resources and a communication tool .
basic principles of the portal creation are the next : * discussion the requirements of the portal with all stakeholders before development ; * developing a hierarchical taxonomy of the portal ; * constant testing the portal for compliance with technical requirements ; * maintaining transparency of the portal development ; * publishing a description of the portal ; * incorporation of representatives of all interested organizations to the group of developers .
interface that provides a mechanism for interaction between applications and the users of the portal is the main element of the portal .
interface provides coordination between teams and individuals , with a convenient and quick search and navigation .
other elements are the next : electronic document management system to ensure the preparation of the document with the required level of quality , intelligent search , categorization of information ; project management , including project planning , establishing project objectives , project schedule control of resources , planning and allocation of human and financial resources ; e - library ( documents collection , knowledge repository ) consisting of various electronic materials , reports , technical documentation , regulatory documents , training materials , etc . ; learning content management system with a system of courses and distance learning and the ability to develop and improve the courses of studies ; forums on the main areas of activity ; news feeds and other applications that are integrated into the portal .
thus , portal of nuclear knowledge will be simultaneously 1 ) a vertical portal ( portal - niche ) having a thematic focus and oriented on full coverage of stated themes ; 2 ) public portal open for the general internet public interested in nuclear subjects ; and 3 ) enterprise collaboration portal @xcite .
its main difference from the usual web site is availability of interactive services ( mail , news , forums , tools for collaborative work and individual users including distance learning tools ) .
draft structure of the portal is presented in fig.[fig1 ] .
content of portal is divided by subjects and marked by labels .
portal content means all information forming portal .
main subjects are the next : nuclear physics , nuclear materials , engineering and instrumentation , elementary particle physics , atomic physics , molecular physics , condensed matter physics , life sciences , chemistry , nuclear power and safety , safeguards , isotopes , fusion research and technology , nuclear fuel cycle and radioactive waste , etc .
labels can be the following : image , photo , video , audio , archive , news , countries , organizations , etc .
it is no need to place by copying all the information .
it is enough to make the necessary links to corresponding portals and sites containing this information .
distance learning system , which will be available within the on - line mode of portal should contain video lectures and animated lessons ( perhaps , the last ones should be broken into short modules ) , online tutorials , interactive quizzes and other materials developed by the best professors of the country .
such systems are actively being developed worldwide last 20 years .
distance learning system in the framework of nuclear knowledge portal will enhance the prestige and quality of education in the field of nuclear science and technology .
together , the e - library materials , training courses , databases , electronic documents ( photos , videos , etc . ) and other portal content will be organized in the nk base that contains knowledge in the field of nuclear technology , including nuclear and reactor physics , ionizing radiation , the application of nuclear methods in various fields of science and technology , radiation and radiochemistry , nuclear medicine , etc .
it is necessary to establish within the portal open areas , open and restricted areas and restricted areas depending on user access rights .
moreover , users with fewer rights should not even see a reference to restricted areas .
so , the structure of portal is a 4d matrix with the following layers : content , area , subject , label .
the main stages of the work on portal development consist of two parts .
the first stage with duration of two years includes the next steps : * identifying the source of the portal funding ; * determining the owner of the portal ; * defining the project team ( responsibility , roles , functions ) ; * determining the structure and the platform of the portal ; * identifying the tools , techniques and sources for data collection / accumulation and storage of information ( life cycle of documents ) ; * determining the necessary hardware and software ; * developing the taxonomy ; * developing a specification for the portal ; * creating a portal prototype based on selected technology ; * starting the collection of information ; * testing the portal in on - line mode .
full implementation of the portal on the second stage consists of purchasing computer equipment , installation the software , and experimental implementation of the portal : testing and evaluation of response time and accuracy of data ; checking the portal for the safety and effectiveness ; refining the portal ; developing a guide for users ; supporting the portal and filling it with information .
the novelty of the work presented can be formulated as creating the belarusian educational and research portal of nuclear knowledge , taking into account the specific conditions of the republic of belarus on the basis of free software developed by belarusian it specialists : e - lab electronic document management system @xcite , @xcite .
this system will be the basis of the interface , e - library , document management system , project management system , training materials , etc . in 2008 ,
the computer program `` laboratory information management system e - lab '' received the certificate no .
051 of the national intellectual property center of the republic of belarus .
it is implemented in the educational process of leading belarusian universities ( bsu , belarusian state technological university , belarusian national technical university ) , introduced in the chemical - toxicological laboratory of the minsk drug treatment clinic .
e - lab is on the basis of management of specimens , measurements and passports of fuels and lubricants of belarusian army ( since 2012 ) and belarusian branch of russian company gazpromneft ( since 2013 ) .
e - lab is an electronic system of the client - server architecture designed on the basis of free software : debian gnu / linux , web server apache , firebird database server using the application server php .
it runs under windows and linux .
it gives web based multi - user operation with different rights of access through widely used browsers .
e - lab operates reliably without interruption , completely secure from unauthorized access and has a fast response to user requests .
the system provides visibility and accessibility of information .
archival storage of materials on the site is provided by the close adjustment it to the place of storage with the control of the storage conditions .
it is provided a single interface for a wide range of integrated applications .
e - lab is the system easily modifiable and adaptable to the conditions of the project .
the presence of adaptable to the conditions of the project document management system e - lab based on free software is very important because the lack of the necessary software @xcite is a serious problem in the development of web portals .
however , it is now apparent that the existing e - lab system must be radically revised and modernized in order to simultaneously ensuring the smooth operation of a large number of users , as well as providing opportunities for e - learning .
development of belarusian educational and research web portal of nuclear knowledge will provide quick access to necessary information and create conditions for its exchange , accumulation and integrity of knowledge at the level ensuring a safe , sustainable and efficient development of nuclear energy and industry of the country , as well as the promotion of nuclear knowledge to attract to this area the most able young people and to create a positive image of nuclear science .
it is obvious that the portal development has no end especially in the part of its filling by information .
it is necessary to accomplish persistent content inventory . in the future , on the base of the proposed portal it may be changing its themes and the development of educational and research portals with distance learning system of the various types .
international atomic energy agency gc(47)/res/10 .
strengthening of the agency s activities related to nuclear science , technology and applications .
part b : nuclear knowledge .
vienna : iaea , 2003 . 7
p. international atomic energy agency .
knowledge management for nuclear industry operating organizations , iaea - tecdoc-1510 .
vienna : iaea , 2006 .
185 p. knowledge management for nuclear research and development organizations .
iaea - tecdoc-1675 .
vienna : iaea , 2012 .
74 p. status and trends in nuclear education .
iaea nuclear energy series , no .
ng - t-6.1 .
vienna : iaea , 2011 .
239 p. fast reactor knowledge preservation system : taxonomy and basic requirements .
iaea nuclear energy series , no .
ng - t-6.3 .
vienna : iaea , 2008 .
89 p. lobko a.s .
, sytova s.n . , charapitsa s.v .
educational and scientific nuclear knowledge portal . proc . of the iv congress of physicists belarus .
april 24 - 26 , 2013 , minsk .
p.419 - 420 .
sytova s.n . ,
charapitsa s.v .
, lobko a.s .
ability to use electronic document management system e - lab for the creation of educational and scientific nuclear knowledge portal .
congress csist2013 ,
november 4 - 7 , 2013 , minsk .
p.254 - 259 .
dubovskaya i.ya . ,
savitskaya t.a .
, lobko a.s .
et al . creating of creating of educational and research portal of nuclear knowledge .
dedicated to the 50th anniversary of mrti - bsuir ( minsk , march 18 - 19 , 2014 ) .
p.450 - 451 .
basyrov r.i .
1c - bitrix corporative portal . improving the efficiency of the company . st .
petersburg : piter , 2010 .
320 p. charapitsa s.v .
et al . electronic management system of accredited testing laboratory e - lab .
`` mathematical modelling and analysis '' , june 6 - 9 , 2012 , tallinn .
charapitsa s. v. et al
. electronic system of quality control and inventory management of fuels and lubricants `` e - lab fuel '' . deposited at the state organization `` belarusian institute of system analysis and information support for scientific and technical sphere '' ( so `` belisa '' ) 26.03.2013 no . d201310 .
85 p. klevtsov a.l .
, orlov v.y . , trubchaninov s.a . principles of creating a knowledge portal on the safety of nuclear installations .
nuclear and radiation safety .
p.53 - 57 . | the main objectives and instruments to develop belarusian educational and research web portal of nuclear knowledge are discussed .
draft structure of portal is presented .
keywords : portal , electronic document management system , nuclear knowledge . |
in recent years , classical behavior of non - hermitian hamiltonian systems@xcite as well as classical motion of hermitian systems for complex energies@xcite have attracted much interest .
investigation of classical mechanics in the complex domain is useful for understanding various classical and quantum mechanical phenomena such as barrier tunneling , dynamical tunneling@xcite , classical and quantum chaos@xcite , quantum correspondence principle , complex forms of uncertainty relations and the semiclassical limit of complex quantum field theories . since the earlier work on classical motion of non - hermitian systems @xcite , there have been several interesting results found on the various aspects of the subject@xcite .
numerical and analytical investigations have revealed that when energies are real , classical trajectories of complex @xmath3-symmetric non hermitian systems are closed and periodic . however , when energies of these systems are complex , the periodic trajectories usually become non periodic and open@xcite .
recently it was shown that even though most of the trajectories corresponding to complex energies are open and non periodic , for some systems , there are special discrete sets of curves in the complex - energy plane for which the trajectories are periodic@xcite . on the other hand , in non - hermitian and
non @xmath3-symmetric hamiltonian systems , even for real energies , almost all trajectories except a few are non - periodic and open .
it was also shown recently that when time is taken as a complex quantity with a specific fixed phase angle or as a specific complex function , non periodic trajectories of 1-d hamiltonian systems become periodic and closed@xcite . in this paper
we investigate the classical trajectories from a different point of view . here
we examine clasical behavior of the complex hamiltonian @xmath4 , ( where @xmath7 , @xmath9 , and @xmath5 is real such that @xmath16 is not @xmath3-symmetric ) for complex parameter @xmath6 and real energy @xmath17 .
outline of the paper is as follows . in the section [ sec:2 ] , analytic expressions for complex trajectories
are derived.expressions for periods of the periodic trajectories as well as time taken by unbounded trajectories to escape to infinity are found in terms of @xmath6 and energy @xmath17 .
we will show that for given real energy , the classical trajectories of the above quartic hamiltonian are open except for a discrete set of parameter values in the complex @xmath6-plane .
in section [ sec:3 ] we study how trajectories behave when energy is real and @xmath6 is a fixed complex parameter .
the classical trajectories of complex hamiltonian @xmath10 ( @xmath18 is investigated for real energies in section [ sec:4 ] and concluding remarks are given in section [ sec:5 ] .
in this section first we study in detail the classical motion of the complex quartic anharmonic oscillator .
we assume that the hamiltonian has the form @xmath19where @xmath5 is a real positive constant , @xmath6 is a complex constant and @xmath7 or @xmath9 .
first we derive expressions for @xmath20 and the period for the above hamiltonian .
when it is needed , value of @xmath21 is chosen as @xmath22 or @xmath9 . throughout this paper mass of the particle
is taken as half @xmath23 the equation of motion is @xmath24 the turning points of this system are taken as @xmath25 and @xmath26 and by integrating equation([eq:2])we have @xmath27where @xmath28 is the constant of integration which depends on initial conditions .
the left - hand side of the above equation is an elliptic integral of the first kind and hence equation ( [ eq:3 ] ) becomes @xmath29 , \frac{\left ( x_{1}-x_{2}\right ) ( x_{0}-x_{3})}{\left ( x_{0}-x_{2}\right ) ( x_{1}-x_{3})}\right)\ ] ] @xmath30 where @xmath31 is an elliptic function .
we invert the above equation in terms of jacobian elliptic function ` @xmath32 ' as @xmath33where @xmath34 and modulus @xmath35 ^{1/2}$ ] and @xmath36 is an arbitrary constant which is determined by the initial conditions .
note that @xmath20 in the above equation is still a solution of ( [ eq:3 ] ) , when @xmath25 and @xmath37 are cyclically changed ( e.g. @xmath38 ) . in order to understand how the trajectories behave ,
we need to recognize the periodic , bounded and unbounded properties of the function @xmath39 .
first we find complementary modulus @xmath40 and complete elliptic functions @xmath41 and @xmath42 .
they are defined by @xmath43 @xmath44 @xmath45 @xmath41 and @xmath42 are evaluated directly from the above equations and they are independent of phase angle @xmath46 .
the trajectory @xmath20 is given by @xmath47 the condition of trajectory become unbounded and the particle escapes to infinity is @xmath48 where @xmath49 satisfied for some real positive @xmath0 and the time taken for the particle to escape to @xmath50 is give by @xmath51 where @xmath52 ` @xmath32 ' is doubly periodic with period @xmath53 where @xmath54 and @xmath14 are integers .
therefore the condition for the trajectory to become periodic and particle does not escape to infinity is @xmath55 and @xmath56 then the trajectory is periodic with the period .
@xmath57 note that if @xmath58 the trajectory is still nonperiodic . by imposing the condition that im@xmath59
, we have @xmath60}{im\left [ k^{\prime } /z\right ] } ; \ \ \ \ m,\ n\ \epsilon \
\mathbb{z } \label{eq:13}\ ] ] where @xmath61 since @xmath14 and @xmath54 are integers and the energy @xmath17 is fixed , @xmath62 is rational and the equation ( 13 ) provides a discrete set of parameter values in the complex @xmath6 plane for which classical trajectories are periodic .
let @xmath63 figures [ fig1]a and [ fig1]b show how the ratio @xmath62 varies with discrete values of @xmath46 for the cases @xmath7 and @xmath64 respectively .
without loss of generality , the energy @xmath17 is taken as unity as it is real .
the results can be generalized for any real energy @xmath17 by simple rescaling of @xmath65 and @xmath0 .
next we consider the case when parameter @xmath6 is a fixed complex number and @xmath66 is a variable ( assume @xmath67 and @xmath68 ) . as a result ,
the equation ( [ eq:13 ] ) allows only a discrete set of values of @xmath17 for which trajectories are periodic .
it was found that these discrete values of @xmath17 can be either real or complex satisfying the condition ( [ eq:13 ] ) .
table [ tab : table_1 ] and table [ tab : table_2 ] show some real discrete values of @xmath17 which make trajectories periodic when @xmath7 and @xmath64 respectively .
figures [ p_1 ] and [ p_2 ] show the periodic trajectories of systems @xmath7 and @xmath64 for two values of real energies .
moreover , it was found that if energy @xmath17 is corresponding to the periodic trajectories of @xmath69 then @xmath70 will be the energy which makes trajectories of @xmath71 ( @xmath72 is the complex conjugate of @xmath6 ) periodic .
further @xmath17 and @xmath70 are solutions corresponding to the same @xmath14 and @xmath54 in the periodic condition ( [ eq:13 ] ) for these two hamiltonians respectively . in other words
if @xmath73 is the discrete set of energies for which classical trajectories of @xmath74 are periodic then @xmath75 is the set of energies for which trajectories of @xmath71 are periodic .
figures [ p_3]a and [ p_3]b show two periodic trajectories illustrating the above claim .
. classical energy spectrum corresponding to periodic trajectories of @xmath76 for various @xmath77 [ cols="^,^,^",options="header " , ]
next we assume that @xmath79 and @xmath80 in the hamiltonian ( [ eq:1 ] ) . then new hamiltonian has the form @xmath81 where @xmath82 is complex and @xmath83 .
the equation of motion is @xmath84where @xmath17 is the total energy .
following the same procedure as in section [ sec:2 ] , we obtain required equations . by integrating ( [ eq:15 ] )
we have @xmath85where @xmath28 is the constant of integration which depends on initial conditions .
the left - hand side of the above equation is an elliptic integral of the first kind and hence equation ( [ eq:16 ] ) becomes @xmath86 , -1\right ) { \small = 2}\left ( \mu e\right ) ^{1/4}{\small t+\alpha } \label{eq:17}\ ] ] where @xmath87 and @xmath31 is an elliptic function .
we invert the above equation in terms of jacobian elliptic function ` @xmath32 ' as @xmath88 note that modulus @xmath89 for the above problem . @xmath41 and
@xmath42 are defined in ( [ eq:7 ] ) and ( [ eq:8 ] ) and @xmath90 then @xmath41 and @xmath42 are obtained as @xmath91 @xmath92 as in the previous sections the condition of trajectory become periodic and particle does not escape to infinity is @xmath93 then the trajectory is periodic with the period @xmath94 since @xmath83 @xmath95 \left ( \cos \left ( \theta /4\right ) -i\sin \left ( \theta /4\right ) \right ) .
\label{eq:24}\ ] ] @xmath96 .
\label{eq:25}\ ] ] since @xmath41 is real and @xmath17 is real and positive , by imposing the condition that im@xmath59 , we have @xmath97 or @xmath98 ; \ \ \ \ m,\ n\ \epsilon \ \mathbb{z } . \label{eq:27}\ ] ] when @xmath99 and @xmath100 , @xmath101 and it is hermitian .
then @xmath16 possesses periodic trajectories and the period @xmath102 becomes @xmath103 but the period is corresponding to the minimum non zero @xmath104and the resulting period is @xmath105 on the other hand when @xmath106 and @xmath107 , @xmath108 and it is the non - hermitian ` wrong sign ' potential which also possesses periodic trajectories .
the period is @xmath109 it is evident from equations ( 29 ) and ( 30 ) that the hamiltonians @xmath110 and @xmath111 have the same classical period ( i.e. @xmath112 ) .
note that these two hamiltonians are the classical limit of the quantum mechanical isospectral hamiltonians as shown in @xcite .
in this paper we have presented three main results .
the first is that , for given real energy , the classical trajectories of quartic hamiltonians @xmath4 , ( where @xmath5 is real , @xmath6 is complex , and @xmath7 @xmath8 @xmath113 ) are closed and periodic only for a discrete set of parameter curves in the complex @xmath6-plane .
the second result is that given complex parameter @xmath6 , the classical trajectories are found to be periodic only for a discrete set of real energies . as a result ,
real classical energies get discretized or quantized by the condition that trajectories are periodic and closed .
this result is analogous to what was obtained by anderson et al in @xcite for real potential parameters with complex @xmath17 ( here it is for complex potential parameters with real energies ) .
further we showed that if @xmath114 is the discrete set of energies for which classical trajectories of @xmath69 are periodic then @xmath115 is the set of energies for which trajectories of @xmath71 are periodic .
we presented our results with illustrations .
it is important to note that when @xmath6 is complex and not pure imaginary , the entire quantum eigen spectrum corresponding to the hamiltonian @xmath16 is complex and eigenenergies do not come as complex conjugate pairs . therefore @xmath16 can not be pseudo hermitian and can not have any antilinear symmetry . as the third result
, we showed that for real energies , the classical trajectories of complex hamiltonian @xmath10 ( @xmath18 are periodic only for discrete values of @xmath46 satisfying the condition @xmath116 $ ] for @xmath14 and @xmath15 . further it was found that hamiltonians @xmath117
@xmath118 and @xmath111 which are the classical limit of the quantum mechanical isospectral hamiltonians introduced in @xcite , have the same classical period .
c. m. bender and s. boettcher , phys . rev . lett .
80 , 5243 ( 1998 ) . | anderson et al have shown that for complex energies , the classical trajectories of _ real _ quartic potentials are closed and periodic only on a discrete set of eigencurves .
moreover , recently it was revealed that , when time is complex @xmath0 @xmath1 certain real hermitian systems possess close periodic trajectories only for a discrete set of values of @xmath2 . on the other hand
it is generally true that even for real energies , classical trajectories of non @xmath3- symmetric hamiltonians with complex parameters are mostly non - periodic and open . in this paper
we show that for given real energy , the classical trajectories of _ complex _ quartic hamiltonians @xmath4 , ( where @xmath5 is real , @xmath6 is complex and @xmath7 @xmath8 @xmath9 ) are closed and periodic only for a discrete set of parameter curves in the complex @xmath6-plane .
it was further found that given complex parameter @xmath6 , the classical trajectories are periodic for a discrete set of real energies ( i.e. classical energy get discretized or quantized by imposing the condition that trajectories are periodic and closed ) .
moreover , we show that for real and positive energies ( continuous ) , the classical trajectories of _ complex _ hamiltonian @xmath10 ( @xmath11 are periodic when @xmath12 $ ] for @xmath13 @xmath14 and @xmath15 .
= 1 |
the study of the nuclear equation of state ( eos ) and the behavior of nuclear matter under extreme conditions is crucial to our understanding of many nuclear and astrophysical phenomena . with the increasing availability of radioactive ion - beams @xcite the study of structure and reactions of exotic nuclei
are now possible , thus providing information concerning the isospin dependence of asymmetric nuclear matter @xcite .
recently , the study of the symmetry energy for such systems has been an active area of interest @xcite .
most studies of the eos involve infinite or semi - infinite nuclear matter and examine the dependence of the eos on the parametrizations of the effective interaction as well as its relation to the macroscopic and macroscopic - microscopic models of nuclear matter @xcite . for finite nuclei
the eos near the equilibrium density can be investigated via collective observables such as the isoscalar monopole vibrations ( breathing mode ) @xcite .
in addition , there are empirical methods relating the compressibility of finite nuclei to that of infinite nuclear matter @xcite .
however , the behavior of the eos for finite nuclei far from the equilibrium density is poorly known . in this manuscript
we introduce a new method for calculating the zero temperature eos and related quantities for finite nuclei within the mean - field description of nuclear properties . in section [ sec : theory ] we outline the general formalism for our calculations .
section [ sec : results ] discusses the application of the formalism to a sample set of nuclei and the results obtained .
the paper is concluded with a summary in section [ sec : summary ] .
in order to study finite nuclei away from their ground state equilibrium we take advantage of the density constrained hartree - fock ( dchf ) method @xcite .
the _ density constraint _ is a novel numerical method that was developed in the mid 1980 s and was used to provide a microscopic description of the formation of shape resonances in light systems @xcite .
recently , we have used the same method to calculate heavy - ion interaction potentials from the tdhf time - evolution of nuclear collisions @xcite . in the traditional constrained hartree - fock ( chf )
notation , density constraint corresponds to the replacement @xmath0 the numerical procedure for implementing this constraint and the method for steering the solution towards a specified @xmath1 is discussed in refs .
the convergence property is as good if not better than the traditional chf calculations with a constraint on a single collective degree of freedom . in practice
, we have obtained accurate densities for spherical nuclei using a radial hartree - fock program .
this density was then fitted to a parametric function of the form @xmath2 where @xmath3 denote the parameters to be fitted to reproduce a particular density profile . in all cases
the resulting non - linear fits were indistinguishable from the fitted density .
subsequently , the constraining density was obtained via a scale transformation of the above density profile @xmath4 followed by a renormalization to produce the correct mass number for the nucleus under study .
this scaling results in the compression of the bulk and stretching of the surface @xcite .
naturally , the minimum of all eos curves occur at the unconstrained hartree - fock minimum corresponding to @xmath5 .
the value of the scale @xmath6 was generally limited to the range @xmath7 . in fig .
[ fig : density ] we show the change in the density for various values of the scaling parameter @xmath6 in the case of the @xmath8ca nucleus . as the primary interaction
we have used the skyrme sly4 force @xcite , with and without the coulomb term , and including all of the spin - dependent terms .
we have performed the calculations using our new hartree - fock program discussed in ref .
@xcite .
in dealing with a finite nucleus we have faced a conceptual problem of deciding what density value to use for plotting the density dependence of the eos . unless otherwise stated we have used the _ central density _ as the reference density value for each value of the scaling parameter @xmath6 .
alternatively , one can choose the nuclear matter equilibrium density , 0.16 @xmath9 , as the reference density from the density profile .
the calculated values for incompressibility is slightly dependent on the choice of the reference density point due to the structure in the density profile .
our calculations show that this is about 10% or less and becomes negligible for heavy systems
. one disadvantage of using the central density as the reference value is that the equilibrium densities for different nuclei do not occur at the same value , thus making the calculation of the symmetry energy , which is essentially taking the difference of two eos curves along an isotope chain , erroneous .
for this reason , in the calculation of the symmetry energy we have used the density value 0.16 @xmath9 as the equilibrium density .
( color online ) scaled density obtained for the @xmath8ca nucleus for various values of the scaling parameter @xmath6 .
the inset shows the same quantity on a logarithmic scale . the force used was sly4 without the coulomb contribution . ] in order to extract the incompressibility coefficient , @xmath10 , we have expanded the eos ( binding energy per particle as a function of density ) around the equilibrium density @xmath11 using the expression @xmath12 where @xmath13 is the binding energy at the equilibrium density , and @xmath10 is the incompressibility coefficient @xmath14 we found in practice that this expression provides an excellent fit except at the extreme values of the density . to fit the entire curve perfectly a small linear contribution as well as a cubic term
could be included .
the results for the incompressibility is approximately 5%-10% higher if the full curve is used in the fit . finally , to extract the symmetry energy we use the expression @xmath15 where the isospin asymmetry parameter is defined as @xmath16 .
the @xmath17 dependence of the symmetry energy is generally acknowledged to be @xmath18 where the higher order terms in @xmath17 are assumed to be small .
traditionally , the symmetry energy can also be expanded around the equilibrium density as @xmath19 where we have defined the quantities @xmath20 and @xmath21 , which are related to the symmetry pressure and symmetry compressibility @xmath22
using our approach we have first investigated the eos for the @xmath23o nucleus , with two forces , sly4 and skm@xmath24 @xcite , without the coulomb interaction .
we found very little difference between the two forces as expected since most modern skyrme forces have similar nuclear matter incompressibility values .
for the incompressibility we find @xmath25 mev .
( color online ) eos for the @xmath26ca , @xmath8ca , and @xmath27ca nuclei as a function of density .
please see the comments in the manuscript regarding the density values . the force used was sly4 without the coulomb contribution . ]
we have then repeated the same two force study for the @xmath26ca nucleus , yielding an incompressibility of @xmath28 mev . in order to investigate the behavior of neutron rich systems
we have performed calculations for @xmath8ca and @xmath27ca systems . in fig .
[ fig : ca1 ] we show the eos for all of these ca nuclei without the coulomb force .
again , we stress that the density scale shown in fig . [
fig : ca1 ] is determined by choosing the central density in the density profile of each nucleus .
consequently , the equilibrium value for the eos is different for each nucleus . as we can see from fig .
[ fig : ca1 ] , the @xmath26ca and @xmath8ca systems have a very similar eos behavior , the primary difference being the energy shift due to the difference in the two binding energies .
the calculated incompressibility of @xmath8ca is @xmath29 mev , only slightly higher than the one for @xmath26ca .
however , the situation for @xmath27ca is significantly different , in addition to the large shift in the energy scale the incompressibility decreases to a value of @xmath30 mev , indicating a somewhat softer nucleus .
the values for the incompressibility obtained here , perhaps with the exception of the @xmath27ca system , are in general agreement with those given in ref .
@xcite .
we have repeated some of the calculations by including the coulomb interaction as well . in general ,
eos curves are shifted up due to the decrease in the binding energy per nucleon .
in addition , we see a small decrease in the incompressibility modulus . for @xmath26ca we find @xmath31 mev .
the incompressibility for @xmath8ca decreases to @xmath32 mev , and this value drops down to @xmath33 mev for @xmath27ca .
( color online ) the quantity @xmath34 for the @xmath8ca and @xmath27ca nuclei as a function of density . ]
we have also investigated the symmetry energy obtained from eq .
[ eq : symmetry1 ] , which is essentially the difference between the curves shown in fig .
[ fig : ca1 ] . the results for the quantity @xmath34 defined in eq .
[ eq : symmetry2 ] , corresponding to nuclei @xmath8ca and @xmath27ca having @xmath17 values of 0.16667 and 0.33334 , respectively , are shown in fig .
[ fig : sym1 ] . as we observe again
, the two systems behave very differently .
unlike infinite nuclear matter the curves do not cross at the equilibrium density since the binding energy per nucleon is different for different nuclei .
the values of @xmath35 around the equilibrium density are about 10 - 14 mev , which is considerably lower than the estimated range for nuclear matter value of 32 mev for the sly4 force . using the results of fig .
[ fig : sym1 ] we have extracted the quantities @xmath20 and @xmath21 for @xmath8ca and @xmath27ca .
the results show a very large variation between the two systems . for @xmath8ca
we find 66.6 mev and -403 mev for @xmath20 and @xmath21 , while for @xmath27ca these numbers become 1.55 mev and -193 mev , respectively .
we have introduced a new method for calculating the nuclear eos for finite nuclei including all of the terms in the nuclear effective interaction .
the calculated values agree well with known values obtained by other means , such as those deduced from experimental giant monopole resonances .
much work has gone into understanding the effects of various components of the effective interaction which vanish or become very small in the infinite nuclear matter limit . with the increased availability of new radioactive neutron and proton rich nuclei
the study of eos and symmetry energy along isotope chains of finite nuclei have become more urgent .
we believe that the density constrained hf method is a step in this direction . | we present a new approach for calculating the nuclear equation of state and compressibility for finite nuclei using the density - constrained hartree - fock method .
= 10000 = 10000 = 10000 |
the purpose of this talk is to explain to statisticians how astrophysicists , mostly using fourier transform techniques , go about analyzing x - ray time - series data obtained from accreting compact objects ( neutron stars and black holes ) , and to point out a few problems with the usual approaches .
the point will be made , that the conglomerate of statistical methods that is being applied in this branch of high - energy astrophysics , even though most definitely not always rigorous , on the whole serves its purpose well and is providing astronomers with the quantitative answers they require .
this talk will aim , however , at a few areas where we run into problems , and where more statistical expertise might help . in thinking about how to solve the problems that i shall outline , it will be important to keep an eye on what the capabilities of the existing methods are , as those capabilities will need to be preserved in whatever new approach
one would like to propose .
mostly , accreting neutron stars and black holes occur in double star systems known as x - ray binary stars , where a normal star and the compact object are in a close orbit around each other ( fig.[lmxb ] ) .
matter flows from the normal star to the compact object by way of a flat , spiraling flow called an accretion disk , and finally accretes onto the compact object .
a large amount of energy is released in this process ( typically 10@xmath0 to 10@xmath1erg / sec ) , and is emitted , mostly in the form of x - rays .
the characteristic variability timescale for the x - ray emitting regions is predicted ( and observed ) to be very short ( less than a millisecond ) . by studying the properties of this rapid x - ray variability it is possible to extract a great deal of information about the flow of matter onto the compact object , and , indirectly , about the object itself .
see van der klis ( 1995 ) for a recent review of the results of studies of this type . in order to understand the character of the date we are dealing with ,
i shall follow the flow of information from star to computer .
an x - ray binary star emits x - ray photons at a very large rate ( say , 10@xmath2photons / sec ) . for all practical purposes ,
the x - ray photon rate produced by the star can be considered as a continuous function of time @xmath3 .
these photons are emitted isotropically , or at least over a solid angle of order 4@xmath4sterad . because the x - ray detector onboard the x - ray satellite spans only a very small solid angle as seen from the x - ray star ,
only a very small fraction @xmath5 ( say , 10@xmath6 ) of the photons is detected in the instrument .
the time series of photon arrival times @xmath7 ( with @xmath8 the total number of detected photons ) is the information that , ideally , we would like to have available for analysis .
however , typically , instrumental limitations ( and the maximum telemetry rate ) prevent the registration and transmission of all these arrival times .
therefore , onboard the satellite , the data are _ binned _ into equidistant time bins .
the information that is finally telemetered to the ground station consists of a sequence of numbers @xmath9 , where @xmath10 is the total number of time bins , @xmath11 representing the number of photons that was detected during time bin @xmath12 . in most cases , and unlike the usual case in astronomy , the time bins are equidistant and contiguous ( no gaps ) .
the statistical problem facing us can be summarized as follows : `` _ given @xmath13 , reconstruct as much as possible about @xmath3 _ '' .
we shall assume that for the bright x - ray binaries that i am discussing here , the rate of background photons can be considered to be negligible , and that there are no other relevant effects affecting the photon arrival time series than the huge geometrical dilution factor just described .
therefore , if the x - ray star does not vary intrinsically , the @xmath14 will to a high degree of precision be uniformly and randomly distributed , so that the @xmath15 follow poisson statistics appropriate to a rate @xmath16 , with a standard deviation @xmath17 equal to @xmath18 .
of course these stars _ do _ vary .
the variability time scales of interest are @xmath191millisecond , and the detected photon rates 10@xmath2010@xmath21photons / sec , which means that the time bins must be chosen such that typically there are on average only a few ( sometimes @xmath221 ) photons per time bin , and @xmath17 is of order @xmath23 ( sometimes @xmath24 ) .
as the intrinsic variability of the star often has a relatively small amplitude , only a few percent of the total flux , it is clear that we are in a low signal - to - noise regime ( with the `` signal '' the intrinsic stellar variability and the `` noise '' the poisson fluctuations ) .
fortunately , typical observations span 10@xmath25 to 10@xmath21sec , so that the number of time bins @xmath10 ( the number of `` measurements '' ) is 10@xmath26 to 10@xmath27 , which allows us to recover the signal from the noise .
the techniques used for this are described in the next section .
the standard approach ( see van der klis 1989 ) is to divide the time series into @xmath28 equal - length segments of @xmath29 time bins @xmath30 each ( so , ideally @xmath31 ) , to calculate the discrete fourier transform of each segment : @xmath32 to convert this into a power spectrum for each segment @xmath33 and then to _ average _ these power spectra ( see below ) .
note that in our application @xmath34 , the number of photons detected in the segment . with this power spectral normalization , due to leahy et al .
( 1983 ) , it is true that if the @xmath15 are distributed according to the poisson distribution , then the @xmath35 follow the @xmath36 distribution with 2 degrees of freedom , so @xmath37 and @xmath38 .
this white noise component with mean level 2 and standard deviation 2 , induced in the power spectrum due to the poisson fluctuations in the time series , is called the `` poisson level '' .
it can be considered as `` background '' in the power spectrum , against which we are trying to observe the other power spectral features , which are caused by the intrinsic variability of the x - ray binary .
the physical dimension of the thus defined powers is the same as that of the time series : @xmath39=[a_j]=[x_k]$ ] .
often the y - axis of plots of power spectra in this normalization is just labeled `` power '' , which reflects the fact that the physical interpretation of the @xmath35 in terms of properties of the star is inconvenient .
for this reason , in recent years another power spectral normalization has become popular where the powers are reported as @xmath40 , with @xmath41 the `` count rate '' , the number of detected photons per second : @xmath42 , where @xmath43 is the duration of a segment .
the @xmath44 are dimensionless , and can be interpreted as estimates of the power density @xmath45 near the frequency @xmath46 , where @xmath47 is a function of frequency whose integral gives the fractional root - mean - square amplitude @xmath48 of the variability .
this latter quantity is defined as @xmath49 ( this follows directly from parseval s theorem . )
the fractional rms amplitude @xmath50 due to fluctuations in a given frequency range @xmath51 is given by @xmath52 so , the physical interpretation of @xmath47 is easy : it is the function whose integral gives you the square of the fractional rms amplitude of the variability in the original time series .
the physical unit used for @xmath47 is ( rms / mean)@xmath20/hz , where `` rms '' and `` mean '' both refer to the time series ; `` rms / mean '' is just the dimensionless quantity @xmath48 .
the averaging of the power spectra mentioned above is usually performed both by averaging individual power spectra ( from different segments ) _ together _ ( averaging the @xmath35 s with the same @xmath53 from the @xmath28 different segments ) and by averaging powers at adjacent frequencies ( @xmath54 to @xmath55 , say ) .
the main purpose of this is of course to decrease the standard deviation of the power estimates , which in the raw spectra is equal to the mean power .
the reason to calculate many power spectra of segments of the data rather than one very large power spectrum of the whole data set , apart from computational difficulties with this approach , is that in this way it is possible to study the variations in the power spectrum as a function of time .
the final step in the analysis is to fit various functional shapes @xmath56 to the power spectra using the method of @xmath36 minimization that is also used in x - ray spectroscopy ( the levenberg - marquardt method described in press et al .
1992 , chapter 15 ) .
because many individual power estimates have been averaged in the analysis process , the central limit theorem ensures that the uncertainties of the final power estimates @xmath57 are approximately normally distributed , as required for this method to work well .
a problem is what `` uncertainties '' @xmath58 to assign to these power estimates .
usually @xmath59 is assumed , where @xmath60 is the number of individual powers averaged to obtain @xmath57 ( @xmath61 stands for the `` width '' , the number of adjacent powers averaged ; @xmath28 for the number of averaged power spectra ) .
this is approximately correct , and was experimentally verified ( fig.[cesme ] ) , in the case that the dynamic range of the power spectrum is dominated by the intrinsic differences in the mean amplitude of the star s variability as a function of frequency rather than by the stochastic fluctuations in power .
if instead the stochastic variations dominate then this procedure for estimating the uncertainties can lead to severe underestimation of the power , as accidentally low powers will get high weights in the fitting procedure and vice versa . a solution to this problem that is sometimes adopted
is to estimate the uncertainty in @xmath62 as @xmath63 , where @xmath64 is the fit function and @xmath65 the frequency corresponding to @xmath57 .
the method described in the previous section works .
it allows astronomers to quickly characterize the variability properties of large amounts of data , to study the changes in the properties of the variability as a function of time and other source characteristics , and to measure amplitudes and characteristic time scales of the variability .
the method straightforwardly extends to simultaneous multiple time series , for example time series obtained in different photon energy bands . for @xmath66 simultaneous time series
, it is possible to calculate @xmath67 different cross - spectra between them , and to look for systematic time delays in the variability as a function of energy .
a comprehensive description of the methods used for this can be found in the paper by vaughan et al .
( 1994 ) .
a very important aspect is the possibility to identify different `` power spectral components '' in the variability .
this is done by studying the changes in the shape of the power spectra as a function of time and other source properties , such as brightness or photon energy spectrum .
it usually turns out that the simplest way to describe the changes in the power spectrum as a function of time is in terms of the sum of a number of components whose properties ( strength , characteristic frequency ) depend in a smooth , systematic and repeatible way on , for example , brightness .
if this is the case , then it is natural to interpret these power spectral components as due to different physical processes ( or different aspects of the same process ) that are all affecting the count rates at the same time , and that have been disentangled from each other in the analysis just described .
examples of power spectral components that are distinguished in practice are `` power law noise '' , `` band - limited noise '' and `` quasi - periodic oscillations ( qpo ) '' .
all of these are all presumed to be stochastic processes in the time series that cause , respectively , a component in the power spectra that fits a function @xmath68 , one that fits @xmath69 and one that fits a lorentzian @xmath70 . for a component to be called
qpo it is sufficient that the corresponding power spectral component has a shape approximately described by @xmath71 with the peak full - width - at - half - maximum ( fwhm ) @xmath72 less than half its centroid frequency @xmath73 ; of course this definition is arbitrary .
figure[bhcps ] shows a number of actually observed power spectra , plus the functions that were fit to them in order to describe them in terms of the power spectral components just mentioned .
qpo peaks , band - limited noise and power law noise can all be seen .
sometimes the shape of the observed power spectrum is ambiguous with respect to the decomposition into power spectral components as described here .
an advantage of the @xmath36 minimization method is that it allows to quantify the degree of this ambiguity by comparing the @xmath36 values of the different possible combinations of fit parameters .
one problem is , that in order to reach the lowest frequencies in the first place , it is necessary to choose the length of the time segments @xmath43 relatively large and therefore @xmath28 relatively small .
this means that at the lowest frequencies the statement made in section[ana ] , that a large number of individual powers has been averaged and therefore the average power is approximately normally distributed is no longer true .
other ( e.g. , maximum - likelihood ) fitting procedures are required to take the true distribution of the average powers into account , but such methods are not usually applied .
see papadakis and lawrence ( 1993 ) for a discussion of a method to remedy some of these problems .
a more serious problem is , that the true power spectrum at these frequencies often seems to be a quite steep power law .
the finite time window @xmath43 in those situations leads to so - called `` low - frequency leakage '' ( see deeter 1984 and references therein ) : power shows up at a higher frequency in the power spectrum than where it belongs .
one way to see this is by noting that the lowest frequency accessible in the discrete fourier transform is @xmath74 .
if there is a lot of variability at lower frequencies than this , then due to these slow variations the time series of an individual time segment will usually have a large overall trend .
the fourier transform of this trend produces a power law with index @xmath752 in the power spectrum .
another way to describe the effect of low - frequency leakage is by noting that according to the fourier convolution theorem the fourier transform calculated in the finite time window @xmath43 is related to the true fourier transform by a convolution of the true transform with the transform of the window function .
as the window function is a boxcar , its fourier transform is the well - known sinc function , with a big central lobe and upper and lower sidelobes that gradually decrease in amplitude . for true power spectra steeper than a power law of index @xmath752
, the contribution to the convolution of the lower side lobes overwhelms that of the central lobe , and the result is a power spectrum that is a power law with index @xmath752 ( e.g. , bracewell 1965 ) .
so , for any true power spectrum steeper than @xmath752 , the actually measured power spectrum will have slope of @xmath752 .
there are well - known solutions to this well - known problem .
the most famous one is data tapering : instead of a boxcar window a tapered window is used , i.e. , a window that makes the data go to zero near its end points more gradually than by a sudden drop to zero .
other methods are polynomial detrending ( fitting a polynomial to the data and subtracting it ) and end - matching ( subtracting a linear function that passes through the first and the last data point ) .
deeter et al .
( 1982 ) and deeter ( 1984 ) have explored a number of non - fourier methods .
all these methods work in the sense that to some extent they suppress the side lobes of the response function and therefore they are able to recover power laws steeper than with index @xmath752 ( the value of the power law index where the method breaks down is different in each case ) .
however , typically these methods have only been evaluated for the case where the time series is pure power - law noise , and in many cases even only with respect to their effectiveness in recovering the power law index , not even the noise strength .
some methods require that the index of the power law is known in advance ! nearly nothing is known about the way in which these methods affect the results of fits to power spectra with complicated shapes such as those described above .
these methods may recover the correct power law index for the low - frequency part of the power spectrum , but what will be the effect on the fractional rms values and time scales of all the other components ? for this reason , these methods are not usually applied .
the high - frequency end of the power spectra , near the nyquist frequency , is of particular interest to astrophysicists , as it is there that we expect to find the signatures of the fastest accessible physical processes going on in the star .
a question that is often asked in this context is : `` what is the shortest variability time scale @xmath76 we can detect in the data ? '' .
generally , what is observed at high frequency is that the power spectrum more or less gradually slopes down towards the poisson level .
the problem is to decide out to which frequency intrinsic power is observed above the poisson background , and _ what this number means_. an , i think , decidedly misleading way to answer the question about the shortest time scale is by determining the _ shortest time interval @xmath77 within which significant variations can be detected_. it is obvious , that for a slowly varying source , by zooming in on some gradual slope in its light curve @xmath3 , this shortest time interval can be made as short as one wishes , just by improving the quality of the data ( by increasing @xmath5 for example ) . yet
, most of the work on `` shortest time scales '' seems to aim for measuring @xmath78 rather than @xmath76 .
it seems clear that one must also take into account the amplitude of the variations , not just the time within which they occur to do something that is physically useful .
following fabian ( 1987 ) one can for example define the variability time scale as where the dot denotes the time derivative .
defined this way , @xmath76 is a measure of how steeply @xmath80 changes with @xmath81 , and depends itself on @xmath81 . using a power spectrum
, one would measure some average of this quantity by determining the fractional rms amplitude @xmath82 near some high frequency @xmath83 in the way described in section[ana ] , and then write where @xmath85 is a constant of order unity depending on assumptions on exactly how the variations causing the power in the spectrum took place .
however , for precise work one has to worry about low - frequency leakage , too .
exactly the same problem as described in section[low_f ] can occur here : power from lower frequencies can leak up to higher frequencies due to the sinc response associated with the power estimators . that this problem is serious is apparent from the fact that the observed power spectrum of for example the famous black - hole candidate cygx-1 has an index of @xmath752 for frequencies above 10hz ( see fig.[bhcps ] ) , just the value at which low - frequency leakage begins to worry us .
it would be of great interest to have a foolproof way to subtract power that has leaked up from lower frequencies , or even to have a way to make a conservative estimate of ( obtain a lower limit on ) the true high - frequency power .
it is not clear to what extent this can be accomplished . obviously , as in any convolution problem
, some information has been lost , but how much can be recovered is a problem x - ray astronomers do not know the answer to .
i wonder to what extent standard deconvolution procedures might be useful here .
i note parenthically that a method that has been applied for determining the shortest variability in a time series by meekins et al .
( 1984 ) seems to suffer from the same problem of low - frequency leakage . in this method ,
the time series is divided up in very short segments of , for example , @xmath86 bins each . in each segment a quantity called `` chi - squared '' is calculated as follows : here @xmath15 is the number of photons detected in bin @xmath88 and @xmath89 is the total number of photons in the segment .
one recognizes an `` observed over expected '' variance ratio for an expected poisson distribution .
the distribution of this quantity is then compared to that expected if all variability in the time series would be due to poisson fluctuations and if a significant excess is found , this is interpreted as detection of variability on time scales between the length @xmath43 of a segment and the duration of one bin .
one would expect low - frequency leakage to be as much of a problem here as in the power spectral method .
it is easy to see that if there are variations in the time series on time scales much longer than the segment length , the data points in the segments will usually follow steep trends , causing large values of @xmath90 that are not related to variability on time scales shorter than @xmath43 .
indeed , meekins et al . show that @xmath90 is closely related to the fourier power in the segment ,
so from the point of view of low - frequency leakage the meekins et al .
method is less effective than the standard power spectral approach , as it requires @xmath43 to be quite short and therefore increases the probability of steep trends in the data segments .
bracewell , r. , the fourier transform and its applications , mcgraw - hill , 1965 .
deeter , j. , astrophys .
j. 281 , 482 , 1984 .
deeter , j. , and boynton , p.e .
, astrophys .
j. 261 , 337 , 1982 .
deeter , j. , astrophys .
j. fabian , a.c .
, in proc .
`` the physics of accretion onto compact objects '' , lecture notes in physics 266 , 229 , 1987 .
leahy , d.a .
, darbro , w. , elsner , r.f . ,
weisskopf , m.c . ,
sutherland , p.g . , kahn , s. , grindlay , j.e .
, astrophys .
j. 266 , 160 , 1983 .
meekins , j.f . , wood , k.s . , hedler , r.l . ,
byram , e.t . ,
yentis , d.j . , chubb , t.a . , friedman , h. , astrophys .
j. 278 , 288 , 1984 .
mitsuda , k. , dotani , t , publ .
japan , 41 , 557 , 1989 .
papadakis , i.e. and lawrence , a. , mon . not .
r. astron .
261 , 612 , 1993 .
press , w.h .
, teukolsky , s.a . , vetterling , w.t . , flannery , b.p .
, `` numerical recipes in fortran '' , 2nd edition , cambridge univ . press , p.650 , 1992 . van der klis , m. , in `` x - ray binaries '' , w.h.g . lewin et al .
( eds . ) , cambridge univ . press , p.252 , 1985 .
van der klis , m. , in `` timing neutron stars '' , gelman and van den heuvel ( eds . ) , nato asi c262 , kluwer , p. 27
vaughan , b.a . ,
van der klis , m. , wood , k.s . ,
norris , j.p . , hertz , p. , michelson , p.f . ,
van paradijs , j. , lewin , w.h.g . , mitsuda , k. , penninx , w. , astrophys .
j. 435 , 362 , 1994 .
vikhlinin , a. , churazov , e. , gilfanov , m. , astron .
287 , 73 , 1994 .
zhang , w. , jahoda , k. , morgan , e.h . ,
giles , a.b .
, astrophys .
j. 449 , 930 , 1995 . | i discuss some practical aspects of the analysis of millisecond time variability x - ray data obtained from accreting neutron stars and black holes .
first i give an account of the statistical methods that are at present commonly applied in this field .
these are mostly based on fourier techniques . to a large extent
these methods work well : they give astronomers the answers they need .
then i discuss a number of statistical questions that astronomers do nt really know how to solve properly and that statisticians may have ideas about .
these questions have to do with the highest and the lowest frequency ranges accessible in the fourier analysis : how do you determine the shortest time scale present in the variability , how do you measure steep low - frequency noise . the point is stressed that in order for any method that resolves these issues to become popular , it is necessary to retain the capabilities the current methods already have in quantifying the complex , concurrent variability processes characteristic of accreting neutron stars and black holes .
# 1*(#1 ) * |
the sun is our nearest active flare star that directly affects terrestrial phenomena and near - earth as well as planetary environments by its various processes of activity .
the term `` solar activity '' means a wide circle of astrophysical actions in spatial , temporal and energy scales .
solar bursts are ones of significant signatures of these processes .
they are present in all parts of electromagnetic spectrum , from gamma rays to radio waves .
an in - depth examination of characteristic parameters in the solar bursts allows one to establish the physical background and initiation mechanisms responsible for one or another observed burst .
exactly this provides the remote diagnostics of solar plasma evolution in corona .
there is the generally accepted classification of solar bursts according to its time - frequency features @xcite . in this paper
we focus our attention on type iv bursts frequently associated with the most powerful and large - scale solar events such as solar flares and coronal mass ejections ( cmes ) .
the type iv bursts represent broadband continuum emissions with a variable time structure @xcite appearing in different forms : fiber bursts @xcite , oscillations of radio emission in intensity @xcite , or ripply narrow - band patterns called zebra structures @xcite .
this radiation was firstly recognized at meter wavelengths by @xcite and referred to the type iv burst .
@xcite initiated to distinguish type iv bursts in height and movement of burst s sources .
subsequently , due to the fact that the radio emission was often related to the continuum at decimeter and centimeter wavelengths with differentiation of various phases within each individual event , this gave rise to overabundance of type iv categories and sub - categories .
@xcite made an attempt to arrange the diversity .
as a result , the continuum radio emission were divided into four sub - categories : two types of flare continuum , slow drift continuum , and moving type iv bursts .
the radiation of the first three sub - categories corresponds to stationary continuum sources .
the radio emission of the last sub - category is associated with motion of emitters usually attributed to cmes . in this connection
, it should be noted that the type iv bursts have two distinct classes : moving ( type ivm ) and stationary ( type ivs ) bursts .
usually each stationary radio burst is the broadband flare - related continuum emission .
it is generated during an explosive phase of solar flares or appearing with some delay after such a flare , and it is situated above the associated flare location .
this long - lasting radiation is thought to be produced by energetic electrons trapped within coronal structures such as magnetic arches .
many observations of low - frequency type iv bursts were performed in the 1970s-80s .
particularly , by means of the clark lake radio telescope ( usa ) the type iv radio emission was investigated at meter - decameter waves @xcite .
furthermore , using the culgoora radio heliograph ( australia ) , before it was put out the service , the type iv bursts were recorded on frequencies 43 , 80 , 160 mhz @xcite .
solar observations with the help of gauribidanur radio heliograph at 80 mhz supplemented by spectral data records within 35 - 85 mhz belong to more modern researches on the low - frequency type iv radiation .
however , the moving type iv bursts were only taken into account in a few studies @xcite .
the type iv bursts at decameter wavelengths are infrequent events of solar activity .
this radiation below 30 mhz was recorded by the utr-2 radio telescope @xcite on several occasions during the summer observational campaigns in 2002 - 2004 years .
this allows some statistical analysis about the decameter bursts @xcite .
in particular , the bursts had durations in the range from one to several hours , as well as their fluxes reached up to @xmath2 s.f.u .
( 1 s.f.u .
= 10@xmath3 w m@xmath4 hz@xmath5 ) .
the most of them were clearly associated with corresponding cmes , but in several cases the bursts appeared without the presence of cmes . recently in the paper of @xcite one stationary type iv burst , observed in 8 - 32 mhz with the uran-2 array , has been analyzed in detail .
the work focused on the interpretation of quasi - periodic oscillations in intensity for the type iv emission as applied to helioseismology .
the preceding review about progress in the study of type iv bursts has shown that the spatial features of type ivs bursts sources in the decameter wavelength range of observations still remain unexplored .
today , due to implementation of novel facilities in space- and ground - based observations , the qualitative analysis of solar events has been improved vastly .
especially , this concerns the success in advanced solar imaging .
in particular , the inner corona is imaged extensively by a few space missions , namely : the extreme ultraviolet imager ( euvi ) on board of the solar terrestrial relations observatory ( stereo ; * ? ? ? * ) , the atmospheric imaging assembly ( aia ) of the solar dynamic observatory ( sdo ; * ? ? ?
* ) , the extreme ultraviolet imaging telescope ( eit ) of the solar and heliospheric observatory ( soho ; * ? ? ?
* ) . to explore the outer corona ,
a number of ground - based radio telescopes developing in recent years has been involved .
the low - frequency antenna arrays focus on imaging and spectral measurements , implemented by the low frequency array ( lofar ; * ? ? ?
* ; * ? ? ?
* ) , the murchison widefield array ( mwa ; * ? ? ?
* ) , and upgraded indian facilities in the gauribidanur radio observatory @xcite . also , the solar - dedicated nanay radio heliograph produces valuable solar images in meter - decimeter range .
the simultaneous examination of both solar spectral and imaging data makes such synergy greatly fruitful . in this connection , it should be pointed out several high - quality papers partly or fully aimed at spatial diagnostics of solar radio bursts , namely : type ii @xcite , type iii @xcite , s - bursts @xcite , and type ivm @xcite .
in fact , the in - depth analysis of spectral and imaging observations of any solar emission is needed for fundamental understanding of observable events .
the approach is essentially important for the study of solar radio bursts generated by different solar processes .
while the two - dimensional radio images allow one to establish the spatial evolution of burst sources in solar corona , the spectrograph measurements in the form of dynamic spectra provide useful information about specific spectral features of the bursts .
since 2010 , using the upgraded the radio heliograph based on the ukrainian t - shaped antenna array , we have obtained spectral records of solar bursts together with their radio imaging data in the frequency range 8 - 33 mhz .
this opens up new possibilities in solar astronomy . in this paper
, we consider the event occurred on 6 september 2014 .
it includes several components : set of type iii bursts , low- and high - frequency type ii bursts , and also low- and high - frequency type iv bursts .
it is interesting that the type iv bursts belong exactly to the class of stationary type iv bursts .
notice also here that we mainly focus on low - frequency imaging measurements , taking into account a close examination of other `` participants '' of this complex event .
the present paper is organized in the following manner .
section [ sect : instrument ] provides an overview of the utr-2 radio heliograph used . in the next section [
sect : event ] we carry out the data analysis through the consideration of each constituent of the observed event by the piece .
the main results are discussed in section [ sect : discussion ] .
conclusions are presented in section [ sect : conclusions ] .
the two - dimensional ( 2d ) radio heliograph of the utr-2 is a radio astronomical instrument for regular low - frequency solar observations within 8 - 33 mhz .
it has been designed with solar imaging programs in mind .
the history of the heliograph development is described thoroughly in @xcite .
the radio heliograph is an updated device for obtaining two - dimensional images of brightness distribution of radio emission from the sun .
although its important design features in full have been introduced in @xcite , below we present a quick overview of the heliograph s architecture as applied to the heliograph imaging technique .
the instrument is based on the utr-2 ( ukrainian t - shaped radio telescope , mode 2 ) antenna system which is located at the s. ya .
braude radio observatory about 70 km from kharkiv in ukraine ( @xmath6 n , @xmath7 e ) , operated by the institute of radio astronomy ( ira ) .
the utr-2 antenna field comprises 2040 fat , therefore broadband , horizontal dipoles divided into three rectangular arrays : north , south , and west .
each of these arrays , in turn , consists of four structurally independent sections .
the north and south arrays make the north - south arm , including eight sections , which are arranged along the meridian , @xmath0 2 km in length .
each of the two arrays includes 120 rows of 6 dipoles .
the short arm , west , perpendicular to the north - south arm , has four sections ( 6 rows of 100 dipoles ) , lined along the parallel , @xmath0 1 km in length .
the horizontal dipoles of all arrays are oriented along the east - west line . along the parallel , they are spaced by 9 m , and 7.5 m along the meridian .
the total area of the antenna field reaches up to @xmath8 m@xmath9 .
this design of the utr-2 antenna array makes the instrument very flexible for radio astronomy observations . to reconfigure it ,
this allows us to change its properties for various scientific goals , including spectral and imaging measurements of the sun .
in particular , the utr-2 antenna pattern can have both one beam and five fan independent beams .
namely , in heliographic measurements we make use of the multi - beam mode realized in the following way .
the signals from the north - south arm are combined to form five knife - shaped ( @xmath10 at 25 mhz ) beams that are multiplied by signals from the west arm to produce five pencil beams .
consequently , the beams at zenith have sizes @xmath11 at half - power width at 25 mhz with spacing from each other on @xmath12 along declination . in the utr-2 arrays the signals of horizontal dipoles are summed and phased with discrete cable delay lines .
moreover , the signals are phased independently in two orthogonal coordinates , @xmath13 and @xmath14 , which are connected to equatorial coordinates with the following relations @xmath15 where @xmath16 and @xmath17 are the hour angle and the declination , respectively , and @xmath18 denotes the geographic latitude of antenna location . it should be pointed out that the heliograph s hardware is embedded in the utr-2 antenna system by connecting additional blocks into the latter .
the most important of them is responsible for changing the five - beam set position along hour angle ( or along _ u _ coordinate ) .
the extra shift module , integrated into the phasing system of west arm , performs the fast scanning of a sky area by five pencil - shaped beams relocating the antenna pattern consecutively over eight discrete positions ( see @xcite for more details ) .
the angular step of the extra shift module along hour angle is taken the same as the separation between beam centers along declination , @xmath19 ( or along _ v _ coordinate ) .
figure [ figure1 ] shows the complete picture of the heliograph field of view at zenith .
it forms a rectangular matrix in _ uv_-plane consisting of 40 elements .
the field of view covers the spatial sector @xmath20 in sky .
the beams of the matrix abut each other at 25 mhz , but they are overlapped for lower frequencies . in this case
the utr-2 tracking system is carried out by superimposition of the field - of - view center with the solar disk center ( see figure [ figure1 ] ) . note that the rectangular raster is preserved only for near - meridian operation . in the other cases , the raster shape may essentially differ from the rectangular one . in the 2d heliograph mode the sun traverses the field of view pointed to an appropriate fixed position in sky , due to diurnal motion of celestial sphere .
the field of view accompanies the sun motion in sky so that after some time it is turned over hour angle to the west forward , waiting for the sun passage through the field of view again .
along declination and hour angle , respectively .
the full width at half maximum ( fwhm ) of the pencil - shaped beams in the zenith direction at 25 mhz is shown by black circles , and the heliocentric distances equal 1 , 2 and 4 solar radii are represented by dashed orange circles .
the separation between beam centroids is equal to @xmath21.,scaledwidth=42.0% ] to observe solar radio emission , the parallel - serial mode can be applied .
then the output signals of all five beams of the north - south and west antennas are simultaneously recorded by five parallel receiver channels .
however , the serial mode requires only one receiver - recorder that significantly reduces the cost of the heliograph equipment , and the output signal from each beam is linked to the digital spectral analyzer one - by - one .
it is that the heliograph configuration was used for our observations on 6 september 2014 . using this mode
, we have to restore each heliograph frame from serial records as a two - dimensional image .
usually the image repetition rate of the utr-2 heliograph is chosen from observation tasks
. the preferable composition rate of heliograms for observations with slow variations of intensity in time ( for example , radio emission of the quiet sun ) is enough to take about 2 minutes , and the utr-2 tracking rate is about some tens of minutes .
the recent results of the utr-2 heliographic observations of solar corona at decameter wavelengths , during low solar activity , have been presented in @xcite . on the other hand , the measurements of high solar activity manifestations such as short - lived bursts and transients , bursts with fine structure and others , demand to increase the image composition rate together with time resolution in records . with reference to the observations on 6 september 2014
the heliographic studies have been fulfilled with recording 1 frame per 3 seconds .
it seems to be sufficient for the reliable tracking of burst s sources in the solar corona from frequency - time records .
the procedure of solar heliographic observations at the utr-2 radio observatory is as follows .
the solar imaging is carried out by the 2d heliograph . to make the observations more informative ,
the output signals from north and/or south telescope arrays are used for producing the multi - frequency spectral data ( dynamic spectra ) .
the representation of solar radio data in the form of a dynamical spectrum ( based on the short - time fourier transform and showing the evolution of radio emission intensity in frequency and time ) is a conventional method for the study of solar bursts .
it allows one to distinguish bursts one from another visually , due to their different time - frequency features , namely frequency drift rate , duration , fine structure and so on .
moreover , the data facilitate the understanding of heliographic records in serial mode to restore properly two - dimensional images of solar burst sources . as a back end
, we make use of the digital receivers dspz specially designed for low - frequency radio astronomical observations ( see details in @xcite ) . in the issue
two sets of dspz are involved .
the first of them serves as a heliograph receiver , and the second gets the spectral data .
the solar event on 6 september 2014 was studied by both space- and ground - based instruments .
they permit us to observe solar phenomena at different stages in detail .
spectral radio observations have been carried out by following radio astronomical facilities ( joint dynamic spectrum in figure [ figure2 ] ) : the orfees spectrograph , which belongs to nanay radio observatory and operates within 140 - 1000 mhz ; the e - callisto spectrograph with three callisto receivers fed by a bicone antenna ( 10 - 100 mhz ) and a log - periodic antenna ( 100 - 200 mhz and 200 - 400 mhz ) at the rosse solar - terrestrial observatory ( http://www.rosseobservatory.ie ) ; the nanay decametric array ( nda ; see * ? ? ?
* ) utilized in the frequency range 10 - 70 mhz ; the utr-2 antenna array designed for observations from 8 mhz up to 33 mhz .
hereafter it is convenient to distinguish between the high - frequency ( hf ) type ii , iv bursts , belonging to the orfees / rsto frequency range , and the low - frequency ( lf ) ones relevant to the nda / utr-2 frequency range .
radio imaging was provided by the nanay radio heliograph ( nrh ; see * ? ? ?
* ) at nine separate frequencies within 150 - 445 mhz and by the utr-2 heliograph in 16.5 - 33 mhz with frequency resolution in 4khz .
notice that the most interesting results of our event were obtained from low - frequency radio imaging measurements . [ cols="^,^,^,^,^",options="header " , ] [ table1 ] on 6 september 2014 the cme onset was detected at 07:45:57 ut by the stereo behind euvi spacecraft .
the observations showed that the cme started as a set of rising loops at the active region ( ar ) core .
the eruption developed above noaa ar 12157 ( s13 e57 ) containing 14 solar spots .
the sunspot group had a complex magnetic structure ( @xmath22 classification ) .
the above - mentioned cme onset coincides in time with the early growing phase of the successive x - ray solar flare of c8.0 class from this ar ( see figure [ x - ray ] ) .
likely , the cme and the flare were connected with the same eruptive process on the sun . at 10:12:00 ut and 10:54:26 ut the eruption was registered in coronagraph observations on both soho and stereo spacecrafts .
according to the soho / lasco c2 , c3 and the stereo cor2b measurements , the sky - plane ( projected ) speeds of the eruption were very close to each other in value : about 414 km s@xmath5 and 390 km s@xmath5 , respectively .
the cme is characterized by the angular width in @xmath0 246 angular degrees , an almost constant linear speed , and acceleration is about + 0.5 km s@xmath4 according to the soho / lasco c2 , c3 observations . moreover ,
the eruption development was accompanied with both coronal wave ( likely , a shock ) formation and coronal structure deflection that will be considered hereinafter . on this day
the manifestation of solar activity in radio was noticed as a large number of hf bursts , first appearing at @xmath0 08:40 ut and ending at @xmath0 09:10 ut .
the bursts are clearly visible in the orfees spectrum .
it was found that the bursts have features typical for type iii bursts .
the bursts are characterized by negative frequency drift rate .
recall that type iii bursts indicate the presence of subrelativistic electron beams propagating along open magnetic field lines in the solar corona towards the interplanetary medium .
however , it should be noted that even positive frequency drift rate of type iii bursts can be explained with outward motions of electron beams from the sun ( see @xcite ) .
type ii radio emission , observed on the dynamic spectrum in the form of two bursts , appeared after the type iii bursts .
we divide type ii radio bursts into lf and hf parts , where the latter shows a harmonic structure . the hf type ii burst is shown in figure [ figure2 ] as well as in figure [ hf_ivs ] with more details .
it consists of two branches ( lanes ) , fundamental ( f ) and harmonic ( h ) components , and the h - f ratio is about 1.9 .
the f and h bands started with @xmath0 09:12 ut at 224 mhz and 425 mhz , respectively . according to the dynamic spectrum on figure [ figure2 ] , the fundamental radiation was restricted abruptly in frequency and time , at 144 mhz and at 09:22 ut , respectively , while the harmonic radiation lasted until 10:15 ut and has come to naught at @xmath0 120 mhz .
the frequency drift rate of the hf type ii radio burst is equal to @xmath0 0.122 mhz s@xmath5 .
the lf type ii burst is represented in the nda dynamic radio spectrum in figure [ figure2 ] .
following the records , the lf type ii burst started just a little before its hf companion .
it originated at 09:01 ut near 67.2 mhz and continued until 09:23 ut at @xmath0 39.4 mhz .
the frequency drift rate of the lf type ii burst is equal to 0.024 mhz s@xmath5 , typical for type ii bursts at meter wavelengths .
it is generally accepted that radio sources of the type ii bursts are shock waves .
the shocks can be produced by cmes , and the electrons , leading to the radio emission , are accelerated usually at nose or / and at flanks of the shocks .
regarding the considered type ii bursts , at that time there were no other suitable agents which could induce shock waves , except for the cme .
therefore , it is probably that hf and lf type ii radiations are generated by shocks produced by the single eruption .
the event of 6 september 2014 was interesting not only in the contribution of two type ii bursts , but also in the presence of continuum radiation , occupying a wide frequency range , the continuum being separated visibly on hf and lf parts .
each of them shows special features representative of stationary type iv radio emission .
it is now believed that the origin , responsible for the stationary long - lasting continuum in radio emission from the sun , are electrons , moving in solar magnetic traps . as both the hf and
lf type ivs bursts emerged nearly from the same area in the solar corona ( see below for details ) , reasonably one assumes that their sources are the same . in the paper of @xcite
it has been proposed that solar structures , such as high coronal arches , lead to low - frequency radio continuum in the outer solar corona .
notice that this concept is supported by the radio imaging measurements that will be presented below . the lf stationary type iv radiation extended from meter to decameter wavelengths . in the decameter range
the radio burst was observed by the 2d radio heliograph of utr-2 as well as other radio telescope arrays .
the north and south arrays of utr-2 were also involved for spectral measurements ( to produce high - sensitivity solar dynamic spectra ) with a high resolution in time and frequency ( @xmath0100 msec and @xmath04 khz , repectively ) within 17.0 - 33.0 mhz .
the most low - frequency part of the radio spectrum is shown in figure [ figure2 ] and was received by the north array .
the decameter radiation has been observed for 3 hours from 09:55 ut till the end of this observation session at 13:00 ut .
the decameter continuum was similar to a steady irregular pattern in dynamic spectra .
the small - scale analysis of its fine structure reveals that the radio emission was composed of scrappy fragments like fiber bursts .
the largest value of radio flux density in the burst was @xmath23 s.f.u .
unfortunately , no polarization measurement is available for the utr-2 radio telescope . at higher frequencies
the radio emission was recorded with nda . from the nda spectrum
it shows that in meter wavelengths the stationary type iv burst started from 09:50 ut , and after about 5 minutes it became visible in decameter wavelengths . toward the end of observations , according to the utr-2 spectral data on 6 september 2014 , the radiation of the stationary type iv continuum almost ceased at low frequencies , but it lasted till @xmath013:20 ut at higher frequencies . due to the abundance of strong radio frequency interferences ( radio noise ) in nda spectra below 30 mhz ,
spectral features of the stationary type iv burst are not clearly recognized , while the high sensitivity and high resolution of utr-2 allow us to find out many peculiar spectral features .
figure [ decam_profiles ] shows a superposition of time profiles at six frequencies within 23 - 33 mhz as well as x - ray goes profile within 0.05 - 0.4 nm . using the data
, one can identify two enhancements of intensity ( humps in radio profiles ) on the background against the decameter continuum .
it should be pointed out that the emergence of two consecutive solar flares of c2.5 and c1.7 classes from noaa ar 12157 were close to the enhancements in time .
the green filled strips in figure [ decam_profiles ] indicate the durations of observed flares , from start to end , and the dashed red line inside the strips is the flare peak instant in time .
there are close matchings between the main tops of x - ray emission and enhancements of intensity in decameter radiation .
this temporal correlation between manifestations in x - ray and in radio implies a physical relationship in radiation sources .
if one assumes that the decameter emission source is connected with a coronal loop to the ar , then the radio enhancements can be explained in terms of injections of flare - induced energetic electron beams in the loop . unlike the stationary lf type iv burst
, the analysis of hf type iv radio emission requires more effort .
the point is that the orfees instrument used now has a poor sensitivity .
figure [ hf_ivs ] displays a high - frequency part ( 144 - 450 mhz ) of the dynamic spectrum , obtained with orfees .
this spectral representation has been combined from many consecutive 30-min records into one .
to highlight useful spectral features , the raw data need the preliminary processing of background subtraction .
as the background is removed from dynamic spectra , they become satisfactory for examination as presented in figure [ hf_ivs ] .
it contains evidently the hf type ii burst with harmonic structure ( marked by dashed lines in blue ) as well as radio continuum beginning at @xmath009:50 ut and lasting about 4 hours .
the radio emission takes the form of bright filament patterns distributed in a wide frequency range ( 144 - 400 mhz ) .
such features indicate the presence of broadband continuum radiation , and the filament patterns are the most intensive parts of radio emission , received by the orfees spectrograph , rather than the hf type iv continuum entirely . because of the lack of better spectral data from other radio spectrographs in overlapping frequency bands , it is difficult to get a better representation of the radio emission . therefore , one of possible ways to confirm the existence of continuum radiation is to find out its source , using any heliographic observation .
the nrh maps clearly show a steady radio source in the frequency range from 150 mhz to 327 mhz .
the weak presence of radio source is observed also at 408 mhz .
based on these measurements we have determined brightness temperatures of the burst at several frequencies ( 173 mhz , 228 mhz , 270 mhz , 327 mhz and 408 mhz ) available in measurements .
they were superimposed on the orffes dynamic spectrum . in this case
the characteristic values of brightness temperatures reached up to @xmath24 k that is an actual manifestation of the radio source .
it should be noticed that the spectral and imaging studies are consistent with each other .
this is not surprising , as their data have a common physical origin .
taking into account these results , the recorded radiation can be attributed to stationary type iv bursts .
it is interesting to examine whether spectral humps , found in the lf type ivs radio emission , exist in the hf one . because of a poor sensitivity , a survey of the orfees radio profiles does not reveal the presence of intensity enhancements . on the other hand ,
the nrh temperature profiles show oscillations that are too strong to reveal the enhancements .
another possible reason for this lack of similar spectral humps in the hf type ivs burst is that the two c - class flares were too weak to produce observable signature , which was found in the lf utr-2 records only due to the radio telescope s high sensitivity .
the above analysis , which has been done for lf and hf type ivs bursts , concerned only their spectral properties .
the most important fact which deserves our attention is that following the records of different instruments , the lf and hf type ivs bursts were detected in good agreement with each other in time . according to the nda spectral data , the lf type ivs burst started at @xmath0 09:50 ut , and a little later
the radio emission was received by utr-2 . to determine the start time for the hf type ivs radiation , let us turn to heliographic measurements in view of unclear hf spectral data .
the hf type ivs source has been observable since 09:48 ut at frequencies from 327 mhz to 173 mhz , and then at 150 mhz in the nrh records ( see figure [ nrh_still_frame ] and the accompanying movie m1 ) . on the movie , which covers 09:40@xmath2514:00 ut time range ,
the solar radio emission is seen as two bright regions .
this can be explained by the existence of an earlier distinct radio source ( the first of these regions ) before 09:48 ut .
next , this emission intensity decreased gradually beginning with high frequencies , and almost at once another source , corresponding to the hf type iv emission , appeared . in the consecutive nrh images this is observed as a jump in source locations .
figure [ nrh_still_frame ] represents a still frame of the movie m1 taken at 11:24:02 ut and shows the presence of the radio source corresponding to hf type ivs radio emission .
consequently , the first confident detection time for the hf type ivs burst ( at @xmath0 09:48 ut ) was very close to the lf type ivs burst emergence ( at @xmath0 09:50 ut ) in spectral data .
furthermore , according to the nrh data after 13:10 ut there is a decline in brightness temperature profiles .
the nrh maps show that the hf type ivs source disappeared completely towards @xmath0 13:40 ut at passbands from 327 to 228 mhz , while at 150 mhz and at 173 mhz the brightness temperature decreased by a factor of three in comparison with its value at 13:10 ut .
this indicates that in the nda dynamic spectrum the lf type ivs burst emission came to the end near 13:40 ut . taking into account all pieces of above - stated evidence , we may assume that lf type ivs and hf type ivs bursts had the same origin . to investigate spatial features of corresponding radio sources , we consider the radio imaging measurements provided by the nrh and the utr-2 radio heliograph .
the example of utr-2 radio heliograms is shown in figure [ helio_cube ] . in this case
the data were collected in a so - called three - dimensional ( 3d ) cube , along two spatial coordinates and at frequency . using appropriate electron density models
, one can find out heights of corresponding emitting layers in the solar corona . in figure [ helio_cube ]
such 3d brightness distribution of the lf type ivs burst within 18.5 - 31.0 mhz at 12:00:24 ut has been divided into 50 frequency channels . although the total number of utr-2 heliograms obtained in the observations from 09:55 ut to 13:00 ut within 16.5 - 33.0 mhz ( 4096 spectral channels ) with 3-sec cadence is much larger ( about @xmath1 ) , there is no necessity to account for all the abundance of data . to observe changes of the source position in solar corona
, it suffices to take fifty frequencies in each 3-sec frame .
next , we are going to compare the utr-2 heliograms with the corresponding nrh images . recall that the latter heliographic data are represented in several frequencies .
therefore , we reduce again our amount of utr-2 frequency channels , used in the analysis , to 3 frequencies ( 20 mhz , 25 mhz and 30 mhz ) in each 3d cube .
figure [ imaging ] shows a set of running difference images from soho / lasco c2 and direct sdo / aia 171 observations .
the pictures are formed by the composition of the ground - based radio imaging data and the solar spacecraft records .
the consecutive images should be considered from top to bottom , row by row , from left to right ( each image has the alphabetic character itself ) .
the legend of pictures gives the list of frequencies at which the imaging data were produced .
the intensity maxima on utr-2 heliograms and on nrh maps are marked by special signs to show their relative positions .
the error bars of nrh maxima locations are less than the width of those signs .
as regards to utr-2 radio imaging , the uncertainty of measurements of the lf radio source is determined by the beam size of the antenna pattern .
figure [ imaging](l ) shows the utr-2 half - power beam - width ( hpbw ) with angular sizes [email protected]@[email protected] , [email protected]@[email protected] , [email protected]@[email protected] at 20 mhz , 25 mhz and 30 mhz correspondingly , taken at 12:00:43 ut .
the antenna beam looks like ellipse elongated along _ v_-coordinate .
the angular sizes of hpbw during utr-2 observational session kept almost steady with slight variations .
the hpbw mean values were [email protected]@[email protected] ( at 20 mhz ) , [email protected]@[email protected] ( at 25 mhz ) , [email protected]@[email protected] ( at 30 mhz ) .
the imaging measurements , collected in figure [ imaging ] , show that the radio source locations changed with the cme propagation from the sun .
the most obvious feature of this figure is that the hf sources and the lf sources are in good alignment with each other above the cme - associated ar , despite some southward deflection of the lf parts at certain instants ( see panels ( a ) , ( b ) , ( e - h ) , and ( n - q ) ) .
this is consistent with our presumption that the hf - lf sources are physically related .
initially , both the hf and lf sources were located above the cme - associated ar , clearly shown in panels ( a ) and ( b ) of figure [ imaging ] . the intensity maxima of the lf radio source at panels ( c ) , ( d ) , and partially at ( b ) for 30 mhz maximum , moved sunward to the cme body .
figure [ imaging](e ) shows a very similar radio emitting structure as in figure [ imaging](a ) .
panels ( f)-(i ) of figure [ imaging ] present an ascent of the lf source maxima as well as gradual change of emitting structure from downward - arched shape to line - elongated one within the sun equatorial region .
the latter configuration lasted for @xmath040 minutes as seen from panels ( i)-(m ) of the figure .
then , the positions of the lf source maxima underwent descent in 30 - 25 - 20 mhz sequence .
starting from figure [ imaging](n ) to figure [ imaging](q ) the lf source positions at different heights were moving from east towards south - east .
it should be noticed that the locations of maxima in intensity for the hf and lf type ivs bursts at different frequencies exhibit very similar behavior in time .
the accompanying movie m2 includes the panels of figure [ imaging ] showing the dynamical evolution of both the cme and the radio - emitting structure .
the lf type ivs radio emission was accompanied by hf continuum .
the presence of the latter has been justified confidently by nrh maps and movie . the nrh data allowed to obtain accurate timing of hf type ivs burst .
there are several convincing arguments to believe that the bursts belong to single broadband stationary type iv radio emission .
the absence of transitional emission between lf and hf parts in spectral data ( see figure [ figure2 ] ) can be explained by the following causes .
firstly , this radio emission is not observed on rsto dynamic spectrum ( 100 - 144 mhz ) because of a poor sensitivity of this instrument .
secondly , there is a blank of data within 80 - 100 mhz due to lack of radio measurements .
in addition , the nda dynamic spectrum shows a visible sharp edge of lf radio emission near 65 mhz without the expected prolongation up to upper boundary .
very likely , it is connected with specific gain - frequency pattern of nda .
it is known that irregularity of the frequency response of an antenna results in attenuation of receiving signals at the lower and upper boundaries of working frequency band .
thus , the radio emission received by nda above 65 mhz possibly was depressed by such antenna effect . in support
that lf and hf type iv bursts had the same origin , one should realize there is modest possibility that two type ivs bursts could exist quite separately at the same time , in the same spatial quadrant , as well as showing similar dynamical evolution of radio emitting structure in imaging measurements .
the intensity variations , found in decameter part of broadband type ivs continuum , are possibly affected by the amount of energetic electrons as well as the evolution of their velocity distribution function inside the radio emitting structure . following figure [ decam_profiles ] two humps in radio range appear when two c - class flares occur . probably , during the flares new injection of energetic electrons into the emitting structure took place .
as a result , we observe two apparent enhancements in the radio profiles .
these enhancements were detected only in utr-2 data but they are not recorded by other instruments .
the main reason is that utr-2 antenna array is the most sensitive instrument among those used here .
based on observations in radio range it is commonly accepted that type ivs bursts show up in several minutes after flares . however , the present case does not support this general statement . according to common dynamic spectrum ( figure [ figure2 ] )
the stationary type iv radio burst started near 09:50 ut . the x - ray c8.0 class solar flare had peak time at 08:14 ut .
after then the prolonged declining phase , preceding the type ivs radio emission , took place .
the delay between the flare peak time and the start time of the type iv burst is about 2 hours . from this point
a role of the flare in initiation of type iv burst seems to be indirect .
it should be noted that there is another feasible scenario for stationary type iv bursts development which relates to cmes .
in contrast to flares , which show relatively short characteristic times , a cme is a long - duration process .
such eruptions also involve processes that are able to accelerate electrons accounting for the type iv emission as well .
for instance , it can drive a shock ahead if the eruption is fast enough , or trigger magnetic reconnection along the post - cme current sheet or else along certain magnetic separatrixes within or surrounding the eruptive structure .
we suggest that this scenario also plays a role here
. very likely , the cme process may induce further acceleration and injection of energetic electrons into the type iv emitting structure ( presumably a loop structure ) , in addition to those caused by the c8.0 flare .
note that the c8.0 flare , having an 1.5-hour - long declining phase , may continuously inject energetic electrons into the overlying magnetic structure .
tentatively , we assume the hf and lf type ivs sources emitted from a single high - lying loop with one foot located around the ar from which the cme emerged
. the loop can be delineated by connecting the hf and lf sources as shown in figure [ imaging ] .
the nrh measurements reveal a movement of the hf radio emission source inside the magnetic loop close to its base .
the utr-2 heliographic images show very similar features , but only higher , in the outer corona . according to utr-2 radio images the uttermost altitudes for 20 mhz , 25 mhz
, 30 mhz emission layers were around 4.5@xmath28 , 3.9@xmath28 , and 3.0@xmath28 , correspondingly .
the declined southward deflection , the ascent - and - descent motion may indicate the response of the presumed loop structure to the cme disturbance , including the dynamical displacement of the structure and the changing inner density or turbulence levels which may affect the emitting intensities and frequencies .
in the present work the heliographic observations of stationary type iv radio emission at the lowest frequencies ( close to ionospheric cut - off ) have been presented in detail .
the main research includes solar spectral and imaging measurements carried out by the utr-2 radio telescope in the range 16.5 - 33.0 mhz .
the spectral analysis of the type ivs radio emission was performed together with radio data from higher frequencies observed by nda ( 33.0 - 80.0 mhz ) , e - callisto ( 100 - 144 mhz ) and orfees ( 144 - 450 mhz ) .
all this allows us to consider many aspects of the complex event on 6 september 2014 , especially the spatial characteristics and dynamical evolution of radio sources of the type ivs radio bursts .
it is found that the type ivs bursts have two components , the hf part and the lf part .
the lf decameter type ivs continuum had a double humped form of radio intensity profiles , correlated in time with the occurrence of two minor solar flares from the cme - associated ar .
it is also found that the starting and ending times of the hf and lf type ivs bursts are considerably close to each other .
in addition , from the nrh and utr-2 imaging measurements , the radio sources of both components show spatially correlated motion along with the cme eruption . on the basis of these observations
, we assume that the hf and lf sources are connected by a high - lying magnetic loop structure , which has one root located at the ar , thus allowing energetic electrons accelerated during flares to be injected into the loop and somehow trapped there .
we suggest that the dynamical motion of radio sources manifests the loop response to the cme disturbance .
this research was supported by grants nsbrsf 2012cb825601 , nnsfc 41274175 , 41331068 , as well as by research grant 0115u004085 from the national academy of sciences of ukraine .
the authors are grateful to goes , sdo , soho , stereo , rsto , and nanay teams for their open access data policy . | first - of - its - kind radio imaging of decameter solar stationary type iv radio burst has been presented in this paper . on 6 september 2014
the observations of type iv burst radio emission have been carried out with the two - dimensional heliograph based on the ukrainian t - shaped radio telescope ( utr-2 ) together with other telescope arrays .
starting at @xmath009:55 ut and throughout @xmath03 hours , the radio emission was kept within the observational session of utr-2 .
the interesting observation covered the full evolution of this burst , `` from birth to death '' . during the event lifetime , two c - class solar x - ray flares with peak times 11:29 ut and 12:24 ut took place . the time profile of this burst in radio
has a double - humped shape that can be explained by injection of energetic electrons , accelerated by the two flares , into the burst source . according to the heliographic observations we suggest the burst source
was confined within a high coronal loop , which was a part of a relatively slow coronal mass ejection .
the latter has been developed for several hours before the onset of the event . through analyzing about @xmath1 of heliograms ( 3700 temporal frames with 4096 images in each frame that correspond to the number of frequency channels )
the radio burst source imaging shows a fascinating dynamical evolution . both space - based ( goes , sdo , soho , stereo )
data and various ground - based instrumentation ( orfees , nda , rsto , nrh ) records have been used for this study .
= 1 |
the intensive investigations of the cr@xmath1-doped laser media testify about their high potential in the tunable generation in mid - ir between 2 5 @xmath2 m .
a special interest is connected with cr@xmath1 : znse active medium due to its excellent material properties : high thermal conductivity which is close to that one of yag , the absence of excited state absorption , and very broad gain band which allows for tunable generation in 2 3 @xmath2 m range .
the latter is very attractive for a lot of applications : remote sensing , spectroscopy , ophthalmology , and neurosurgery . at this moment , the pulsed@xmath3 and continuous - wave operation@xmath4 in cr@xmath5 : znse lasers was demonstrated with ingaas and tm : yalo laser pumping , which demonstrated high efficiency and favorable lasing abilities of this active crystal . a very interesting property of cr@xmath1 : znse is the high coefficient of the nonlinear refraction@xmath6 that results from the relatively small bandgap , i. e. , in fact , from the semiconductor nature of the material .
as it is known , the higher nonlinear refraction is favorable to self - starting of kerr - lens mode locking@xmath7 .
the decrease of kerr - lens mode locking threshold allows to simplify the laser configuration , to improve the ultrashort pulse stability , and to increase the lasing efficiency .
another advantage of cr@xmath1 : znse as media for ultrashort pulse laser is a very broad gain band , which can support the generation of 12 fs pulses .
however , high nonlinear refraction coefficient is the source of some shortcomings .
as it was shown@xmath8 , the kerr - lens mode - locked lasers demonstrate a variety of unstable regimes that can destroy the ultrashort pulse and prevent from the pulse shortening . here , for first time to our knowledge
we analyze the kerr - lens mode locking abilities of cr@xmath1 : znse laser .
the analysis of znse nonlinear properties is presented that is necessary for an optimization of the laser design and for investigation of self - starting ability and ultrashort pulse stability in cr@xmath1 : znse laser .
as it is known@xmath6 , the two - photon absorption , raman scattering and stark effect strongly contribute to nonlinear refractivity in semiconductors .
the main parameter , which defines a value of nonlinear refraction coefficient @xmath9 , is the bandgap energy @xmath10 : @xmath11=\frac{k\sqrt{e_{p}}g_{2}(\frac{\hbar \omega } { e_{g}})}{% n_{0}e_{g}^{4}}.\ ] ] here @xmath12 is the linear coefficient of refraction ; @xmath13 for the most of direct gap semiconductors ; @xmath14 is a material - independent constant ; @xmath15 is the function depending only on the ratio of photon energy and the energy gap of the material . for znse @xmath16 .
for @xmath15 we used the following approximation in the below bandgap region : @xmath17which was obtained from experimental data presented in ref@xmath7 .
in addition to dependence of @xmath15 on wavelength , we have to take into account the dispersion of @xmath12 . for znse
the next approximation is valid : @xmath18(here @xmath19 is in meters ) . a resulting dependence of @xmath20 on wavelength is shown in fig .
1 . there is a strong kerr nonlinearity producing self - focusing effect .
the value of @xmath20 exceeds the typical values of the most of dielectric crystals by two orders of magnitude .
however , this is not sufficient for the estimation of the nonlinear properties of znse . despite the fact that znse has cubical lattice , there is not center of inversion in crystal . as result , the medium possesses the second - order nonlinear properties . the corresponding nonlinear coefficient @xmath21 .
although there is no efficient frequency conversion , the existence of second - order nonlinearity can strongly contributes to the nonlinear refractivity due to so - called cascaded second - order nonlinearity , which can be used for kerr - lens mode locking@xmath22 .
usually , change of the index of refractivity has non - kerr character , but the kerr - like approximation @xmath23 ( where @xmath24 is the field intensity ) is valid if@xmath25 @xmath26 , where @xmath27 ; @xmath28 ; @xmath29 , @xmath30 are the fundamental frequency and wave number , correspondingly ; @xmath31 , @xmath32 are the linear refractive index and wave number for the second harmonic .
our calculations showed , that the kerr approximation for znse is valid up to intensities of @xmath33 at 2 @xmath2 m and @xmath34 at 3 @xmath2 m , and it is used throughout present work .
the nonlinear coefficient of refraction due to cascading second - order nonlinearity is@xmath25 : @xmath35 note , that in the case of the normal dispersion of @xmath12 the nonlinearity due to second - order nonlinearity has a defocusing nature .
the dependence of @xmath9 on the wavelength is shown in fig .
one can see , that the nonlinear refraction is very strong . as result of the joint contribution of third- and second - order nonlinearities ,
a net - coefficient of nonlinear refraction is shown in fig .
3 . the nonlinearity is focusing up to 1.2 @xmath2 m and defocusing in mid - ir range .
an experimental value of @xmath20 measured at 1.06 @xmath2 m demonstrates an excellent agreement with the corresponding value of calculated curve in fig .
the obtained value of @xmath20 enables us to estimate a mode locking efficiency due to kerr - lensing .
we have used the method presented in @xmath36 : for each fixed value of the focal length of the mirror @xmath37 ( see fig .
4 ) we varied a folding distance and the position of active medium trying to maximize the value of @xmath38 , where @xmath39 is the effective transmission of the hard aperture .
this parameter defines the inverse saturation intensity of the efficient fast saturable absorber , induced by joint action of kerr self - focusing in active medium and difractional losses at aperture . for the common kerr - lens mode - locked lasers @xmath40 . in the case of znse
the modulation parameter is shown in fig .
it is seen , that the kerr - lensing efficiency in this case is higher by two orders of magnitude .
it is interestingly , that the defocusing nature of the nonlinearity does not affect the process of the cavity optimization essentially : the change of the sign of @xmath20 only slightly affects the position of active medium and folding distance .
the optimized cavity configuration is shown in fig .
calculated values of @xmath38 suggest a good kerr - lens mode locking ability of cr@xmath1 : znse laser . however , the strong self - phase modulation while being the positive factor facilitating self - focusing , can , on the other hand , destabilize an ultrashort pulse generation due to automodulational instability@xmath8 . in order to estimate the mode locking stability in cr@xmath1 : znse laser we performed the numerical simulation on the basis of fluctuation model and distributed scheme of the laser generation .
the master equation was : @xmath41 where @xmath42 is the field , @xmath43 is the longitudinal coordinate , @xmath44 is the saturated by full field energy gain coefficient , @xmath45 is the inverse bandwidth of the spectral filter , @xmath46 is the net - group delay dispersion , @xmath47 is the self - phase modulation parameter , @xmath48 is the active medium length , @xmath49 is the diffraction loss , defined from configuration of laser , @xmath50 = @xmath51 . the strong self - phase modulation and self - focusing dominate over the gain saturation .
the last factor provides the negative feedback in laser that stabilizes ultrashort pulse .
the critical parameter defining the ultrashort pulse dynamics is @xmath52 , where @xmath53 is the gain saturation fluency .
the decrease of this parameter worsens the pulse stability . to comparison , in znse @xmath54
is ten times lower than in ti : sapphire .
therefore , to obtain the stable pulse generation , we have to increase @xmath54 e.g. by increasing @xmath45 .
the stable pulses were obtained for @xmath55 .
the pulse duration was about 300 fs for 800 mw absorbed pump power .
an additional stabilization factor is the minimization ( down to zero ) of net - dispersion .
the character of the pulse destabilization needs some comments .
the typical scenarios are shown in figs .
the pulse behaviors presented here were calculated for the same set of the laser parameters but for different initial noise samples .
there are the possibilities for multipulse generation ( fig .
6 ) , generation of the breezer - like pulse and its subsequent disintegration ( fig .
7 ) , and the collapse of the pulse into several pulses ( fig . 8) .
note , that breezer - like pulse is formed without action of the soliton mechanism in laser because of net - dispersion in this case is close to @xmath56 .
the regime depicted in fig .
8 is characterized by low - frequency modulation of the generation spectrum that at last leads to the pulse collapse .
it should be noted , that the generation of the stable single pulse is depended on the random noise sample , too .
the analysis of the nonlinear properties of znse
crystal was performed .
it was found , that the strong second - order cascading nonlinearity in the combination with third - order phase nonlinearity causes a strong focusing kerr nonlinearity in near - ir and defocusing quasi - kerr nonlinearity in mid - ir ranges . as result ,
the efficiency of the kerr - lensing in active medium is two - three orders of magnitude higher than in typical active media of femtosecond lasers
. the optimal cavity configuration for kerr - lens mode locking was found .
but , as the analysis testifies , the main problem of the ultrashort pulse generation in cr@xmath1 : znse laser is the pulse destabilization due to strong self - phase modulation .
this requires to decrease the bandwidth of spectral filter .
the last allows to generate 300 fs pulses .
the self - start of kerr - lens mode locking has a statistical character , that is depended on the noise sample .
the investigation of the pulse destabilization testifies about formation of the breezer - like pulses and pulse collapse due to low frequency perturbations .
r.h . page , j.a .
skidmore , k.i . schaffers et al . , trends in optics and photonics , ( osa , washington , 1997 ) .
g. j. wagner , t. j. carring , r. h. page et al .
lett . , 24 , 19 ( 1999 ) .
m. sheik - bahae , d. c. hutchings , d. j. hagan , and e. w. van stryland , ieee j. quantum electr . ,
27 , 1296 ( 1991 ) .
+ 4 . c. radzewicz , g. w. pearson , and j. s. krasinski , optics .
commun . , 102 , 464 ( 1993 ) .
j. jasapara , v. l. kalashnikov , et al .
am . , b 17 , 319 ( 2000 ) .
r. l. sutherland , handbook of nonlinear optics ( ny , 1996 ) .
g. cerullo , s. de silvestry , a. monguzzi et al . ,
letts . , 20 , 746 ( 1995 ) .
v. l. kalashnikov , v. p. kalosha et al . , j. opt .
am . , b 14 , 964 ( 1997 ) . | the theoretical calculation for nonlinear refractive index _
n@xmath0 _ in cr@xmath1 : znse - active medium predicts the strong defocusing cascaded second - order nonlinearity within 2 - 3 @xmath2 m spectral range .
on the basis of this result the optimal cavity configuration for kerr - lens mode locking is proposed that allows to achieve a sub-100 fs pulse duration .
the numerical simulations testify about strong destabilizing processes in the laser resulting from a strong self - phase modulation .
the stabilization of the ultrashort pulse generation is possible due to spectral filtering that increases the pulse duration up to 300 fs
. international laser center , 65 skorina ave .
, bldg .
17 , minsk , 220027 belarus , tel./fax : ( 375 - 0172 ) 326 - 286 , e - mail : [email protected] , http://www.geocities.com/optomaplev |
in the past few years a match among the study of systems defined on lattices by means of statistical mechanics @xcite and the study of networks by means of graph theory @xcite gave origin to very interesting models as the small world magnets @xcite@xcite or the scale free networks @xcite .
however a complete analysis starting from the simple fully connected mean field ising model @xcite@xcite up to these recent complex models @xcite is still not complete ( even thought several important steps have been obtained , examples being @xcite@xcite@xcite ) and important models have not yet been taken into account . among these a certain role is played by the mean field diluted ising model : an erdos - renyi networks @xcite which has spins as nodes and their interactions as links .
the interactions are encoded in a matrix connecting pairs of spins which , when the connectivity allows the link to be present , shares the same value for all the couples . despite their easy formalization @xcite ,
diluted ferromagnets are poorly investigated by rigorous tools @xcite .
inspired by a recent work on these systems in which the authors presented a detailed analysis of the ergodic region and the zero temperature line @xcite we extend recent techniques developed in a series of papers @xcite@xcite@xcite@xcite to this model with the aim of analyzing its critical behavior .
we systematically develop the interpolating cavity field method @xcite and use it to sketch the derivation of a free energy expansion : the higher the order of the expansion , the deeper we could go beyond the ergodic region . within this framework
we perform a detailed analysis of the scaling of magnetization ( and susceptibility ) at the critical line .
the critical exponents turn out to be the classical ones . at the end
we perform extensive monte carlo ( mc ) simulations for different graph sizes and bond concentrations and we compare results with theory . indeed , also numerically
, we provide evidence that the universality class of the diluted ising model is independent of dilution .
in fact the critical exponents we measured are consistent with those pertaining to the curie - weiss model , in agreement with analytical results .
the critical line is also well reproduced .
the paper is organized as follows : in section ( [ laa ] ) we describe the model , in section ( [ cavatappi ] ) we introduce the cavity field technique , which constitutes the framework we are going to use in section ( [ lac ] ) to investigate the free energy of the system at general values of temperature and dilution . section ( [ critico ] ) deals with the criticality of the model ; there we find the critical line and the critical behavior of the main order parameter , i.e. magnetization , we provide its self - averaging and work out a picture by which we explain the breaking of the ergodicity . section ( [ numerics ] ) is devoted to numerical investigations , especially focused on criticality .
finally , section ( [ conclusions ] ) is left for outlook and conclusions .
given @xmath0 points and families @xmath1 of i.i.d random variables uniformly distributed on these points , the ( random ) hamiltonian of the diluted curie - weiss model is defined on ising @xmath0-spin configurations @xmath2 through @xmath3 where @xmath4 is a poisson random variable with mean @xmath5 and @xmath6 is the connectivity .
the expectation with respect to all the ( _ quenched _ ) random variables defined so far will be denoted by @xmath7 , while the gibbs expectation at inverse temperature @xmath8 with respect to this hamiltonian will be denoted by @xmath9 , and depends clearly on @xmath10 and @xmath8 .
we also define @xmath11 .
the pressure , i.e. minus @xmath8 times the free energy , is by definition @xmath12 where we implicitly introduced the partition function @xmath13 too .
when we omit the dependence on @xmath0 we mean to have taken the thermodynamic limit which we assume to exist for all the observables we deal with , in particular for the free energy @xcite@xcite@xcite ( however we will look for a more firm ground on this point by numerical investigation in sec.([numerics ] ) ) .
the quantities encoding the thermodynamic properties of the model are the overlaps , which are defined on several configurations ( _ replicas _ ) @xmath14 by @xmath15 particular attention must be payed on @xmath16 which is called _
magnetization_. when dealing with several replicas , the gibbs measure is simply the product measure , with the same realization of the quenched variables , but the expectation @xmath7 destroys the factorization .
sometimes for the sake of simplicity we will call @xmath17 .
in this section at first we introduce the cavity field technique on the line of @xcite by expressing the hamiltonian of a system made of @xmath18 spins through the hamiltonian of n spins by scaling the connectivity degree @xmath10 and neglecting vanishing terms in n as follows h_n+1 ( ) = -_=1^p_(n+1 ) _ i__j _ ~-_=1^p _ n _ i__j _ - _ = 1^p_2 _ i__n+1 such that we can use the more compact expression @xmath19 with [ decompo ] = , _ n ( ) = - _ = 1^p_2 _
i_. so we see that we can express the hamiltonian of @xmath18 particles via the one of @xmath0 particles , paying two prices : the first is a rescaling in the connectivity ( vanishing in the thermodynamic limit ) , the second being an added term , which will be encoded , at the level of the thermodynamics , by a suitably cavity function as follows : let us introduce an interpolating parameter @xmath20 $ ] and the cavity function @xmath21 given by @xmath22 = \lim_{n\rightarrow \infty } \mathbb{e}\big[\ln \frac{z_{n , t}(\tilde{\alpha},\beta ) } { z_{n}(\tilde{\alpha},\beta)}\big].\end{aligned}\ ] ] the three terms appearing in the decomposition ( [ hsplit ] ) give rise to the structure of the following theorem : [ main ] in the @xmath23 limit , the free energy per spin is allowed to assume the following representation a ( , ) = 2 - + ( , ; t=1 ) * proof * consider the @xmath18 spin partition function @xmath24 and let us split it as suggested by eq .
( @xmath25 @xmath26 where the factor two appears because of the sum over the hidden @xmath27 variable . defining a perturbed boltzmann state @xmath28 ( and its replica product @xmath29 ) as @xmath30 where the tilde takes into account the shift in the connectivity @xmath31 and multiplying and dividing the r.h.s . of eq.([zecca ] ) by @xmath32 , we obtain @xmath33 taking now the logarithm of both sides of eq.([dinox ] ) , applying the average @xmath7 and subtracting the quantity @xmath34 $ ] , we get @xmath35 - \mathbb{e}[\ln z_{n+1}(\tilde{\alpha},\beta ) ] = \ln2 + \mathbb{e}\big[\ln \frac{z_n(\tilde{\alpha},\beta)}{z_{n+1}(\tilde{\alpha},\beta)}\big ] + \psi_n(\tilde{\alpha},\beta;t=1)\ ] ] in the large @xmath0 limit the l.h.s . of eq.([latte ] ) becomes @xmath36 - \mathbb{e}[\ln z_{n+1}(\tilde{\alpha},\beta ) ] & = & \\ \nonumber ( \alpha - \tilde{\alpha } ) \frac{\partial}{\partial\alpha}\mathbb{e}[\ln z_{n+1}(\alpha,\beta ) ] & = & \alpha \frac{1}{n+1}\frac{\partial}{\partial\alpha } \mathbb[\ln z_{n+1}(\alpha,\beta ) ] = \alpha \frac{\partial a_{n+1}(\alpha,\beta ) } { \partial\alpha}\end{aligned}\ ] ] then by considering the thermodynamic limit the thesis follows .
( actually we still do not have a complete proof of the existence of the thermodynamic limit but we will provide strong numerical evidences in section [ numerics ] ) @xmath37 hence , we can express the free energy via the energy and the cavity function .
while it is well known how to deal with the energy @xcite , the same can not be stated for the cavity function , and we want to develop its expansion via suitably chosen overlap monomials in a spirit close to the stochastic stability @xcite@xcite@xcite , such that , at the end , we will not have the analytical solution for the free energy in the whole @xmath38 plane , but we will manage its expansion close ( immediately below ) to the critical line . to see how the machinery works ,
let us start by giving some definitions and proving some simple theorems : we define the t - dependent boltzmann state @xmath39 as @xmath40 where @xmath41 extends the classical partition function in the same spirit of the numerator of eq.([dante ] ) .
as we will often deal with several overlap monomials let us divide them among two big categories : we can split the class of monomials of the order parameters in two families : * we define _ filled _ or equivalently _ stochastically stable _ all the overlap monomials built by an even number of the same replicas ( i.e. @xmath42 , @xmath43 , @xmath44 ) .
* we define _ fillable _ or equivalently _ saturable _ all the overlap monomials which are not stochastically stable ( i.e. @xmath45 , @xmath46 , @xmath47 ) we are going to show three theorems that will play a guiding role for our expansion : as this approach has been deeply developed in similar contexts ( as fully connected ising model @xcite or fully connected spin glasses @xcite or diluted spin glasses @xcite , which are the boundaries of the model of this paper ) we will not show all the details of the proof , but we sketch them as they are really intuitive . the interested reader can deepen this point by looking at the original works .
[ ciccia ] for large @xmath0 , setting @xmath48 we have _
n , t(_i_1_i_2 ... _i_n ) = _
n+1(_i_1_i_2 ... _i_n_n+1^n ) + o()such that in the thermodynamic limit , if @xmath48 , the boltzmann average of a fillable multi - overlap monomial turns out to be the boltzmann average of the corresponding filled multi - overlap monomial .
[ saturabili ] let @xmath49 be a fillable monomial of the overlaps ( this means that there exists a multi - overlap @xmath50 such that @xmath51 is filled ) .
n_t1 q_2n _ t = q_2nq_2n ( examples : for @xmath52 we get @xmath53 ) [ saturi ] in the @xmath54 limit the averages @xmath55 of the filled monomials are t - independent in @xmath8 average .
* proof * in this sketch we are going to show how to get theorem ( [ ciccia ] ) in some details ; it automatically has as a corollary theorem ( [ saturabili ] ) which ultimately gives , as a simple consequence when applied on filled monomials , theorem ( [ saturi ] ) .
let us assume for a generic overlap correlation function @xmath56 , of @xmath57 replicas , the following representation @xmath58 where @xmath59 labels the replicas , the internal product takes into account the spins ( labeled by @xmath60 ) which contribute to the _ a_-part of the overlap @xmath61 and runs to the number of time that the replica @xmath59 appears in @xmath56 , the external product takes into account all the contributions of the internal one and the @xmath62 factor fixes the constraints among different replicas in @xmath56 ; so , for example , @xmath63 can be decomposed in this form noting that @xmath64 , @xmath65 , @xmath66 , where the @xmath67 functions fixes the links between replicas @xmath68 and @xmath69 .
the averaged overlap correlation function is @xmath70 now if @xmath56 is a fillable polynomial , and we evaluate it at @xmath48 , let us decompose it , using the factorization of the @xmath71 state on different replica , as @xmath72 where @xmath73 stands for the number of the unfilled replicas inside the expression of @xmath56 .
so we split the measure @xmath9 into two different subset @xmath74 and @xmath75 : in this way the replica belonging to the @xmath76 subset are always in even number , while the ones in the @xmath59 subset are always odds .
applying the gauge @xmath77 the even measure is unaffected by this transformation @xmath78 while the odd measure takes a @xmath27 inside the boltzmann measure .
@xmath79 at the end we can replace in the last expression the subindex @xmath18 of @xmath27 by @xmath80 for any @xmath81 and multiply by one as @xmath82 . up to orders @xmath83 , which go to zero in the thermodynamic limit , we have the proof .
@xmath37 it is now immediate to understand that the effect of theorem ( [ ciccia ] ) on a fillable overlap monomial is to multiply it by its missing part to be filled ( theorem [ saturabili ] ) , while it has no effect if the overlap monomial is already filled ( theorem [ saturi ] ) because of the ising spins ( i.e. @xmath84 ) .
now the plan is as follows : we calculate the @xmath85-streaming of the @xmath86 function in order to derive it and then integrate it back once we have been able to express it as an expansion in power series of @xmath85 with stochastically stable overlaps as coefficients . at the end
we free the perturbed boltzmann measure by setting @xmath48 and in the thermodynamic limit we will have the expansion holding with the correct statistical mechanics weight .
\nonumber & = & 2\tilde{\alpha}\mathbb{e}[\ln \tilde{\omega } ( e^{\beta \sum_{\nu=1}^{p_{2\tilde{\alpha } t } } \sigma_{i_\nu } + \beta\sigma_{i_0 } } ) ] - 2\tilde{\alpha}\mathbb{e}[\ln \tilde{\omega } ( e^{\beta \sum_{\nu=1}^{p_{2\tilde{\alpha } t } } \sigma_{i_\nu } } ) ] = 2\tilde{\alpha}\mathbb{e}[\ln \tilde{\omega}_t ( e^{\beta\sigma_{i_0}})]\end{aligned}\ ] ] using now the equality @xmath88 , we can write the r.h.s . of eq.([giura ] ) as @xmath89 = 2\tilde{\alpha}\log\cosh\beta - 2\tilde{\alpha}\mathbb{e}[\ln(1 + \tilde{\omega}_t(\sigma_{i_0})\theta)].\ ] ] we can expand the function @xmath90 in powers of @xmath91 , obtaining @xmath92 we learn by looking at eq.([sedici ] ) that the derivative of the cavity function is built by non - stochastically stable overlap monomials , and their averages do depend on @xmath85 making their @xmath85-integration non trivial ( we stress that all the fillable terms are zero when evaluated at @xmath93 due to the gauge invariance of the model ) .
we can escape this constraint by iterating them again and again ( and then integrating them back too ) because their derivative , systematically , will develop stochastically stable terms , which turn out to be independent by the interpolating parameter and their integration is straightforwardly polynomial .
to this task we introduce the following [ stream ] let @xmath94 be a function of s replicas
. then the following streaming equation holds @xmath95 \quad \\
\nonumber & + & \quad 2\tilde{\alpha}\theta^2 [ \sum_{a < b}^{1,s}\langle f_s \sigma_{i_0}^a\sigma_{i_0}^b \rangle_{t,\tilde{\alpha } } - s \sum_{a=1}^s\langle f_s \sigma_{i_0}^a\sigma_{i_0}^{s+1 } \rangle_{t,\tilde{\alpha } } + \frac{s(s+1)}{2!}\langle f_s \sigma_{i_0}^{s+1}\sigma_{i_0}^{s+2 } \rangle_{t,\tilde{\alpha } } ] \quad \\ \nonumber & + & \quad 2\tilde{\alpha}\theta^3 [ \sum_{a < b < c}^{1,s}\langle f_s \sigma_{i_0}^a\sigma_{i_0}^b\sigma_{i_0}^c \rangle_{t,\tilde{\alpha } } - s \sum_{a < b}^{1,s}\langle f_s \sigma_{i_0}^a\sigma_{i_0}^b\sigma_{i_0}^{s+1 } \rangle_{t,\tilde{\alpha } } \\ \nonumber & + & \frac{s(s+1)}{2!}\sum_{a=1}^{s}\langle f_s \sigma_{i_0}^a \sigma_{i_0}^{s+1}\sigma_{i_0}^{s+2 } \rangle_{t,\tilde{\alpha}}\quad + \frac{s(s+1)(s+2)}{3 ! }
\langle f_s \sigma_{i_0}^{s+1}\sigma_{i_0}^{s+2}\sigma_{i_0}^{s+3 } \rangle_{t,\tilde{\alpha } } ] \,\end{aligned}\ ] ] where we neglected terms @xmath96 .
* proof * the proof works by direct calculation : @xmath97 \\ \nonumber & & = 2\tilde{\alpha } \mathbb{e } \big [ \frac{\sum_{\{\sigma\}}f_s e^{\sum_{a=1}^s(\beta\sigma_{i_0}^a + \beta\sum_{\nu=1}^{p_{\tilde{\alpha } n } } \sigma_{i_\nu}^a\sigma_{j_\nu}^a + \beta \sum_{\nu=1}^{p_{2\tilde{\alpha } t } } \sigma_{i_\nu}^a ) } } { \sum_{\{\sigma\ } } e^{\sum_{a=1}^s(\beta\sigma_{i_0}^a + \beta\sum_{\nu=1}^{p_{\tilde{\alpha } n } } \sigma_{i_\nu}^a\sigma_{j_\nu}^a + \beta \sum_{\nu=1}^{p_{2\tilde{\alpha } t } } \sigma_{i_\nu}^a)}}\big ] - 2\tilde{\alpha}\langle f_s \rangle_{t,\tilde{\alpha } } \\ \nonumber & & = 2\tilde{\alpha } \mathbb{e}\big [ \frac{\tilde{\omega}_t ( f_s e^{\sum_{a=1}^s\beta\sigma_{i_0}^a } ) } { \tilde{\omega}_t ( e^{\sum_{a=1}^s\beta\sigma_{i_0}^a})}\big ] - 2\tilde{\alpha}\langle f_s \rangle_{t,\tilde{\alpha } } \\ \nonumber & & = 2\tilde{\alpha } \mathbb{e}\big [ \frac{\tilde{\omega}_t ( f_s \pi_{a=1}^{s}(\cosh\beta + \sigma_{i_0}^a\sinh\beta ) ) } { \tilde{\omega}_t ( \pi_{a=1}^{s}(\cosh\beta + \sigma_{i_0}^a\sinh\beta))}\big ] - 2\tilde{\alpha}\langle f_s \rangle_{t,\tilde{\alpha } } \\ \nonumber & & = 2\tilde{\alpha } \mathbb{e}\big [ \frac{\tilde{\omega}_t ( f_s \pi_{a=1}^{s}(1 + \sigma_{i_0}^a\theta ) ) } { ( 1 + \tilde{\omega}_t(\sigma_{i_0}^a)\theta)^s}\big ] - 2\tilde{\alpha}\langle f_s \rangle_{t,\tilde{\alpha } } = \end{aligned}\ ] ] now noting that @xmath98 we obtain @xmath99 - 2\tilde{\alpha}\langle f_s \rangle_{t,\tilde{\alpha}}.\end{aligned}\ ] ] from which our thesis follows @xmath37 .
now that we exploited the machinery we can start applying it to the free energy .
let us at first work out its streaming with respect to the plan @xmath38 : @xmath100 = \alpha \sum_{k=1}^{\infty } \pi(k-1,\alpha n ) \mathbb{e } \big[\frac{\omega(\sigma_{i_k}\sigma_{j_k}e^{\beta\sigma_{i_k}\sigma_{j_k}})_{k-1 } } { \omega(e^{\beta\sigma_{i_k}\sigma_{j_k}})_{k-1}}\big ] \\
\nonumber & & = \alpha \mathbb{e } \big[\frac{\omega(\sigma_{i_k}\sigma_{j_k}(\cosh\beta + \sigma_{i_k}\sigma_{j_k}\sinh\beta ) ) } { \omega(\cosh\beta + \sigma_{i_k}\sigma_{j_k}\sinh\beta)}\big ] = \alpha \mathbb{e}\big[\frac{\omega(\sigma_{i_k}\sigma_{j_k } ) + \theta } { 1 + \omega(\sigma_{i_k}\sigma_{j_k})\theta}\big]\end{aligned}\ ] ] by which we get ( and with similar calculations for @xmath101 that we omit for the sake of simplicity ) : @xmath102 now remembering theorem ( [ main ] ) and assuming critical behavior ( that we will verify a fortiori in sec .
( [ critico ] ) ) we move for a different formulation of the free energy by considering the cavity function as the integral of its derivative . in a nutshell
the idea is as follows : due to the second order nature of the phase transition for this model ( i.e. criticality that so far is assumed ) we can expand the free energy in terms of the whole series of order parameters .
of course it is impossible to manage all these infinite overlap correlation functions to get a full solution of the model in the whole ( @xmath103 ) plane but it is possible to show by rigorous bounds that close to the critical line ( that we are going to find soon ) higher order overlaps scale with higher order critical exponents so we are allowed to neglect higher orders close to this line and we can investigate deeply criticality , which is the topic of the paper . to this task let us expand the cavity functions as @xmath104 where @xmath105 for @xmath23 . now using the streaming equation
as dictated by proposition ( [ stream ] ) we can write the overlaps appearing in the expression of @xmath86 as polynomials of higher order filled overlaps so to obtain a straightforward polynomial back - integration for the @xmath86 as they no longer will depend on the interpolating parameter thanks to theorem ( [ saturi ] ) .
for the sake of simplicity the @xmath106-dependence of the overlaps will be omitted keeping in mind that our results are all taken in the thermodynamic limit and so we can quietly exchange @xmath106 with @xmath10 in these passages .
the first equation we deal with is : @xmath107\ ] ] where @xmath108 is not filled and so we have to go further in the procedure and derive it in order to obtain filled monomials : @xmath109 + \tilde{\beta}\theta [ \langle m_1m_2q_{12 } \rangle - 4\langle m_1m_2q_{13 } \rangle_t + 3\langle m_1m_2q_{34 } \rangle_t].\ ] ] in this expression we stress the presence of the filled overlap @xmath110 and of @xmath111 which can be saturated in just one derivation . wishing to have an expansion for @xmath112 up to the third order in @xmath91
, it is easy to check that the saturation of the other overlaps in the last derivative would carry terms of higher order and so we can stop the procedure at the next step @xmath113 + \tilde{\beta}[\mbox{unfilled terms } ] + o(\theta^2)\ ] ] from which integrating back in @xmath85 @xmath114t\ ] ] inserting now this result in the expression ( [ m1m2 ] ) and integrating again in @xmath85 we find @xmath115 and coming back to @xmath112 we get @xmath116 which is the attempted result . let us move our attention to @xmath117 , analogously we can write @xmath118 + \tilde{\beta}\theta [ \langle q_{12}^2 \rangle - 4\langle q_{12}q_{13 } \rangle_t + 3\langle q_{12}q_{34 } \rangle_t]\ ] ] and consequently obtain @xmath119 with the two expansion above , in the @xmath54 limit , putting @xmath48 we have @xmath120 at this point we have all the ingredients to write down the polynomial expansion for the free energy function as stated in the next : a general expansion via stochastically stable terms for the free energy of the diluted ising model can be written as @xmath121 it is immediate to check that the above expression , in the ergodic region where the averages of all the order parameters vanish , reduces to the well known high - temperature ( or high connectivity ) solution @xcite ( i.e. @xmath122 ) .
of course we are neglecting @xmath123 higher order terms because we are interested in an expansion holding close to the critical line , but we are not allowed to truncate the series for a general point in the phase space far beyond the ergodic region .
now we want to analyze the critical behavior of the model : we find the critical line where the ergodicity breaks , we obtain the critical exponent of the magnetization and the susceptibility and at the end we show that within our framework the lacking of ergodicity can be explained as the breaking of commutativity of the infinite volume limit against our cavity field , thought of as a properly chosen field , vanishing in the thermodynamic limit too , accordingly with the standard prescription of statistical mechanics @xcite .
let us firstly define the rescaled magnetization @xmath124 as @xmath125 . by applying the gauge transformation @xmath126 in the expression for the quenched average of the magnetization ( eq .
( [ magneto ] ) ) and multiplying it times @xmath0 so to switch to @xmath127 , setting @xmath48 and sending @xmath128 we obtain @xmath129 by which we see ( again remembering criticality and so forgetting higher order terms ) that the only possible divergence of the ( centered and rescaled ) fluctuations of the magnetization happens at the value @xmath130 which gives @xmath131 as the critical line , in perfect agreement with @xcite
. the same critical line can be found easier simply by looking at the expression ( [ a ] ) as follows : remembering that in the ergodic phase the minimum of the free energy corresponds to a zero order parameter ( i.e.@xmath132 ) , this implies the coefficient of the second order @xmath133 to be positive .
anyway immediately below the critical line values of the magnetization different from zero must be allowed ( by definition otherwise we were not crossing a critical line ) and this can be possible if and only if @xmath134 . consequently ( and using once more the second order nature of the transition ) on the critical line we must have @xmath135 and this gives again @xmath136 .
phase diagram : below @xmath137 there is no giant component in the erdos - renyi graph , @xmath138 defines the percolation threshold . above at left of the critical line the system behaves ergodically , conversely on the right ergodicity is broken and the system displays magnetization.,width=226,height=188 ] now
let us move to the critical exponents : critical exponents are needed to characterize singularities of the theory at the critical line and , for us , these indexes are the ones related to the magnetization @xmath139 and to the susceptibility @xmath140 .
we define @xmath141 and we write @xmath142 and @xmath143 , where the symbol @xmath144 has the meaning that the term at the second member is the dominant but there are corrections of higher order . remembering the expansion of the squared magnetization that we rewrite for completeness @xmath145 and considering that using the same gauge transformation @xmath126 on ( eq.([overlap ] ) ) we have for the two replica overlap the following representation @xmath146 we can arrive by simple algebraic calculations to write down the free energy , close to the critical line of course , depending only by the two parameters @xmath147 and @xmath148 @xmath149 by a comparison of the formula obtained by deriving @xmath150 as expressed by eq.([38 ] ) and the expression we have previously found ( eq . ( [ dada ] ) ) that we report for the sake of readability , @xmath151 it is immediate to see that we have @xmath152= \theta \langle m_1 ^ 2 \rangle.\ ] ] if we put ourselves close to the value @xmath153 and make a change of variable @xmath154 with @xmath155 we get @xmath152 \sim \frac{\theta}{2}\frac{\partial}{\partial\tau}[\tau\langle m_1 ^ 2 \rangle ] = \frac{\theta}{2}\langle m_1 ^ 2 \rangle + \frac{\theta\tau}{2}\frac{\partial\langle m_1 ^ 2 \rangle}{\partial \tau } = \theta \langle m_1 ^ 2 \rangle\ ] ] by which we easily obtain @xmath156 therefore we get that the critical exponent for the magnetization , @xmath157 , which turns out to be the same as in the fully connected counterpart @xcite@xcite , in agreement with the disordered extension of this model @xcite . again , by simple direct calculations , once we get the critical exponent for the magnetization it is straightforward to show that the susceptibility @xmath140 @xcite obeys ~||^-1 close to the critical line , by which we find its critical exponent to be once again in agreement with the classical fully connected counterpart @xcite .
now we want to show some wrong results which a naive calculation would suggest so to emphasize the importance of the bounds relating different monomials that we are going to discuss immediately later @xcite@xcite .
then in the next subsection , we explain what is the physics behind this picture by providing a mechanism for the breaking of the ergodicity .
the point on which we focus is the following : if we wish to perform the same procedure we performed on @xmath147 , applying blindly saturability below to the first critical line , to the @xmath158 replica overlap @xmath159 we would gain @xmath160 identifying @xmath161 as another critical temperature , or better , the critical temperature typical of @xmath159 . in the same way we could find @xmath162 for every @xmath163 obtaining @xmath164 such that
, at the end , we obtain a scenario with several transition lines , one for every order parameter .
this is not a possible scenario , as generally explained for instance in @xcite and as dictated , in this model , by the following as soon as the first order parameter ( the magnetization ) starts taking values different from zero , the same happens to all the other order parameters @xmath165 \geq ( e_k e_i [ \omega(\sigma_{i_1}\sigma_{i_2})])^n = \langle m^2 \rangle^n \quad \forall n\ ] ] we omit the proof details as they are a simple application of jensen inequality .
so far we showed that it is not possible to have several transition lines , one for every order parameter .
now we want to understand why there is just one critical line by applying the theory developed in @xcite to this model . starting from theorem ( [ main ] ) that we recall for simplicity @xmath166 we want to show the phase transition expressed by the non - commutativity among the thermodynamic limit and the vanishing perturbation .
again for simplicity we report the expansion of @xmath150 we have previously built @xmath167 that we obtained by considering the cavity function as the integral of its @xmath85-derivative @xmath168 and performing , via the streaming equation , a saturating procedure of consecutive @xmath85-derivatives upon the overlaps in order to express them as functions of higher order filled terms .
then the only thing we had to do was sending @xmath0 to infinity , carrying out of the integral the overlaps and then putting @xmath48 to evaluate @xmath169 .
the result of this procedure brings to equation ( [ afinale ] ) .
but what if we exchanged the limit order by sending @xmath170 first and taking the thermodynamic limit after ?
it is easy to note that all the overlaps appearing in eq.([cavita ] ) are fillable such that we can avoid the saturation procedure simply by setting @xmath48 first and then sending n to infinity . in this way , thanks to theorem ( [ saturabili ] ) , each fillable overlap is transformed in a filled @xmath85-independent one and this kills all the correlations among different replicas and allows us to write @xmath171 being clearly @xmath172 \\ \nonumber & = & 2\alpha\ln\cosh\beta - 2\alpha \sum_{n=1}^{\infty } \frac{(-1)^n}{n } \theta^{n } \langle q_{1, ..
,n}^2 \rangle = 2\alpha \frac{\partial a(\alpha,\beta)}{\partial\alpha}.\end{aligned}\ ] ] in particular , retaining just the first two terms of the expansions , we show the difference between the two results . * @xmath173 + @xmath174 * @xmath175 + @xmath176 we immediately recognize that eq.([51 ] ) is the correct expression holding also below the critical line .
when the system lives in the ergodic region all the order parameters are zero and it reduces to @xmath177 which is the ergodic solution , if we cross the critical line the formula takes into account the phase transition encoded in the coefficient of the second order and gives the correct expression immediately below .
it is also straightforward to recognize as the ergodic solution eq.([52 ] ) which can be the correct one only above the critical line @xcite . at the end we saw that there exists only and only one critical line for all the order parameters .
we saw that this line can be depicted as the breaking of commutativity among the infinite limit operation and the setting of @xmath48 ( relaxing the boltzmann factor from the interpolant avoiding the trivial way @xmath93 ) . hence , which can be the genesis of the transition for all the other order parameters ?
the correlations among them and the magnetization , as clearly put in evidence in the free energy expansion ( see eq.([a ] ) ) . and
lastly , which is the origin of these correlation ?
saturability property _ that the model exhibits as stated by theorem ( [ saturabili ] ) .
in fact , at the critical line ( which is the last line , from above , where gauge invariance is a symmetry of the boltzmann state too , thanks to the continuity of the transition ) saturability easily shows that _
n m_1 q_12 _ t = m_1 m_2 q_12 , explaining the birth of the correlations among the various order parameters ( we reported just the first two as an example ) .
we have previously shown how filled overlaps become asymptotically @xmath85-independent when n grows to infinity .
starting from this we can find identities stating the self - averaging of the order parameters as @xmath147 .
in particular we are going to take these overlaps and calculate their derivative with respect to @xmath85 ; then by applying the gauge transformation and setting @xmath178 we can write down the self - averaging relations . @xmath179 \rightarrow [ \langle m_1 ^ 4 \rangle - \langle m_1 ^ 2m_2 ^ 2 \rangle ] = \mathbb{e}[\omega(m^4 ) - \omega^2(m^2)].\ ] ] coherently with @xcite we find standard self - averaging for the magnetization .
more interesting is the situation concerning the overlap but we actually lack a complete mathematical control . by applying the same trick as before we can equate to zero ( in the large @xmath0 limit ) the @xmath85-derivative of the squared overlap ( which is stochastically stable ) ,
consequently we get two terms .
@xmath180 + \beta'\theta[\langle q_{12}^3 \rangle_t - 4\langle q_{12}^2q_{13}\rangle_t + 3\langle q_{12}^2q_{34 } \rangle_t ] \\
\label{ipotesi } & \rightarrow & [ \langle m_1 ^
2q_{12}^2 \rangle - \langle m_3 ^ 2q_{12}^2 \rangle ] = 0 \qquad
[ \langle q_{12}^4 \rangle - 4\langle q_{12}^2q_{13}^2\rangle + 3\langle q_{12}^2q_{34}^2 \rangle ] = 0\end{aligned}\ ] ] a - priori we can not assume factorization of the series so to put to zero each term separately ( as we did in eq.([ipotesi ] ) ) , however , close to the critical line surely the second term is an higher order and we can reasonably set to zero the first .
furthermore as the second term at the r.h.s . has a pre - factor @xmath181 differing from the first term it would be difficult to imagine the opposite .
it is in fact very natural to assume that the two terms can be set to zero separately in the whole @xmath38 plane and this is very interesting because the second term is a very well known relation in the field of spin glasses @xcite@xcite@xcite@xcite@xcite suggesting a common structure among different kinds of disordered systems , the only sharing feature among diluted ferromagnets and spin - glasses being some kind of disorder ( topological in the former , frustrating in the latter ) .
we are not going to deepen this point as it is under investigation in @xcite where the same set of relations ( and more ) is found and discussed .
in this section we analyze , from the numerical point of view , the ferromagnetic system previously introduced by performing extensive monte carlo simulations with the metropolis algorithm @xcite .
the erdos - renyi random graph is constructed by taking @xmath0 sites and introducing a bond between each pair of sites with probability @xmath182 , in such a way that the average connectivity per node is just @xmath183 . clearly , when @xmath184 the complete graph is recovered .
the simplest version of the diluted curie - weiss hamiltonian has a poisson variable per bond as @xmath185 , and it is the easiest approach when dealing with numerics . for the analytical investigation
we choose a slightly changed version ( see eq.([ham ] ) ) : each link gets a bond with probability close to @xmath186 for large @xmath0 ; the probabilities of getting two , three bonds scale as @xmath187 therefore negligible in the thermodynamic limit .
working with directed links ( as we do in the analytical framework ) the probability of having a bond on any undirected link is twice the probability for directed link ( i.e. @xmath188 ) .
hence , for large @xmath0 , each site has average connectivity @xmath189 .
finally in this way we allow self - loop but they add just @xmath190-independent constant to @xmath191 and are irrelevant , but we take the advantage of dealing with an @xmath191 which is the sum of independent identically distributed random variables , that is useful for analytical investigation . when comparing with numerics consequently
we must keep in mind that @xmath192 . in the simulation , once the network has been diluted , we place a spin @xmath193 on each node @xmath194 and allow it to interact with its nearest - neighbors .
once the external parameter @xmath8 is fixed , the system is driven by the single - spin dynamics and it eventually relaxes to a stationary state characterized by well - defined properties .
more precisely , after a suitable time lapse @xmath195 and for sufficiently large systems , measurements of a ( specific ) physical observable @xmath196 fluctuate around an average value only depending on the external parameters @xmath197 and @xmath183 . moreover , for a system @xmath198 of a given finite size @xmath0 the extent of such fluctuations scales as @xmath199 with the size of the system .
the estimate of the thermodynamic observables @xmath200 is therefore obtained as an average over a suitable number of ( uncorrelated ) measurements performed when the system is reasonably close to the equilibrium regime .
the estimate is further improved by averaging over different realizations of the same system @xmath198 . in summary , @xmath201 , \ ; \ ; t_n = t_0+n\mathcal{t}\ ] ] where @xmath202 denotes the configuration of the magnetic system at time step @xmath85 and @xmath203 is the decorrelation parameter ( i.e. the time , in units of spin flips , needed to decorrelate a given magnetic arrangement ) . in general , statistical errors during a mc run in a given sample result to be significantly smaller than those arising from the ensemble averaging ( see also @xcite ) . finite size scaling for the magnetization and the internal energy ( inset ) for @xmath204 and @xmath205 .
all the measurements were carried out in the stationary regime and the error bars represent the fluctuations about the average values .
we find good indication of the convergence of the quantities on the size of the system and thus of the existence of the thermodynamic limit.,height=188 ] figure ( [ fig : fss ] ) shows the dependence of the macroscopic observables @xmath139 and @xmath206 from the size of the system ; values are obtained starting from a ferromagnetic arrangement , at the normalized inverse temperature @xmath207 .
notice that at this temperature the system composed of @xmath208 is already very close to the asymptotic regime .
analogous results are found for different systems @xmath198 , with @xmath8 far enough from @xmath209 .
log - log scale plot of magnetization ( main figure ) and susceptibility ( inset ) versus the reduced temperature @xmath210 for @xmath211 .
symbols represents data from numerical simulations performed on systems of size @xmath212 , while lines represent the best fit .
, height=158 ] in the following we focus on systems of sufficiently large size so to discard finite size effects . for a wide range of temperatures and dilutions ,
we measure the average magnetization @xmath213 and energy @xmath206 , as well as the magnetic susceptibility @xmath214 , calculated as @xmath215.\ ] ] their profiles display the typical behavior expected for a ferromagnet and , consistently with the theory , highlight a phase transition at well defined temperatures @xmath216 . now
, we investigate in more detail the critical behavior of the system .
we collect accurate data of magnetization and susceptibility , for different values of @xmath183 and for temperatures approaching the critical one .
these data are used to estimate both the critical temperature and the critical exponents for the magnetization and susceptibility . in fig .
( [ fig : crit ] ) we show data as a function of the reduced temperature @xmath217 for @xmath204 and @xmath218 .
the best fit for observables is the power law @xmath219 we obtain estimates for @xmath220 , @xmath221 and @xmath222 by means of a fitting procedure .
results are gathered in tab .
[ tabexp ] .
[ tabexp ] .estimates for the critical temperature and the critical exponents @xmath67 and @xmath223 obtained by a fitting procedure on data from numerical simulations concerning ising systems of size @xmath212 and different dilutions ( we stress that analytically we get @xmath224 and @xmath225 ) .
errors on temperatures are @xmath226 , while for exponents are within @xmath227 .
[ cols="<,<,<,<,<,<",options="header " , ] within the errors ( @xmath228 for @xmath209 and @xmath229 for the exponents ) , estimates for different values of @xmath183 agree and they are also consistent with the analytical results exposed in sec .
( [ critico ] ) we also checked the critical line for the ergodicity breaking , again finding optimal agreement with the criticality investigated by means of analytical tools .
in this paper we developed the _ interpolating cavity field _
technique for the mean field diluted ferromagnet .
once the general framework has been built we used it to analyze criticality : we found analytically the critical line and the critical exponent of the magnetization , whose self - averaging is also proved .
we present an argument to explain the transition from an ergodic phase to a broken ergodicity phase by the breaking of commutativity of two limits , volume and applied field , as dictated by standard statistical mechanics .
we furthermore showed the existence of only one critical line where all the multi - overlaps start taking positive values as soon as the magnetization becomes different from zero .
we proved this both mathematically by a rigorous bound and physically by a mechanism that generates strong correlations among magnetization and overlaps at the ( unique ) critical line : _ saturability_. at the end a detailed numerical analysis of the model is presented : by sharp monte carlo simulations the convergence of the energy density ( and the magnetization ) to its limit is investigated obtaining monotonicity in the system size .
the critical line , as well as scaling of the magnetization and the susceptibility , are also investigated obtaining full agreement among theory and simulations .
future works should extend these techniques to several lateral models as the bipartite diluted mean field ising models , while the need of stronger techniques to go well beyond the critical line is also to be satisfied as well as their practical application to social science or biological networks .
we plan to follow these research lines in the future .
the authors are pleased to thank francesco guerra , pierluigi contucci and peter sollich for useful suggestions .
ea thanks the italian foundation `` angelo della riccia '' for financial support .
ab acknowledges partial support by the cultaptation project ( european commission contract fp@xmath230-nest - path-@xmath231 ) and partial support by the miur within the smart - life project ( ministry decree @xmath232 n.@xmath233 ) .
f. guerra , _ the cavity method in the mean field spin glass model .
functional reperesentations of thermodynamic variables _ , in `` advances in dynamical systems and quantum physics '' , s. albeverio _
eds . , singapore ( 1995 ) .
f. guerra , _ sum rules for the free energy in the mean field spin glass model _
, in _ mathematical physics in mathematics and physics : quantum and operator algebraic aspects _ , fields institute communications * 30 * , american mathematical society ( 2001 ) . | we perform a detailed study of the critical behavior of the mean field diluted ising ferromagnet by analytical and numerical tools .
we obtain self - averaging for the magnetization and write down an expansion for the free energy close to the critical line .
the scaling of the magnetization is also rigorously obtained and compared with extensive monte carlo simulations .
we explain the transition from an ergodic region to a non trivial phase by commutativity breaking of the infinite volume limit and a suitable vanishing field .
we find full agreement among theory , simulations and previous results . |
compact accreting x - ray sources in nearby galaxies with luminosities exceeding 10@xmath11 erg s@xmath12 are referred to as ultraluminous x - ray sources ( ulxs ) .
current evidence suggests that ulxs might be a mixed bag of compact objects including stellar - mass black holes ( stmbhs : 3 - 25 @xmath9 ) powered by super - eddington accretion ( e.g. , king et al .
2001 , begelman 2002 , gladstone et al .
2009 ) , intermediate - mass black holes ( imbhs : a few@xmath2(100 - 1000 ) @xmath9 ; kaaret et al .
2001 , 2006 , matsumoto et al .
2001 , farrell et al .
2009 , pasham et al .
2014 , mezcua et al .
2015 ) , and neutron stars ( bachetti et al . 2014 ) .
one of the biggest challenges in understanding ulxs is to estimate their compact object masses . because their optical counterparts are faint ( v - band magnitudes of 22 - 24 ; e.g. , gladstone et al . 2013 ,
tao et al . 2011 ) , doppler tracking their optical counterparts to derive their mass functions as done for galactic stmbhs has been extremely challenging ( e.g. , roberts et al . 2011 , cseh et al . 2013 ) .
however , in a few ulxs , such optical measurements have yielded mass constraints that suggest lower - mass black holes ( @xmath13 30 @xmath9 ; liu et al .
2013 , motch et al .
2014 ) .
it has been suggested that the detection of the 3:2 frequency ratio high - frequency quasi - periodic oscillations ( qpos ) can resolve the ulx mass problem ( abramowicz et al .
stmbh high - frequency qpos ( frequency range of 100 - 450 hz ; mcclintock & remillard 2006 ) that appear in a 3:2 frequency ratio scale inversely with the black hole mass .
moreover , it has been demonstrated that the power density spectra ( pds ) of both the stellar - mass and the supermassive black holes are qualitatively similar .
the pds break timescale of both simply scale with the black hole mass , after accounting for the differences in the accretion efficiency between sources ( e.g. , mchardy et al .
2006 , krding et al .
one can also use this break timescale to estimate black hole masses .
thus , under this black hole variability unification paradigm , 3:2 high - frequency qpo analogs of stmbhs should also be detectable from imbhs , but with centroid frequencies scaled down according to their respective black hole masses ( vaughan & uttley 2005 ) .
for example , a few 1000 @xmath9 imbh should exhibit high - frequency qpos with centroid frequencies in the range of a fraction of hz .
in fact , such a 3:2 ratio qpos have already been detected from the ulx m82 x-1 . in that source ,
the two qpos ( 3.3 and 5 hz ) allowed pasham et al .
( 2014 ) to estimate its black hole mass to be 428@xmath4105 @xmath9 . here ,
we report evidence for a second such high - frequency pair from another ulx , ngc 1313 x-1 .
as of the writing of this paper , 22 of the 24 _ xmm - newton _ observations of ngc 1313 are public .
the three brightest x - ray sources in this field , the two ulxs , ngc 1313 x-1 ( hereafter , x-1 ) , x-2 and the x - ray bright supernova sn1978k are well separated in the xmm images ( see figure 1 ) .
previous energy spectral studies of x-1 suggest it may host an imbh with a mass of @xmath8 1000 @xmath9 ( e.g. , miller et al .
2003 , 2013 ) . given the frequency range we are interested in , we used data primarily from epic - pn , utilizing events in the entire 0.3 - 10.0 kev band pass .
both the pn and the mos detectors were operated in the so - called full - frame mode during all the observations . while pn s full - frame data mode offers a time resolution of 73.4 ms , i.e. , nyquist frequency of 6.82 hz , mos data is limited to a nyquist frequency of only 0.19 hz .
we reduced all the observations using the standard data reduction procedures and removed datasets that were severely affected by background flaring .
this preliminary screening left us with fifteen observations ( table 1 ) .
source events were extracted from a circular region of radius 33@xmath14 ( dashed circle around x-1 in figure 1 ) when x-1 was clear of a ccd gap . when x-1 was close to or on a ccd gap we extracted events from a smaller region of radius 25@xmath14 excluding the ccd gap .
circular background regions of the same radius , and free of any point sources , were chosen away from the source s readout column and as close to the telescope pointing as possible . in order to assess x-1 s variability , we first extracted the source and the background light curves from each of the fifteen observations .
background flaring was prominent for only brief durations in some observations .
we constructed good time intervals ( gtis ) accounting for both the background flares and times when the detector was turned off .
figure 2 contains sample background - subtracted x - ray light curves ( black ) and their respective backgrounds ( red ) from all the observations whose power spectra are described in this article . using the gtis shown in figure 2
, we constructed a leahy normalized ( poisson noise level of 2 ; leahy 1983 ) pds of x-1 from each of the individual observations .
all the power spectra were sampled only up to 4 hz , a value safely below the nyquist frequency of 6.82 hz , in order to avoid any aliasing affects .
we started our timing analysis with the three longest observations ( obsids : 0405090101 , 0693850501 , and 0693851201 ) during which x-1 was positioned on - axis , giving the best sensitivity for detecting qpos .
the top panel of figure 3 shows the combined epic ( pn+mos ) pds from obsid 0405090101sampled in the frequency range from 0.0022 to 4 hz .
we first extracted a pn - only pds and found a qpo - like feature at 80 mhz , a frequency well below the mos detectors nyquist frequency .
therefore , in order to improve the signal , we used pn+mos data in this single instance .
it is evident that the overall shape of the power spectrum is flat - topped at the lowest frequencies , breaking into a power - law , and white noise at the highest frequencies .
two broad , qpo features at centroid frequencies of roughly 30 and 80 mhz are also apparent .
we modeled the continuum with a constant plus a bending power law model@xmath15 } \ ] where c , n , $ \nu_{bend}$ are the poisson noise level , the normalization of the bending power - law and the bending frequency , respectively while $ \alpha$ and $ \beta$ are the power - law indicies below and above the bending frequency , respectively.}$ ] similar to other ulxs ( e.g. , pasham & strohmayer 2012 , 2013 ) , and stmbhs in the steep power - law state ( mcclintock & remillard 2006 ) .
this gave a best - fit @xmath16 of 419 for 252 degrees of freedom ( dof ) .
we then added a lorentzian component to model the qpo at 80 mhz .
this improved the @xmath16 by 89 , i.e. , @xmath16 of 330 for 249 dof . using the f - test
, this corresponds to a significance of 8@xmath210@xmath17 or @xmath5 7@xmath6 .
the best - fit qpo has a centroid frequency , width , normalization , and a root - mean - squared ( rms ) amplitude of [email protected] mhz , [email protected] mhz , [email protected] , and [email protected]% , respectively . adding a second qpo improved the @xmath16 by 40 ( 290 for 246 dof ) , which corresponds to an f - test probability of 6@xmath210@xmath18 ( @xmath5 4.9@xmath6 ) .
the best - fit centroid frequency , width , normalization and rms amplitude of the second qpo were [email protected] mhz , [email protected] , [email protected] , and [email protected]% , respectively . however , protassov et al . ( 2002 ) pointed out the problems with applying the f - test in additive models .
therefore , in order to estimate the qpo significances independent of the f - test , we employed a rigorous monte carlo approach as follows .
\1 ) we estimated the baseline bending power - law plus a constant model parameters along with their uncertainties , and then randomly sampled @xmath19 = 1.8@xmath210@xmath20 model parameter sets from within the best - fit parameter error bars ( assuming normal distribution ) .
\2 ) for each of these parameter sets , we simulated a light curve of the same length as the observed one following the algorithm described in timmer & koenig ( 1995 ) , and then extracted a pds from this simulated light curve .
\3 ) after binning these pds in the same way as the original pds , we modeled them with a bending power law .
we then added a qpo to this model , with the qpo frequency constrained to lie between 0.01 and 1 hz .
the maximum improvement in @xmath16 ( @xmath21@xmath22 ) was recorded from each simulated pds .
\4 ) the significance of the 80 mhz qpo was estimated as 1 - @xmath23/@xmath19 , where @xmath23 is the number of simulated @xmath24 values greater than the observed @xmath25 . for estimating the significance of the 30 mhz qpo
, we assumed the baseline model to be the best - fit continuum plus 80 mhz qpo of the observed pds , and recorded the maximum @xmath26 values by adding an additional qpo to this base model .
this methodology is similar to estimating the significances of spectral lines in energy spectra ( e.g. , tombesi et al .
2010 , zoghbi et al . 2015 ) . for the 80 mhz feature we ran 1.8@xmath210@xmath20 simulations .
the maximum @xmath26 was 25 , which is much lower than the observed value of 89 .
thus , we conclude that the 80 mhz feature is significant at least at the 5@xmath6 level .
we ran 200,000 simulations to test the significance of the 30 mhz feature and found that one run exceeded the observed @xmath26 of 40 .
thus , we conclude its significance is 1 - ( 1/200,000 ) or @xmath1 4.4@xmath6 .
the pds from observation 0693850501 exhibited a strong feature at roughly 0.3 hz ( bottom - left panel of figure 3 ) .
the continuum looks flat because we only sampled up to roughly 0.02 hz .
a flat - topped , followed by a power - law like continuum can be seen when we sample up to 0.001 hz using longer light curve segments .
modeling the continuum with a constant yielded a best - fit @xmath16 of 186 for 127 dof . adding a lorentzian component improved the @xmath16 by 56 with an addition of three parameters ( @xmath16 of 130 for 124 dof ) .
this corresponds to an f - test probability of 1.1@xmath210@xmath27 or @xmath5 6@xmath6 . using
the monte carlo approach to test the significance , with 1.8@xmath210@xmath20 simulations , we find a lower limit of 5@xmath6 . for the monte carlo simulations
, we modeled each of the simulated pds with a constant , and a constant plus a lorentzian model , and recorded the maximum improvement in the @xmath26 .
out of the 1.8@xmath210@xmath20 simulations , the maximum @xmath26 improvement was 27 , a value much less than the observed @xmath26 improvement of 56 .
the best - fit centroid frequency , width , normalization and the rms of the qpo were [email protected] hz , [email protected] hz , [email protected] , and [email protected]% , respectively .
we also estimated the significance using another independent method .
first we rescaled the pds ( rescaling factor of 1.01 , and the rescaled power at 0.29 hz in the highest bin is 2.173 ) so that the local mean around 0.5 hz is equal to 2 , the value expected from a purely poisson ( white noise ) process .
we then computed the probability with 3@xmath6 and the 4@xmath6 confidence , of obtaining a power that is at or higher than some threshold p@xmath28 , probability(p @xmath5 p@xmath28 ) = n@xmath29@xmath2q(p@xmath28@xmath2898@xmath24@xmath302@xmath2898@xmath24 ) , where q(p@xmath28@xmath2898@xmath24@xmath302@xmath2898@xmath24 ) is the probability of obtaining a @xmath16 value of p@xmath28@xmath2898@xmath24 or higher from a @xmath16 distribution with 2@xmath2898@xmath24 dof .
we used this @xmath16 distribution because we averaged in frequency by a factor of 4 and averaged 898 individual power spectra each derived from 128 s light curve segments .
n@xmath29 account for the total number of trials ( frequency bins within 0.1 - 4 hz ) .
the confidence contours are marked by horizontal dotted lines in the bottom - left panel of figure 3 .
clearly , the highest bin in the 0.3 hz qpo is significant at greater than the 4@xmath6 level . after establishing the lower - frequency continuum and qpos , and the 0.3 hz qpo
, we searched for a signal at 2/3 and 3/2 of 0.3 hz separately from all the gtis longer than 7 ks .
the pds from the 4@xmath31 gti ( exposure @xmath1 22 ks ) of this observation ( top panel of figure 4 ) showed evidence for excess power at [email protected] hz , a value consistent with 3/2 of 0.3 hz .
this feature is significant at the 7.6@xmath210@xmath32 level ( the significance is 1@xmath210@xmath33 if all frequencies were searched ) .
we estimated the significance of this feature as follows .
first , we estimated the probability of detecting a false peak with a power value of 2.13 .
this is the probability of getting a @xmath16 value of 2.13@xmath2174@xmath216 or higher from a @xmath16 distribution with 2@xmath2174@xmath216 dof .
this value is 3.8@xmath210@xmath32 .
however , after securing the 0.3 hz feature , we searched in two bins , one at 2/3 and the other at 3/2 ( bin width of 0.125 hz ) .
considering the two trials , the significance in the 4@xmath31 gti is [email protected]@xmath210@xmath32 .
this significance level does not take into account the number of gtis searched yet .
we estimate its global significance considering all gtis in section 3.1.5 .
a feature at a frequency of [email protected] hz ( middle panel of figure 4 ) as observed in the pds of observation 0693850501 ( taken roughly a week before this observation ) was again present , albeit at a lower significance of @xmath7 3.3@xmath6 .
similar to the bottom - left panel of figure 3 , the dotted horizontal lines represent the 3 and the 4@xmath6 confidence contours considering all frequency bins ( trials ) between 0.1 and 4 hz .
we then constructed pds from each of the shorter observations .
observation 0150280401 showed evidence for a qpo with a centroid frequency of [email protected] hz ( rms amplitude of 12@xmath42% ) , a value consistent with 3/2 of 0.3 hz .
the significance of this feature again considering a search around 2/3 and 3/2 of 0.3 hz is 2.8@xmath210@xmath32 estimated as follows .
first , we estimated the probability of getting a power value of 2.451 , i.e. , the probability of getting a @xmath16 value of 2.451@xmath281@xmath24 or higher from a @xmath16 distribution with 2@xmath281@xmath24 dof .
this value is 7@xmath210@xmath34 .
however , considering that we searched in two bins ( width per bin of 0.03125 hz ) around 3/2 and 2/3 of 0.3 hz , this translates to 4@xmath27@xmath210@xmath34 ( @xmath5 3.5@xmath6 ) .
the significance of the feature considering a full frequency search is 6@xmath210@xmath35 .
observation 0205230601 showed weak evidence for qpo - like feature at the @xmath5 2@xmath6 level ( bottom panel of figure 4 ) .
the pds from the remaining observations were essentially featureless with evidence for red - noise in a handful of them . to ensure that these qpo features are not associated with the background
, we extracted all the background pds from these fifteen observations .
they are all consistent with being featureless ( flat , white noise ) within the error bars . in order to estimate the global significance of the 0.45 hz qpo we considered all the 24 gtis
longer than 7 ks where we searched for qpos ( see table 1 ) .
figures 3 & 4 show all the statistically significant qpos detected during this search . the qpo s global probability can be calculated straightforwardly using the binomial distribution formula which gives the probability of happening of a certain event @xmath36 times in @xmath37 trials as , @xmath38 where @xmath37 is the total number of gtis searched , and @xmath36 is the number of gtis where the signal was detected at a probability of @xmath39 . using the above formula
, the global probability of detecting the 0.45 hz feature at @xmath5 7.6@xmath210@xmath32 significance in two ( @xmath36 ) out of the twenty four ( @xmath37 ) gtis is 1.6@xmath210@xmath32 ( @xmath5 3.5@xmath6 ) .
similarly , we estimate the global probability of the 0.3 hz qpo , using @xmath37 = 24 , @xmath36 = 1 , and @xmath39 = 5.6@xmath210@xmath18 ( 5@xmath6 ) , to be @xmath5 3.5@xmath6 . _
note that this is very conservative lower limit as we have not included the case where the 0.3 hz feature was detected at @xmath5 3.3@xmath6_.
ngc 1313 x-1 s pds has all the features of a typical stmbh in the steep power - law state , but with characteristics time scales @xmath81000@xmath2 longer .
typical stmbhs have three components : ( 1 ) a continuum often flat - topped at the lowest frequencies , with a power - law like decline beyond a certain break frequency , followed by white noise at the highest frequencies ; ( 2 ) low - frequency qpos ( frequencies of a few hz ) ; and , finally ( 3 ) high - frequency qpos ( frequency range of 100 - 450 hz ) exhibited by some systems .
three stmbhs with known masses show high - frequency qpos in harmonic pairs with centroid frequencies in a ratio consistent with 3:2 ( e.g. , miller et al .
2001 , strohmayer et al .
2001a , 2001b , remillard et al .
2002 ) . in these systems , the two qpos are often not simultaneous ( e.g. , see table 1 of remillard 2002 & strohmayer 2001b ) . furthermore , unlike the low - frequency qpos ,
these are stable in frequency with changes in source luminosity ( mcclintock & remillard 2006 ) . also , the timescales associated with them ( @xmath80.01 s ) are comparable to the keplerian orbital periods of a test particle close to the innermost stable circular orbit .
the commonalities of the high - frequency qpos in these three stmbh systems suggests a common physical origin . under the assumption that these originate from a radius fixed in gravitational units ( @xmath40/@xmath41 , where @xmath42 , @xmath43 , and @xmath44 are the gravitational constant , the black hole mass and the speed of light , respectively ) ,
their frequency should simply scale inversely with the black hole mass .
indeed , the three stmbhs with high - frequency qpos do agree with this inverse scaling law ( see figure 4.17 of mcclintock & remillard 2006 , zhou et al . 2015 ) .
we suggest that the observed lower 30 and 80 mhz , and the higher 0.3 and 0.45 hz 3:2 ratio qpos are the analogs of the low and the high - frequency qpos of stmbhs , and that x-1 may be in an x - ray accretion state similar to the steep power - law state of stmbhs .
this result agrees with prior work by feng & kaaret ( 2006 ) , who studied x-1 s x - ray ( 0.3 - 10 kev ) energy spectral variability using _ xmm - newton _ data taken between 2000 and 2005 .
they concluded that the source resides in either the steep power - law state at high luminosities or in the low / hard state at lower luminosities , but never enters the thermal dominant state ( see bachetti et al .
2013 for alternate arguments ) . using the dynamical masses and the high - frequency qpos of stmbhs@xmath45 xte j1550 - 564 , gro j1655 - 50 , and grs 1915 + 105
, we estimate based on the inverse mass scaling that x-1 s black hole mass is 5000@xmath41300 @xmath9 .
we also measured the mass using the break frequency
black hole mass - accretion rate scaling as derived by mchardy et al .
( 2006 ) . using a break frequency of 16@xmath43 mhz from obsid 0405090101 , and assuming a lower limit on the bolometric luminosity of 2@xmath210@xmath3 erg s@xmath12 implies a black hole mass of @xmath5 7600@xmath45600 @xmath9@xmath46 .
this value is consistent with the measurement from the 3:2 qpo pair .
yang et al .
( 2011 ) estimated x-1 s bolometric correction factor ( ratio of the x - ray to optical flux ) to be similar to that of typical stmbhs , implying that a significant fraction of the emission is in the x - rays .
x-1 s average x - ray ( 0.3 - 10.0 kev ) luminosity of 2@xmath210@xmath3 erg s@xmath12 ( e.g. , feng & kaaret 2006 ) implies an eddington ratio of @xmath5 [email protected] .
this value is significantly low compared to the typical eddington ratios of stmbhs in the steep power - law state ( @xmath5 0.2@xmath47 ; mcclintock & remillard et al . 2006 ) . assuming the 0.2 value reported by mcclintock & remillard ( 2006 )
, x-1 s low eddington ratio is inconsistent with the interpretation of it being in the steep power - law state .
bachetti et al .
( 2013 ) carried out timing analysis of two of the data sets described here ( 0693850501 and 0693851201 ) .
however , they did not report any evidence for qpos at 0.3 hz .
we suspect the reason for this discrepancy is because they extracted their pds using both the pn and the mos data .
as described earlier , mos is limited to a frequency of @xmath1 0.19 hz .
in addition to any signal beyond 0.19 hz being unreliable , even at frequencies lower than , but close to 0.19 hz , signal suppression can be severe ( van der klis 1989 ) . to test this , we extracted a pds using a combined pn and mos event list from the observations 0693850501 and 0693851201 . while we still see evidence for a feature at 0.3 hz and and at @xmath8 90 mhz , they were statistically less significant than in the analysis presented above .
the first ulx 3:2 pair from m82 x-1 was detected using the rossi x - ray timing explorer ( _ rxte _ ) , while the detection reported here is from _ xmm - newton_. currently , _
xmm - newton _ is the only x - ray observatory that can provide both large enough effective area and the required time resolution to detect these oscillations from ulxs .
therefore , deeper x - ray observations of other variable ulxs , _ viz .
_ , ngc 5408 x-1 ( middleton et al .
2011 ) , ngc 6946 x-1 ( rao et al .
2010 ) are strongly encouraged . also , the reported 3:2 pair from x-1 boosts confidence in the prospects of detecting high - frequency qpos from relatively isolated ulxs with neutron star interior composition explorer ( _ nicer_)with an anticipated effective area of 1.5 times greater than epic - pn .
lccccccc + obsid & date ( utc ) & observation & count rate & effective & number of + & & time ( ks ) & ( counts s@xmath12 ) & exposure(ks ) & gtis @xmath5 7 ks + + + 0106860101 & 2000 - 10 - 17 & 42.4 & @xmath48 & 31.6 & 1 + + 0150280301 & 2003 - 12 - 21 & 16.2 & @xmath49 & 9.7 & 1 + + 0150280401 & 2003 - 12 - 23 & 20.9 & @xmath50 & 10.5 & 1 + + 0150280501 & 2003 - 12 - 25 & 21.4 & @xmath51 & 9.8 & 1 + + 0150280601 & 2004 - 01 - 08 & 53.2 & @xmath52 & 12.4 & 1 + + 0150281101 & 2004 - 01 - 16 & 8.9 & @xmath53 & 7.0 & 1 + + 0205230301 & 2004 - 06 - 05 & 11.9 & @xmath54 & 10.0 & 1 + + 0205230401 & 2004 - 08 - 23 & 18.0 & @xmath55 & 14.9 & 1 + + 0205230501 & 2004 - 11 - 23 & 16.0 & @xmath56 & 14.0 & 1 + + 0205230601 & 2005 - 02 - 07 & 14.3 & @xmath57 & 12.4 & 1 + + 0301860101 & 2006 - 03 - 06 & 21.8 & @xmath58 & 19.9 & 1 + + 0405090101 & 2006 - 10 - 15 & 123.1 & @xmath59 & 121.1 & 3 + + 0693850501 & 2012 - 12 - 16 & 125.2 & @xmath60 & 123.0 & 4 + + 0693851201 & 2012 - 12 - 22 & 125.2 & @xmath61 & 123.0 & 4 + + 0722650101 & 2013 - 06 - 08 & 30.7 & @xmath62 & 28.8 & 2 + + * figure 1 : * _ xmm - newton_/epic - pn x - ray ( 0.3 - 10 kev ) image of ngc 1313 , from observation 0693850501 , produced with ds9 .
the dashed circles have a radius of 33@xmath14 and indicate the size of our typical source extraction region .
the dynamical center of the host galaxy , shown as an encircled cross , is based on h i maps by ryder et al .
the north and the east arrows are each 100@xmath14 long .
[ fig : figure1 ] * figure 2 : * background - subtracted epic - pn x - ray ( 0.3 - 10 kev ) light curves of ngc 1313 x-1 ( black ) , and their backgrounds ( red ) during five different _ xmm - newton _ observations .
the pds from these data sets are shown in figures 3 & 4 . for each panel ,
their observation i d and time zero in seconds since 1998.0 tt are indicated at the top . also shown are the are the good time intervals ( gtis ) with duration longer than 500 s. a dashed vertical green line and a solid vertical blue line mark the beginning and end of a gti , respectively .
[ fig : figure1 ] | following the discovery of 3:2 resonance quasi - periodic oscillations ( qpos ) in m82x-1 ( pasham et al .
2014 ) , we have constructed power density spectra ( pds ) of all 15 ( sufficiently long ) _ xmm - newton _ observations of the ultraluminous x - ray source ngc1313x-1 ( @xmath0 @xmath1 2@xmath210@xmath3 erg / sec ) .
we detect a strong qpo at a frequency of [email protected] hz in data obtained on 2012 december 16 .
subsequent searching of all the remaining observations for a 3:2/2:3 frequency pair revealed a feature at [email protected] hz on 2003 dec 13 ( frequency ratio of [email protected] ) .
the global significance of the 0.29 hz feature considering all frequencies between 0.1 and 4 hz is @xmath5 3.5 @xmath6 .
the significance of the [email protected] hz qpo is @xmath5 3.5@xmath6 for a search at 2/3 and 3/2 of 0.29 hz .
we also detect lower frequency qpos ( [email protected] and [email protected] mhz ) .
all the qpos are super - imposed on a continuum consisting of flat - topped , band - limited noise , breaking into a power - law at a frequency of 16@xmath43 mhz and white noise at @xmath7 0.1 hz .
ngc1313x-1 s pds is analogous to stellar - mass black holes ( stmbhs ) pds in the so - called steep power - law state , but with the respective frequencies ( both qpos and break frequencies ) scaled down by a factor of @xmath8 1000 . using the inverse mass - to - high - frequency qpo scaling of stmbhs
, we estimate ngc1313x-1 s black hole mass to be 5000@xmath41300 @xmath9 , consistent with an inference from the scaling of the break frequency
. however , the implied eddington ratio , l@xmath10 @xmath5 [email protected] , is significantly lower compared to stmbhs in the steep power - law state ( l@xmath10 @xmath7 0.2 ) . |
the early experiments on turbulent rayleigh - benard ( rb ) convection in air cells with prandtl number @xmath20 , summarized by davis @xcite , showed a power law increase of the nusselt number @xmath0 with the rayleigh number @xmath2 , namely , @xmath21 with @xmath22 . ''
we mean `` scales as '' throughout the text , _ not _ `` order of magnitude '' .
the prefactors are determined in section iv .
] however , in these early experiments only relatively small rayleigh numbers @xmath23 were achieved .
later , when rb experiments with larger rayleigh numbers and in water cells with @xmath24 were done , the power law exponent @xmath25 turned out to be larger than @xmath26 .
malkus elegant theory of marginal stability , resulting in @xmath27 , seemed to describe those experiments @xcite . in the late 80 s , libchaber et al.s experiments done at the university of chicago on high rayleigh number rayleigh - benard convection in a helium gas cell with prandtl number @xmath20 revealed new and unexpected scaling for the nusselt number as a function of the rayleigh number , namely @xmath28 with @xmath29 @xcite .
the reynolds number @xmath1 , characterizing the wind near the walls , i.e. , the large eddy mean flow , scaled as @xmath30 with @xmath31 @xcite .
these results were reproduced and extended in many experiments and numerical simulations @xcite ; for review articles , which also summarize the results of earlier experimental , theoretical , and numerical work , we refer to refs.@xcite . from all these experiments at first sight
it seems that at least the scaling exponent @xmath32 is very robust .
various theories were put forward to account for the scaling of the nusselt number @xmath0 and the reynolds number @xmath1 as functions of the rayleigh number @xmath2 and the prandtl number @xmath3 .
these include the chicago mixing zone model @xcite and the shraiman - siggia theory @xcite , both reviewed in refs .
the main result of the chicago model is @xmath33 .
the chicago group did not focus on the @xmath3 dependence as only experiments with @xmath20 were done @xcite .
later , cioni et al .
@xcite added the @xmath3 dependence in the spirit of the chicago model and obtained @xmath34 here , @xmath35 refers to the velocity _ fluctuations _ and not to the large scale mean velocity ( often denoted as the `` wind of turbulence '' ) as @xmath1 does .
this prandtl number dependence is only expected to hold for @xmath36 @xcite . on the other hand , the shraiman - siggia model , which assumes a turbulent boundary layer ( bl ) and a thermal boundary layer nested therein ( i.e. , shraiman and siggia implicitly assume a large enough prandtl number ) , states @xmath37 for an extension of the shraiman - siggia theory to position
dependent shear rates see ref .
the prandtl number dependence of the nusselt number resulting from the chicago model @xcite and from the shraiman - siggia theory @xcite are not contradictory , as eq .
( [ nu_chicago ] ) has been suggested for small @xmath3 number fluids and eq .
( [ nu_ss ] ) for the large @xmath3 number case . indeed , in the large @xmath3-limit zaleski @xcite derives the same @xmath3-dependence eq .
( [ nu_ss ] ) as in the shraiman - siggia theory also from the chicago model .
however , both theories are based on rather different assumptions . in recent years , the prandtl number dependence of the nusselt number has been measured and comparison with the two theories @xcite became possible .
the first experiments were done with water and helium rb cells . however , comparing @xmath0 in water and in helium convection only allows for a small variation of the prandtl number . from such experiments a small decrease of @xmath0 with increasing @xmath3 at given @xmath2
was reported @xcite : for @xmath38 belmonte et al .
@xcite measured @xmath39 for @xmath40 and @xmath41 for @xmath42 .
much larger @xmath3 variations are possible if rb convection in mercury or liquid sodium is studied . those experiments with a mercury rb cell ( @xmath43 ) by rossby @xcite , by takeshita et al .
@xcite , and by cioni et al.@xcite and with a liquid sodium rb cell ( @xmath44 ) by horanyi et al .
@xcite reveal that @xmath0 _ increases _ with @xmath3 , which is consistent with the cioni et al .
@xcite extension of the chicago mixing zone model . however , there seems to be indication that also the chicago mixing zone model can not account for all phenomena observed in recent experiments :
one of the most startling observations is that there seems to be a small but significant trend of the scaling exponent @xmath25 as a function of @xmath3 . for @xmath45 ( water )
one has @xmath46 @xcite for @xmath2 up to @xmath47 and an even larger @xmath48 for larger @xmath49 @xcite ; for @xmath50 ( helium gas ) it is @xmath29 @xcite ; for @xmath43 ( mercury ) @xmath51 @xcite and @xmath52 @xcite ( for @xmath53 ) have been measured ; and for @xmath44 ( liquid sodium ) it is @xmath54 @xcite .
those and further experimental results are summarized in table [ tabneu ] .
all exponents between @xmath55 and @xmath56 have been measured ! for thermal convection in a water rb cell ( @xmath57 ) with self similarly distributed balls on the top and bottom wall , the scaling exponent can even be as large as @xmath58 , presumably depending on the ball size distribution @xcite . .
power law exponents @xmath25 of the power law @xmath21 for various experiments .
the experiments were done with different aspect ratios , however , no strong dependence of the scaling exponent @xmath25 on the aspect ratio is expected ( in contrast to the prefactors ) . [
cols="^,^,^,^,^",options="header " , ] ( 11,12 ) ( 0.5,0.5 ) ( 11,12 ) ( 0.5,0.5 ) the increase with @xmath3 seems to cease for @xmath3 between @xmath59 and @xmath60 . as stated above
, other experimental data @xcite even suggest a _ decrease _ of @xmath0 with increasing @xmath3 in that regime .
this is compatible with and explained by the present theory which gives a ( @xmath2 number independent ) transition from the @xmath61 regime with @xmath62 to @xmath63 with @xmath64 , both for fixed @xmath2 . as there is even a controversy in literature
whether @xmath0 for water with @xmath42 or @xmath0 for helium gas with @xmath40 is larger , it is hard to say where exactly the transition from @xmath61 to @xmath63 takes place . according to cioni et al .
@xcite it is fair to say that within experimental accuracy @xmath65 .
we adopt this point of view and use it to calculate the transitional @xmath3 number @xmath66 to be about @xmath67 .
this experimental information thus defines the line @xmath68 in the phase space and separates the lower subregime @xmath61 with @xmath69 from the upper subregime @xmath63 with with @xmath70 .
next , we compare the predicted scaling exponent @xmath25 of @xmath0 vs @xmath2 , which in regime i is according to the presented theory the same for the lower and the upper subregime . for small @xmath3 ( mercury , sodium ) and ( relatively ) small @xmath2 the theoretically obtained value @xmath22 for the scaling exponent of @xmath0 vs @xmath2 has been measured in several experiments : @xmath71 in @xcite , @xmath72 in @xcite , and @xmath73 in @xcite .
also regime @xmath74 ( intermediate @xmath2 , low @xmath3 ) with @xmath75 seems to have been observed by cioni et al .
@xcite . for larger @xmath3 ( helium , water )
the measured scaling exponent is larger , @xmath76 , see table [ tabneu ] .
here we argue that this results from the superposition of the scaling in regime i , with those in the regimes @xmath77 and @xmath78 . to substantiate this , we plot the expected @xmath0 vs @xmath2 dependence for @xmath79 , nu = 0.27 ra^1/4 + 0.038 ra^1/3 , [ eq2733 ] in figure [ nu_vs_ra_log_2733 ] . here
we have already made use of the prefactors from table [ tab_scal_laws ] , which will be calculated in the next section .
now we fit eq .
( [ eq2733 ] ) with _ one _ power law in as large a regime as @xmath80 .
this fit which is nearly indistinguishable from the superposition ( [ eq2733 ] ) reads nu = 0.22 ra^0.289 .
[ eq27allein ] the power law exponent is very close to @xmath81 and definitely consistent with the experimental data of table [ tabneu ] . changing the fit regime of course changes the exponent of the power law ( [ eq27allein ] ) .
e.g. , for a linear regression in the regime @xmath82 which is typical for many experiments one obtains @xmath83 , i.e. , an exponent which is even closer to @xmath84 . by plotting compensated plots or local slopes @xmath85 as done in figure [ local_slope ] for eq .
( [ eq2733 ] ) one may be able to get hints that there is no pure power law .
note , that on first sight a compensation with @xmath86 may erroneously even be considered as `` better '' .
chavanne et al .
@xcite find hints for a transition to a regime with a visibly larger scaling exponent .
according to our theory this could be regime @xmath78 or @xmath87 or @xmath88 or , most likely , a mixture of all of them .
no clean scaling exponent could hitherto be determined experimentally .
one reason is that in that regime both @xmath2 and @xmath3 change .
we will discuss the possible nature of this transition below .
also figure 3 of ref .
@xcite suggests such a transition .
we now turn to the large @xmath3 regime .
there are very few data for large @xmath89 .
recently , ashkenazi and steinberg @xcite performed convection experiments with sf@xmath90 close to its critical point . in these experiments
both @xmath2 and @xmath3 change considerably at the same time . to
what degree the rb convection is still boussinesq close to the critical point is extensively discussed in ref .
@xcite .
ashkenazi and steinberg obtain @xmath91 in @xmath92 and @xmath93 .
based on the phase diagram [ fig_sketch ] we judge that for these ranges of the @xmath2 and @xmath3 numbers we should be in regimes @xmath63 and @xmath78 . from table [ tab_scal_laws ] we see that the @xmath2 exponent of @xmath0 is @xmath26 and @xmath94 , respectively .
the measured exponent of @xmath95 in between is consistent with this .
the @xmath3 exponent of @xmath0 is expected to be in between @xmath96 and @xmath97 , slightly smaller ( modulus - wise ) than the value of @xmath98 reported in @xcite .
the results of this subsection clearly demonstrate the importance of superimposing the power laws of adjacent phase space regimes .
this can really mimick different scaling behavior , as demonstrated in figures [ nu_vs_ra_log_2733 ] and [ fig_re_ra ] .
this characteristic feature holds because the power law exponents of the neighboring regimes are rather similar
. they will commonly show up if data in a crossover range are examined .
we emphasize that these crossover ranges appear to be rather extended , reaching well into the corresponding regimes , due to the small differences of the scaling exponents .
therefore , rather than writing pure power laws , one should allow for superpositions .
table [ tab_scal_laws ] suggests nu ~ra^1/4 pr^1/8 ( 1 + \ { cc c_iii_u & ra^5/28 pr^-15/56 + + c_iv_u & ra^1/12 pr^-1/8 + + c_iv_l & ra^1/4 pr^3/8 + + c_ii_l&ra^-1/20pr^3/40 + . ) [ eq35 ] and re ~ra^1/2 pr^-3/4 ( 1 + \ { ll c^_iii_u & ra^1/14 pr^-3/28 + + c^_iv_u & ra^-1/18 pr^1/12 + + c^_iv_l & pr^1/4 + + c^_ii_l&ra^-1/10pr^3/20 + . ) , [ eq36 ] respectively . here
, we have separated the exponents of regime @xmath61 .
which of these corrections and how many are to be taken depends on the @xmath3 number and on the aspect ratio .
_ locally _ , i.e. , for a limited @xmath2 range , the suggested @xmath0 vs @xmath2 power law exponents @xmath99 @xcite ( cf . figure 2 of ref .
@xcite or figure 3 of @xcite ) , or @xmath100 @xcite can still be considered as an appropriate representation of the experimental data .
_ globally _ , for larger @xmath2 intervals , however , we claim that eq .
( [ eq35 ] ) is a better description . in previous publications @xmath0 , compensated by the expected scaling @xmath86 , was plotted against @xmath2 in a log - log plot , see figure 2 of ref .
@xcite . from that plot
one realizes that the 2/7-scaling is slightly too steep between @xmath101 and @xmath102 and not steep enough beyond the crossover at @xmath103 .
the analogously compensated plot with the expected scaling ( [ eq29 ] ) is shown in figure [ fig_nu_vs_ra ] . now ( for @xmath20 ) one obtains a horizontal line up to @xmath104 , showing that eq .
( [ eq29 ] ) nicely agrees with the data .
however , beyond @xmath104 one observes deviations .
we suggest that these corrections originate from the different scaling in the neighboring regime @xmath78 .
the reason that it is regime @xmath78 ( and not regime @xmath77 ) is that in the chavanne et al .
experiments the large @xmath2 measurements also have large @xmath3 ; the trajectory in control parameter space @xmath105 is not a straight line . at @xmath106 one
typically has @xmath20 , but at @xmath107 chavannve et al.typically have @xmath108 .
the power law exponent @xmath109 , following from table [ tab_scal_laws ] , is consistent with the experimental data for large @xmath2 , see figure [ fig_nu_vs_ra ] . for the mercury data of cioni et al .
@xcite we do not have such a complication as @xmath43 is roughly constant for all chosen @xmath2 .
as plotted in figure [ fig_nu_vs_ra ] , lower curve , we observe a straight line up to about @xmath110 and then a decay , signaling contributions from regime @xmath74 . the power law exponent @xmath111 of the correction term in eq .
( [ eq35 ] ) is consistent with the data shown in figure [ fig_nu_vs_ra ] .
a more stringent way to test the superpositions of type ( [ eq35 ] ) is to make a _ linear _ plot @xmath112 vs @xmath113 or vs @xmath114 or vs @xmath115 , etc , depending on which neighboring regime the corrections originate .
this is done in figure [ fig_leo ] , assuming , as argued above , that the most relevant corrections originate from regime @xmath78 .
if the theory is correct , the data points must fall on a straight line .
indeed , they do so with satisfying precision .
note that this kind of linear plot is very sensitive to what combinations of @xmath2 and @xmath3 are chosen as @xmath116 and @xmath117 axes .
e.g. , plotting @xmath112 vs @xmath115 ( the subleading correction characterizing regime @xmath87 ) does not lead to a straight line at all , see the inset of figure [ fig_leo ] .
clearly the variable @xmath113 on the abszissa is superior , adding confirmation that the chavanne et al .
@xcite large @xmath2 experiments represent the physics of regime @xmath78 .
we now consider the experimental values for the scaling exponents @xmath118 of the reynolds number vs the rayleigh number . for @xmath24 ( water )
xin et al .
@xcite find @xmath119 and qiu and xia @xcite find @xmath120 .
both experiments where done in the ra interval between @xmath110 and @xmath121 , i.e. , in regime @xmath63 , where exactly this power law exponent @xmath122 is expected . for @xmath20 both castaing et al .
@xcite and chavanne et al .
@xcite find @xmath123 for all @xmath2 which suggests that possibly the regimes @xmath61 and @xmath87 are seen where this value is predicted , or also regime @xmath78 , where the exponent is only slightly higher ( @xmath124 ) . for @xmath125 ( mercury ) cioni et al .
@xcite find @xmath126 from a fit to all available @xmath2 .
this value is in between the derived values @xmath122 in regime @xmath61 and @xmath127 in regime @xmath74 .
we compare this experimental finding @xmath128 ( based on a fit to the data in the range up to @xmath129 @xcite ) with the @xmath61-@xmath74 superposition according to table [ tab_scal_laws ] re= 0.59 ra^1/2 + 4.30 ra^2/5 , [ superpos_re ] cf .
[ fig_re_ra ] . in this relatively short @xmath2 interval
the theoretical superposition ( [ superpos_re ] ) is again hardly distinguishable from its straight line fit re=3.5 ra^0.446 [ slf_re ] whose exponent reasonably well agrees with the measured one .
moreover , also the theoretically obtained @xmath3 number dependence of @xmath1 very nicely agrees with available experimental information : chavanne et al .
@xcite did experiments with ( slightly ) varying @xmath3 .
they then plotted @xmath130 vs @xmath2 and obtained the law @xmath131 .
the exponent @xmath132 of the prandtl number was determined by minimizing the scattering of points around a straight line in the log - log plot .
it very well agrees with the calculated @xmath3 scaling exponent 3/4 in eq .
( [ eq30 ] ) ( regime @xmath61 ) .
a possible reason for the slight difference between theory and data fit is that part of the experimentally realized @xmath2 and @xmath3 already belong to regimes @xmath78 and @xmath77 , where according to eqs .
( [ eq22 ] ) and ( [ eq22u ] ) the expected @xmath3 scaling exponent is @xmath133 and @xmath134 , respectively .
however , the deviation is clearly within the experimental uncertainty . in figure [ fig_re_he ]
we replot the experimental reynolds number data of chavanne et al .
@xcite . in fig .
[ fig_re_he]a we show @xmath135 vs @xmath2 .
the fit gives a @xmath2-exponent @xmath136 in very good agreement with the theory s exponent @xmath137 . in fig .
[ fig_re_he]b we display @xmath138 vs @xmath3 .
the data fit results in a @xmath3-exponent @xmath139 , also in excellent agreement with the theoretical expectation which is @xmath140 in @xmath61 and @xmath141 in @xmath63 .
the large @xmath2 values in both figures will turn out to belong already to the regimes @xmath78 and @xmath77 , where the @xmath3-scaling is similar .
the available range is too small to perform a more detailed comparison .
the only large @xmath3 data available are again those by ashkenazi and steinberg @xcite .
they obtain @xmath142 in @xmath143 and @xmath144 .
the expected @xmath2 exponent of @xmath1 is between @xmath137 and @xmath124 , distinctly larger than the measured one of @xmath145 .
similarly , the theoretically expected @xmath3 exponent of @xmath1 is between @xmath141 and @xmath146 , also larger than the measured exponent @xmath147 .
we have no explanation .
( 11,12 ) ( 0.5,0.5 ) ( 11,12 ) ( 0.5,0.5 ) ( 18,18 ) ( 4.0,0.5 ) ( 4.75,9.0 ) ( 1.0,5.5 ) ( 1.0,13.5 )
to obtain the values of the prefactors in the power laws within the presented scaling theory , we need further input from experiment . but what data to choose ? as pointed out in the introduction , a huge variety of data is around , often disagreeing with each other even in the scaling exponents , not to speak of prefactors .
those often vary by as much as @xmath148 from experiment to experiment , even for cells with the same aspect ratio .
another reason which makes the adoption to one experiment and the comparison to others difficult is that different definitions are used for the reynolds numbers .
e.g. , cioni et al .
@xcite define the reynolds number as @xmath149 , where @xmath150 is a distinguished frequency at the small frequency edge in the temperature spectrum .
chavanne et al .
@xcite define @xmath151 , where @xmath152 is a typical frequency in the cross correlation spectrum of two temperature signals , measured at a vertical distance of @xmath153 mm and @xmath67 cm off the axis of the cell . in spite of these difficulties
, we decided to calculate the prefactors of the suggested scaling laws by employing the following choice for the input information as a reasonable , realistic example .
our reasons are first , to be able to draw a phase diagram with more or less realistic values .
second , to stress the importance of the prefactors .
but we caution the reader that our input choice is somehow arbitrary ; other possibilities can equally well be rationalized , sometimes shifting the various regime boundaries considerably . above , as an input from experiment ,
we had already chosen @xmath154 as the prandtl number for which @xmath0 is maximal ( for fixed @xmath155 , cf . fig .
[ fig_nu_vs_pr ] ) .
in addition , we will use the following experimental information : 1 .
the observed transition rayleigh number for the transition from regime @xmath61 to regime @xmath156 , @xmath157 at @xmath79 @xcite ; 2 .
the observed transition rayleigh number for the transition from regime @xmath61 to regime @xmath74 , @xmath158 at @xmath43 @xcite ; 3 .
the experimental values , taken from ref .
@xcite , for the reynolds and the nusselt number at the middle point @xmath159 in the phase diagram figure [ fig_sketch ] ; 4 . and the prefactor 0.0372 of the scaling law @xmath160 measured in regime @xmath61 @xcite .
information ( 1 ) specifies the prefactor of the rhs of eq .
( [ eq32 ] ) ( which we call @xmath161 ) to be @xmath162 .
information ( 2 ) gives the prefactor of the rhs of eq .
( [ eq31 ] ) ( which we call @xmath163 ) to be @xmath164 .
the two curves ( [ eq31 ] ) and ( [ eq32 ] ) cross at ( ra_m , pr_m ) = ( 1.03 10 ^ 11 , 0.94 ) .
[ n1 ] this middle point @xmath165 is defined by the conditions @xmath166 and @xmath167 .
equation ( [ n1 ] ) specifies the prefactors of the rhs of eqs .
( [ eq33 ] ) and ( [ eq34 ] ) to be @xmath168 and @xmath169 , respectively .
we see that @xmath170 so that there is a regime @xmath156 , but no regime @xmath171 , as already anticipated .
the line @xmath172 hits the boundary between @xmath173 and @xmath174 at ( ra_m , pr_m ) = ( 7.1 10 ^ 10 , 2.0 ) , [ mprime ] which fixes the prefactors of eqs.([eq32u ] ) and ( [ tll3 ] ) to be @xmath175 and @xmath176 , respectively . correspondingly , one obtains the prefactors of eqs .
( [ eq33u ] ) and ( [ tll4 ] ) to be @xmath177 and @xmath178 , respectively .
these are the data on which the phase diagram figure [ fig_sketch ] is based ; they are summarized in table [ tab_bounds ] .
apart from regime @xmath156 all regimes turned out to have at least one decade of extension both in @xmath2 and in @xmath3 and should therefore in principle be visible .
however , we should always expect one or more subleading corrections .
regime @xmath156 will clearly not be detectable .
we again stress how dependent this phase diagram drawn in figure [ fig_sketch ] is on the choice of experimental information .
e.g. , if we had adopted glazier et al.s @xcite point of view that there is no transition towards a steeper @xmath2 dependence of @xmath0 at least up to @xmath179 , regimes @xmath180 and @xmath181 would have shifted further to the right or would even not exist at all . but as shown in figure [ nu_vs_ra_log_2733 ] , apparent smooth scaling behavior does _ not _ allow to exclude a transition .
making use now of the experimental information ( 3 ) and ( 4 ) we can calculate the prefactors in the power laws for @xmath0 and @xmath1 . from figs . 2 and 3 of chavanne et al.s work @xcite we can extract the reynolds and nusselt numbers at the middle point @xmath165 of the phase diagram which touches all four regimes , namely re_m = 1.20 10 ^ 4,nu_m = 2.78 10 ^ 2 , [ n2 ] which is information ( 3 ) above .
the definition of the middle point @xmath165 , i.e. , the conditions @xmath166 and @xmath167 , allows to calculate the prefactors @xmath182 , @xmath183 , and @xmath184 on the rhs of eqs .
( [ eq17 ] ) , ( [ eq18 ] ) , and ( [ eq19 ] ) , respectively .
one obtains c__u , bulk = nu_m ra_m 2 pr^2_m re_m^3 = 9.38 , [ n3 ] c _ _ , bulk = nu_m 2 pr_m re_m = 0.0123 , [ n4 ] c__u , bl = nu_m ra_m 2 pr^2_m re_m^5/2 = 1028 .
[ n5 ] finally , the prefactor @xmath185 on the rhs of relation ( [ eq26 ] ) is adopted to chavanne et al.s @xcite experimentally determined prefactor ( see figure 3 of that paper ) in the relation ( [ eq30 ] ) , i.e. , @xmath186 valid throughout regime @xmath61 ( information ( 4 ) above ) . in that regime @xmath187 ; with eqs.([eq11 ] ) , ( [ eq19 ] ) , and ( [ n5 ] ) we get c_nu = ( 3.72 10 ^ -2)^2 c__u , bl = 1.42 . [ n6 ] the prefactors referring to the upper halve of the phase diagram are calculated from the matching conditions for @xmath1 and @xmath0 on the @xmath188 line . with eqs .
( [ n3 ] ) ( [ n6 ] ) and the matching conditions now _ all _ prefactors of the power laws in the four different regimes are determined .
we have summarized all these power laws in table [ tab_scal_laws ]
. note that from the condition @xmath189 for @xmath190 the prefactor of eq .
( [ eq14 ] ) is also obtained .
it is @xmath55 , i.e. , @xmath191 .
also the prefactors to the @xmath192- and the @xmath193-borders of validity and the crossover lines to turbulence of the laminar bl flow automatically follow , and are included into table [ tab_scal_laws ] .
we now would like to compare the absolute agreement of the power laws summarized in table [ tab_scal_laws ] , whose prefactors result from an adoption to above four pieces of experimental information , with _ further _ experimental data .
we first focus on the chavanne et al.s @xcite rb measurements in helium gas . from the previous section we know that in these experiments due to the large @xmath3 at large @xmath2 it is mainly regime @xmath78 which causes additional contributions to regime @xmath61 .
therefore , we have plotted the superposition @xmath194 with @xmath195 into figure [ fig_nu_vs_ra ] .
we again stress that in the experiments @xmath3 is not constant at all .
nevertheless , the data are satisfactorily described .
note that the solid curve in figure [ fig_nu_vs_ra ] is no fit ! a much better way to check whether the obtained prefactors of the theory agree with the measured ones is to do a linear regression of the straight line as offered in figure [ fig_leo ] .
such a straight line fit gives @xmath196 .
the found prefactors are in good agreement with the expectation @xmath197 from table [ tab_scal_laws ] .
we also compare the theoretical prefactors of the reynolds number scaling in regime @xmath61 with experiment .
as the theoretical prefactors have been adopted to the experimental @xmath135 vs @xmath2 power law , it only makes sense to check the prefactors in @xmath138 vs @xmath3 . from table [ tab_scal_laws ]
the expected slope is @xmath140 and the expected prefactor @xmath198 .
linear regression gives a slope of @xmath139 and a prefactor of @xmath199 , cf .
figure [ fig_re_he ] .
next , we compare the experimental @xmath0 vs @xmath2 scaling for mercury with @xmath43 with theory . the measured relations
@xmath200 @xcite , @xmath201 @xcite , and @xmath202 @xcite are all in reasonable agreement with the regime @xmath61 expectation @xmath203 from table [ tab_scal_laws ] . the same holds for a comparison in regime @xmath74 : the reported experimental fit is @xmath204 @xcite , theory gives @xmath205 , again , remarkable agreement of both the power law exponent _ and _ the prefactor .
remember that the only experimental input from this experiment into the theory is @xmath206 .
let us also check the prefactors of @xmath0 as a function of @xmath3 for fixed @xmath101 .
a power law fit to all available experimental data points @xcite included in figure [ fig_nu_vs_pr ] gives @xmath207 which is in agreement with the theoretical expectation @xmath208 in regime @xmath61 .
leaving out the data points for water ( @xmath209 ) which strictly speaking already belongs to regime @xmath63 gives a slightly larger power law exponent and a slightly larger prefactor , @xmath210 . both exponent and prefactor are consistent with the theoretical expectation .
as stated above , the present theory also gives the absolute widths of the thermal and the laminar viscous boundary layers , _
= 0.5 l nu^-1,[newlamth ] _ u = 0.25 l re^-1/2 .
[ newlamu ] the results for the widths of the bls for @xmath43 ( mercury ) and @xmath211 ( water ) are shown in figure [ fig_bl ] . in both cases
there are three regimes involved , namely , @xmath61 , @xmath74 , and @xmath87 for @xmath43 and @xmath63 , @xmath78 , and @xmath77 for @xmath211 .
as expected , for the larger @xmath3 numbers the thermal boundary layer is always nested in the viscous one which agrees with the experimental observations @xcite .
for the lower @xmath3 numbers it is the other way round .
if the laminar bl becomes turbulent , the blasius estimate @xmath212 for its width must be replaced by the thickness of the turbulent bl .
to give an idea about this length scale , we calculate the width @xmath213 of the viscous sublayer of the turbulent bl within the prandtl theory @xcite , applied to couette flow . in the large @xmath1 limit it holds @xcite = 1.38 ( k^2 re)k^2 re .
[ eqy0 ] here , @xmath214 is the experimentally known van krmn constant .
we have included @xmath215 in figure [ fig_bl ] for the relevant large @xmath2 .
the turbulent viscous sublayer is thiner than the laminar bl .
one also notes that for the larger prandtl number @xmath209 ( water ) shraiman and siggia s assumption @xcite of the thermal boundary layer being nested in the turbulent one is just fulfilled . for lower @xmath3
this is not the case any more .
( 11,12 ) ( 0.5,0.5 ) many experiments justify the identification of the thermal bl width with the inverse nusselt number , @xmath216 . for a detailed discussion
we refer to the review articles or to belmonte et al .
@xcite or to the more recent work by lui and xia @xcite .
the situation is more complicated for the width of the kinetic bl @xmath217 .
its measurement is experimentally difficult .
moreover , experimental results on @xmath217 seem to exist only for regime i. belmonte et al .
@xcite tried to measure @xmath217 in an indirect way , namely , through the detection of a spectral cutoff frequency in gas convection which in water convection ( at one @xmath2 ) is found to have a similar dependence on the height @xmath218 in the rb cell as the velocity profile @xmath219 .
for @xmath20 they found @xmath220 in the regime @xmath221 and @xmath222 in @xmath223 .
we have no idea about the origin of the measured scaling exponents .
more recently , xin et al .
@xcite , xin and xia @xcite , and qiu and xia @xcite measured the thickness of the kinetic bl in a water cell in a more direct way .
they define @xmath217 as the distance from the wall at which the extrapolation of the linear part of the velocity profile @xmath219 equals the maximum velocity @xmath224 , the velocity of the large scale wind . in the interval
@xmath225 they find @xmath226 for the thickness of the top and bottom kinetic bl @xcite and @xmath227 for the thickness of the kinetic bls at the side walls @xcite .
the first exponent ( for the top and the bottom plates ) is different from the value of this theory @xmath228 .
we can only speculate on the origin of this discrepancy .
perhaps , if @xmath217 is defined as the distance of the velocity maximum to the wall , the @xmath2-scaling would be different .
the power law for the thickness of the kinetic bls at the side walls , however , is well in agreement with the expectation @xmath228 . in any case
, the experimentally found very weak dependence of @xmath217 on @xmath2 supports the assumed _ laminar _ nature of the kinetic bl .
if the width of the bl were identified with the width @xmath213 of the viscous sublayer of a _ turbulent _ bl , one would expect a stronger dependence on @xmath1 , namely @xmath229 @xcite , i.e. , when neglecting logarithmic corrections , one would have @xmath230 . according to the presented theory with the chosen prefactors the breakdown of laminarity in the shear bl happens at @xmath231 for @xmath79 and @xmath232 for @xmath43 .
these values are calculated from table [ tab_bounds ] .
hitherto , there are no experiments for these regimes .
however , one may want to argue that the transition to a turbulent shear bl may already occur earlier , be it because of a different aspect ratio , or , in view of more recent work @xcite , because the critical reynolds number ( which we had assumed to be 420 , cf .
( [ s2 ] ) ) is smaller , or because of a different choice of the experimental input information to which the prefactors of the theory are adopted .
e.g. , if one assumes a laminar shear layer with width @xmath233 as suggested by landau and lifshitz ( ref .
@xcite , section 39 ) for the ( related ) case of a flat plate shear flow , the transition to turbulence in the shear bl already occurs at @xmath234 ( for @xmath235 ) .
we note that this is just at about that @xmath2 number where a marked transition in the ( thermal ) dissipative spectral power has been measured by a probe placed in the bl @xcite .
this transition was towards a weaker increase with @xmath2 . in the context of this section
we also interpret the above mentioned recent experiment by ciliberto et al .
@xcite in which the boundary layers are disturbed by constructing a rough bottom plate with a mean roughness comparable to the thermal boundary layer thickness .
the experiments are performed in water . according to the theory of this paper one would expect larger bulk contributions to both the thermal and the kinetic dissipation and therefore an earlier onset of regime @xmath236 and @xmath237 .
indeed , experimentally the increase of @xmath0 with @xmath2 is found to be much steeper . for the experiment described by ciliberto et al .
@xcite the data can be fitted to power laws @xmath238 or @xmath239 , depending on the features of the rough bottom and upper plate .
we summarize the central ingredients of the theory presented in this paper : the scaling laws for the nusselt number and the reynolds number are based on the decomposition of the global thermal and kinetic energy dissipation rates into their bl and bulk contributions .
these in turn are estimated from the dynamical equations , taking the wind @xmath240 as the relevant velocity in the heat conduction cell .
the resulting estimates are inserted into the rigorous relations eqs .
( [ eq11 ] ) and ( [ eq12 ] ) for the global kinetic and thermal energy dissipation , respectively .
four regimes arise , depending on whether the bulk or the bl contributions dominate the two global dissipations .
each of the four regimes in principle divides into two subregimes , depending on whether the thermal bl ( of width @xmath241 ) or the kinetic bl ( of width @xmath217 ) is more extended .
in addition to these main regimes there is a range for very large @xmath3 numbers in which the wind reynolds number is @xmath242 ; here the whole flow is viscosity dominated , and the theory looses its applicability .
there also is the range of very small @xmath3 numbers in which the @xmath0 goes down to @xmath243 , and again the theory no longer holds .
finally , for large @xmath2 the laminar kinetic bl becomes turbulent . beyond turbulence onset
we feel the flow is bulk dominated .
all scaling exponents follow from this theory .
if one in addition introduces only four pieces of experimental information , also all the prefactors can be determined .
therefore the theory has predictive power not only for the power law exponents but also for the prefactors . these , however , depend on the chosen experimental information input . to nail the prefactors more input information for various aspect ratios
is necessary .
the phase diagram of the theory , the main result of this work , is shown in figure [ fig_sketch ] . the power laws with the prefactors based on the chosen experimental information are summarized in table [ tab_scal_laws ] , the power laws of the boundaries between the different regimes in table [ tab_bounds ] .
a detailed comparison of the theoretical power law exponents _ and _ the prefactors with the experimental data gives reasonable and encouraging agreement .
we emphasize that to accurately account for the dependences of @xmath0 and @xmath1 on @xmath2 and @xmath3 single power laws are often not sufficient , as additive corrections from neighboring regimes can be considerable .
this can be viewed as one of the main insights obtained in this paper .
a particularly striking example is that @xmath19 mimicks a @xmath84 power law scaling over at least _
nine _ order of magnitude in @xmath2 , see figure [ nu_vs_ra_log_2733 ] .
the theory also offers a possible explanation why a transition to a steeper increase of @xmath0 vs @xmath2 is seen in the chavanne et al .
data @xcite , but not in the chicago group data @xcite .
it may be that in the large @xmath2 chicago experiments the @xmath3 number was smaller than in the chavanne et al .
@xcite measurements .
then for the chicago data one had to expect a transition towards regime @xmath77 where the @xmath2 scaling exponent is @xmath244 and thus , as demonstrated , in superposition with the leading @xmath26 exponent , is indistinguishable from a @xmath84 scaling . in the chavanne et al .
@xcite experiments , on the other hand , one has the transition to the large @xmath3 number regime @xmath78 where the @xmath2 scaling exponent is @xmath94 which is much better distinguishable from @xmath26 .
whether this explanation is true remains to be seen .
finally , we want to stress and discuss one of the basic assumptions of the theory , namely , that a large scale `` wind of turbulence '' exists , defining the reynolds number @xmath245 , creating a shear bl and stirring the turbulence in the bulk .
clearly , this assumptions breaks down in the shaded area in figure [ fig_sketch ] beyond the line @xmath246 where the flow is viscosity dominated .
but even below this line we do not exclude that the convection rolls break down and that the heat is exclusively transported by the fluctuations .
e.g. , for a water cell ( @xmath24 ) tanaka and miyata @xcite do not note the wind of turbulence , in contrast to zocchi et al .
@xcite , who do observe the wind of turbulence in their experiment with only a slightly lower aspect ratio .
also all latest experiments with various aspect ratios doubtlessly detect the wind of turbulence @xcite whose existence we therefore consider as a weak and in particular controllable assumption .
the present theory does _ not _ make any statement _ how _ the heat is transported from the bottom to the top , i.e. , whether it is mainly convective transport or mainly transport through plumes @xcite . both processes may contribute , as both create thermal and viscous dissipation . for even larger prandtl numbers @xmath247 , the spontaneous formation of a wind of turbulence may seem more and more unlikely . to initiate such a wind so that the presented theory can be applied and results can be compared we suggest to slightly tilt the rb cell , thus breaking the symmetry and creating a preferred direction for the wind of turbulence .
if this wind of turbulence can be created , we are confident that the suggested theory holds .
presently , we do not have any experimental information which contradicts the theory .
however , for further verification more experiments will be valuable
. * acknowledgments : * d. l. gratefully acknowledges s. ciliberto s hospitality during his visit in lyon .
it was him who renewed our interest in rayleigh - benard convection .
we thank him and s. cioni and x. chavanne for various discussions and also for supplying us with their experimental data and allowing us to reproduce them .
we also thank g. de bruin , l. kadanoff , v. steinberg , and a. tilgner for hints and discussions . support for this work by the deutsche forschungsgemeinschaft ( dfg ) under the grant lo 556/3 - 1 , by the german - israeli foundation ( gif ) , and by the chicago mrsec - grant is also acknowledged .
this work is part of the research programme of the `` stichting voor fundamenteel onderzoek der materie ( fom ) '' , which is financially supported by the `` nederlandse organisatie voor wetenschappelijk onderzoek ( nwo ) '' . | a systematic theory for the scaling of the nusselt number @xmath0 and of the reynolds number @xmath1 in strong rayleigh - benard convection is suggested and shown to be compatible with recent experiments .
it assumes a coherent large scale convection roll ( `` wind of turbulence '' ) and is based on the dynamical equations both in the bulk and in the boundary layers .
several regimes are identified in the rayleigh number @xmath2 versus prandtl number @xmath3 phase space , defined by whether the boundary layer or the bulk dominates the global kinetic and thermal dissipation , respectively .
the crossover between the regimes is calculated . in the regime
which has most frequently been studied in experiment ( @xmath4 ) the leading terms are @xmath5 , @xmath6 for @xmath7 and @xmath8 , @xmath9 for @xmath10 . in most measurements these laws
are modified by additive corrections from the neighboring regimes so that the impression of a slightly larger ( effective ) @xmath0 vs @xmath2 scaling exponent can arise .
the most important of the neighboring regimes towards large @xmath2 are a regime with scaling @xmath11 , @xmath12 for medium @xmath3 ( `` kraichnan regime '' ) , a regime with scaling @xmath13 , @xmath14 for small @xmath3 , a regime with @xmath15 , @xmath16 for larger @xmath3 , and a regime with scaling @xmath17 , @xmath18 for even larger @xmath3 .
in particular , a linear combination of the 1/4 and the 1/3 power laws for @xmath0 with @xmath2 , @xmath19 ( the prefactors follow from experiment ) , mimicks a 2/7 power law exponent in a regime as large as ten decades . for very large
@xmath2 the laminar shear boundary layer is speculated to break down through nonnormal - nonlinear transition to turbulence and another regime emerges .
the presented theory is best summarized in the phase diagram figure [ fig_sketch ] .
-1 cm u v |
checkpoint - restart is now a mature technology with a variety of robust packages @xcite .
this work concentrates on the dmtcp ( distributed multithreaded checkpointing ) package and its sophisticated plugin model that enables process virtualization @xcite .
this plugin model has been used recently to demonstrate checkpointing of 32,752 mpi processes on a supercomputer at tacc ( texas advanced computing center ) @xcite .
dmtcp itself is free and open source .
the dmtcp publications page @xcite lists approximately 50 refereed publications by external groups that have used dmtcp in their work .
this work concentrates on the recent advances in the dmtcp programming model that were motivated by work with intel corporation .
while intel works with multiple vendors of hardware emulators , this work reflects the three - way collaboration between the dmtcp team , intel , and mentor graphics , a vendor of hardware emulators for eda .
further information specific to eda ( electronic design automation ) is contained in @xcite . in particular ,
the ability to save the state of a simulation _ including the state of a back - end hardware emulator _ is a key to using checkpoint - restart in eda . for background on how dmtcp is used generally at intel ,
see @xcite .
the focuses of the ongoing work at intel is best described by their statement of future work : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ `` within intel it , we will focus on the development and enhancement of the dmtcp technology for use with graphical eda tools , with strong network dependencies .
@xmath0 there is also additional engagement with third - party vendors to include native dmtcp support in their tools , as well as engagement with super - computing development teams on enabling dmtcp for the xeon phi family of products . ''
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ a hardware emulator may entail a thousand - fold slowdown , as compared to direct execution in silicon .
there are two natural use cases of checkpointing in the context of eda . in both cases ,
the natural strategy is to run until reaching the region of logic of interest . then checkpoint .
later , one can repeatedly restart and test the logic , without worrying about the long initialization times under a hardware emulator .
restarting under dmtcp is extremely fast , especially when the fast - restart flag is used that takes advantage of mmap ( ) to load pages into memory on - demand at runtime ( after the initial restart ) .
the two use cases follow .
run until reaching the logic to be tested ; then repeatedly restart and follow different logic branches ; and run until reaching the logic to be tested ; then repeatedly restart , inject faults in the emulated ( or simulated ) silicon model and run along a pre - determined logic branch to determine the level of fault tolerance for that silicon design . for this work ,
the second case is of greater interest .
this requires running arbitrary code either immediately at the point of restart by injecting faults in the logic design , or by interposing on later logic functions of the simulator / emulator so as to inject transient faults .
the first use case above has been extensively studied using dmtcp in domains as varied as architecture simulation @xcite , formal verification of embedded control systems @xcite , network simulation @xcite , and software model checking @xcite .
while the two use cases are closely related , this work highlights the second use case , by including the possibility of interposing at runtime .
section [ sec : processvirtualization ] presents the tools for such interposition , including the creation of global barriers at an arbitrary point in the program .
section [ sec : casestudies ] presents three particular extensions of checkpointing that were added to the dmtcp plugin model specifically motivated by the concerns observed in our general collaboration on eda .
the dmtcp plugin model is critical in this latter application .
one must stop a computation at a pre - defined location in the simulation , save additional state information ( such as the state of a hardware emulator being used @xcite ) , and then inject additional code ( such as fault injection ) at restart time .
a contribution of the dmtcp plugin model is the ability to virtualize multiple aspects of the computation .
these include : pathnames ( for example , the subdirectory corresponding to the current `` run slot '' of the emulator ) ; environment variables ( for example , modification of the display environment variable , or other environment variables intrinsic to the running of the simulation ) ; interposition of the simulation by a third - party plugin ( for example , for purposes of measuring timings since restart at multiple levels of granularity , or programmatically creating additional checkpoints for analysis of interesting states leading to logic errors ) ; and third - party programmable barriers across all processes ( enabling the acceleration of simulations through the use of parallel processes and even distributed processes within a single computation ) .
the dmtcp plugin model is an example of _ process virtualization _ : virtualization of external abstractions from within a process .
it is argued here that the dmtcp plugin model sets it apart from other checkpointing approaches . to this end , a brief survey of existing checkpointing approaches and process virtualization is provided at the end . in the rest of this paper , section [ sec : processvirtualization ]
motivates the need for a model of process virtualization with a simple example concerning process ids .
it also reviews the dmtcp plugin model .
section [ sec : casestudies ] presents a series of micro - case studies in which dmtcp was extended to support the applications at intel , along with third - party dmtcp plugins developed by mentor graphics for use by intel and other customers ..
section [ sec : relatedwork ] the provides a survey of dmtcp and some other related approaches to checkpointing and process virtualization .
section [ sec : conclusion ] then presents the conclusions .
application - specific checkpointing and system - level transparent checkpointing are two well - known options for checkpointing .
unfortunately , neither one fits the requirements for the proposed use case for simulating fault injection in silicon logic .
application - specific checkpointing is error - prone and difficult to maintain .
system - level transparent checkpointing generally does not provide enough control at runtime to dynamically adjust the type of fault injection .
in particular , it is often necessary to capture control of the target application dynamically at runtime in order to inject faults .
here we show how that can be incorporated in a modular dmtcp plugin , rather than incorporated directly into the simulator / emulator . for a more thorough introduction to the dmtcp plugin model ,
see either @xcite or the dmtcp documentation @xcite .
this section highlights those aspects most likely to assist in adding fault injection through a dmtcp plugin .
the primary features of the model of interest for fault injection are : 1 .
interposition on function / library calls , and their use in virtualization ; 2 .
programmatically defined barriers across all processes on a computer ; and 3 .
programmatically defined choices of when to checkpoint and when to avoid checkpointing .
figure [ fig : pid - virt ] succintly describes the philosophy of process virtualization .
some invariant ( in this case the pid ( process i d ) of a process may have a different name prior to checkpoint and after restart .
a virtualized process will interact only with virtual process ids in the base code .
a dmtcp plugin retains a translation table between the virtualized pid known to the base code and the real pid known to the kernel .
since the base code and the kernel interact primarily through system calls , the dmtcp plugin defines a wrapper function around that system call .
the wrapper function translates between virtual and real pids both for arguments to the system call and for the return value .
this is illustrated both in figure [ fig : pid - virt ] and in the example code of listing [ lst : pidwrapperexample ] .
.... wrapper int kill(pid_t pid , int sig ) { disable_ckpt ( ) ; real_pid = virt_to_real(pid ) ; int ret = real_@@kill(real_pid , sig ) ; enable_ckpt ( ) ; return ret ; } .... additionally , pid s may be passed as part of the proc filesystem , and through other limited means . to solve this , dmtcp implements virtualization of filenames as well as pid names , and
so the `` open '' system call will also be interposed upon to detect names such as ` /proc
/ pid / maps ` . in this way
, a collection of wrapper functions can be collected together within a dmtcp plugin library .
such a library implements a virtualization layer .
the elf library standard implements a library search order such that symbols are searched in order as follows : executable @xmath1 lib1 @xmath1 lib2 ... libc @xmath1 kernel where the symbol is finally replaced by a direct kernel call .
this sequence can also be viewed as a sequence of layers , consistent with the common operating system implementation through layers .
a dmtcp plugin for pids then presents a virtualization layer in which all higher layers see only virtual pids , and all lower layers see only real pids .
this is analogous to an operating system design in which a higher layer sees the disk as a filesystem , and a lower layer sees the disk as a collection of disk blocks . in a similar way
, dmtcp provides layers to virtualize filenames , environment variables and myriad other names . in this way
, an end user can implement a fault injection plugin layer such that all code below that layer sees injected faults , while higher layers do not see the injected faults .
additionally , such a layer can be instrumented to gather information such as the cumulative number of faults .
dmtcp also provides an api for the application or a plugin to either request a checkpoint or to avoid a checkpoint . upon checkpoint ,
each plugin is notified of a checkpoint barrier , and similarly upon restart .
thus , it is feasible to create successive checkpoints available for restart or available as a snaphot for later forensics on the cause of a later error .
optimizations such as forked checkpointing ( fork a child and continue in the parent ) are available in order to take advantage of the kernel s copy - on - write in order to make checkpointing / snapshotting extremely fast .
checkpointing in a distributed application context requires coordination between multiple processes at different virtualization layers .
the use of programmable barriers enables this coordination .
in addition to the checkpoint and restart events , each plugin ( or virtualization layer ) can define its own set barriers and a callback to execute at a barrier .
a centralized dmtcp coordinator forces the application processes to execute the barriers in sequence .
further , a hardware resource , for example , the interface to a hardware emulator , might be shared among multiple processes that share parent - child relationships . to get a semantically equivalent state on restart
, the barriers can be used to elect a leader to save and restore the connection to the hardware emulator on restart .
this section describes three specific real - world use cases where dmtcp was extended to support hardware emulation and simulation software .
the examples are motivated by our work with various hardware and eda tool vendors .
gui - based simulation software presents a unique challenge in checkpointing .
the front - end software communicates with an x server via a socket .
the x server runs in a privileged mode and outside of checkpoint control .
while the connection could be blacklisted for the checkpointing , application s gui context and state is part of the x server and can not be checkpointed .
the context does not exist at restart time and needs to be restored .
dmtcp was extended to transparently support checkpointing of vnc @xcite and xpra @xcite .
the two tools allow x command forwarding to a local x server that can be run under checkpoint control .
@xcite presents an alternate record - prune - replay based approach using dmtcp to checkpoint gui - based applications .
authentication and license services is an important issue for protecting the intellectual property of all the parties . often , the authentication protocols and software are proprietary and specific to a vendor .
further , the licensing services are not run under checkpoint control , which makes it difficult to get a `` complete '' checkpoint of the software .
extensions were added to dmtcp to allow a vendor to hook into the checkpoint and restart events and mark certain connections as `` external '' to the computation . at checkpoint time , the connections marked external are ignored by dmtcp and instead the responsibility of restoring these connections is delegated to the vendor - specific extension .
the vendor - specific plugin also allows the application to check back with the licensing service at restart time in order to not violate a licensing agreement that restricts the number of simultaneous `` seats '' .
the ability to migrate a process among the available resources is critical for efficient utilization of hardware emulator resources .
however , the environment variables , the file paths , and the files that are saved as part of a checkpoint image make such migrations challenging
. we added dmtcp extensions ( plugins ) to virtualize the environment and the file paths .
this allows a process to be restarted on a different system by changing the values and the paths .
another extension that we added to dmtcp allows a user to explicitly control the checkpointing of files used by their application at the granularity of a single file .
hardware emulators communicate with the host software via high - speed interfaces .
any in - flight transactions at checkpoint time can result in the data being lost and inconsistent state on restart .
thus , it is important to bring the system to a quiescent state and drain the in - flight data on the buses before saving the state .
further , checkpointing while the software is in a critical state ( like holding a lock on a bus ) can lead to complications on restart . to help mitigate such issues ,
dmtcp was extended to allow fine - grained programmatic control over checkpointing .
this enables the hardware / eda tool vendor to tailor the checkpointing for their specific requirements .
in particular , it allows a user to invoke checkpointing from within their code , disable checkpointing for critical sections , or delay the resuming of user threads until the system reaches a well - behaved state .
the software toolchain used for simulation and emulation is often put together by integrating various third - party components .
the components may be closed - source and may use proprietary protocols for interfacing with each other and the system .
for example , many software toolchains rely on legacy 32-bit code that s difficult to port to 64-bits , and so , support for mixed 32-/64- bit processes was an important consideration .
checkpointing while holding locks was another interesting issue .
while the locks and their states are a part of the user - space memory ( and hence , a part of the checkpoint image ) , an application can also choose to use an error - checking lock that disallows unlocking by a different thread than the one that acquired it .
on restart , when new thread ids would be assigned by the system , the locks would become invalid and the unlock call would fail .
we extended dmtcp by adding wrapper functions for lock acquisition and release functions to keep track of the state of locks . at restart time , a lock s state is patched with the newer thread ids . more generally , the problem described above is about the state that s preserved when a resource is allocated at checkpoint time and needs to be deallocated at restart time .
while the restarted process inherits its state from the checkpoint image , its environment ( thread ids , in the above case ) might have changed on restart .
an application author with domain expertise can extend the dmtcp checkpointing framework to recognize and virtualize these resources .
the state could be a part of the locks that are acquired by a custom thread - safe malloc library , or the guard regions created by a library to guard against buffer overflows , or the libraries that are loaded temporarily .
high performance computing ( hpc ) is the traditional domain in which checkpoint - restart is heavily used .
it is used for the sake of fault tolerance during a long computation , for example of days . for a survey of checkpoint - restart implementations in the context of high performance computing ,
see egwutuoha @xcite . in the context of hpc , dmtcp and blcr @xcite
are the most widely used examples of transparent , system - level checkpoint - restart parallel computing .
( a transparent checkpointing package is one that does not modify the target application . )
dmtcp ( distributed multithreaded checkpointing ) is a purely user - space implementation .
in addition to being transparent , it also does not require any kernel modules and its installation and execution does not require root privilege or the use of special linux capabilities .
it achieves its robustness by trying to stay as close to the posix standard as possible in its api with the linux kernel . the first version of dmtcp was later described in @xcite .
that version did not provide the plugin model for process virtualization .
for example , virtualization of network addresses did not exist , as well as a series of other constructs , such as timers , session ids , system v shared memory , and other features .
these features were added later due to the requirements of high performance computing . eventually , the current procedure for virtualizing process ids ( see section [ sec : pidvirtexample ] was developed . to the best of our knowledge ,
dmtcp is unique in its approach toward process i d virtualization .
eventually , the plugin model was developed , initially for transparent support of the infiniband network fabric @xcite .
the current extension of that plugin model is described in @xcite . still later , the requirements for robust support of eda in collaboration with intel led to the development of reduction of runtime overhead graphic support using xpra , path virtualization ( for virtualization of the runtime slot and associated directory of a run using a hardware emulator , including different mount points on the restart computer ) , virtualization of environment variables including the x - windows display variable ( for similar reasons ) , robustness across a variety of older and newer linux kernels and gnu libc versions , mixed multi - architecture ( 32- and 64-bit ) processes within a single computation , low - overhead support for malloc - intensive programs , re - connection of a socket to a license server on restart , and whitelist and blacklist of special temporary files that many or may not be present on the restart computer .
blcr supports only single - node standalone checkpointing .
in particular , it does not support checkpointing of tcp sockets , infiniband connections , open files , or sysv shared memory objects .
blcr is often used in hpc clusters , where one has full control over the choice of linux kernel and other systems software .
typically , a linux kernel is chosen that is compatible with blcr , a blcr kernel module is installed , and when it is time to checkpoint , it is the responsibility of an mpi checkpoint - restart service to temporarily disconnected the mpi network layer , then checkpoint locally on each node , and finally re - connect the mpi network layer .
note that blcr is limited in what features it supports , notably including a lack of support for sockets and system v shared memory . quoting from the blcr user s guide : _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ `` however , certain applications are not supported because they use resources not restored by blcr : @xmath0 applications which use sockets ( regardless of address family ) .
@xmath0 ; applications which use character or block devices ( e.g. serial ports or raw partitions ) .
@xmath0 ; applications which use system v ipc mechanisms including shared memory , semaphores and message queues . ''
@xcite _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ the lack of blcr support for shared memory also prevents its use in openshmem @xcite . zapc and cruz represent two other checkpointing approaches that are not currently widely used .
zapc @xcite and cruz @xcite were earlier efforts to support distributed checkpointing , by modifying the kernel to inserting hooks into the network stack using netfilter to translate source and destination addresses .
zapc and cruz are no longer in active use .
they were designed to virtualize primarily two resources : process ids and ip network addresses .
they did not support ssh , infiniband , system v ipc , or posix timers , all of which are commonly used in modern software implementation .
criu @xcite leverages linux namespaces for transparently checkpointing on a single host ( often within a linux container ) , but lacks support for distributed computations . instead of directly virtualizing the process
i d , criu relies on extending the kernel api through a much larger proc filesystem and a greatly extended `` prctl '' system call .
for example , the `` pr_set_mm '' has 13 additional parameters that can be set ( e.g. , beginning end end of text , data , and stack ) . in another example
, criu relies on the `` config_checkpoint_restore '' kernel configuration to allow a process to directly modify the kernel s choice of pid for the next process to be created @xcite . in a general context
, there is a danger that the desired pid to be restored may already be occupied by another process , but criu is also often used within a container where this restriction can be avoided . finally , criu has a more specialized plugin facility @xcite .
some examples are : ability to save and restore the contents of particular files ; and the means to save and restore pointers to external sockets , external links , and mount points that are outside the filesystem namespace of an lxc ( linux container ) .
recall that criu does not try to support distributed computations .
perhaps it is for this reason that criu did not have the same pressure to develop a broader plugin system capable of supporting generic external devices such as hardware emulators .
the term _ process virtualization _ was used in @xcite . that work discusses kernel - level support for such process virtualization , while the current work emphasizes an entirely user - space approach within unprivileged processes .
related to process virtualization is the concept of a library os , exemplified by the drawbridge library os @xcite and exokernel @xcite .
however , such systems are concerned with providing _ extended or modified _ system services that are not natively present in the underlying operating system kernel .
both process - level virtualization and the library os approach employ a user - space approach ( ideally with no modification to the application executable , and no additional privileges required ) .
however , a library os is concerned with providing _ extended or modified _ system services that are not natively present in the underlying operating system kernel .
process virtualization is concerned with providing a semantically equivalent system object using the _ same _ system service .
this need arises when restarting from a checkpoint image , or when carrying out a live process migration from one computer to another .
the target computer host is assumed to provide the same system services as were available on the original host .
although process - level virtualization and a library os both operate in user space without special privileges , the goal of a library os is quite different .
a library os modifies or extends the system services provided by the operating system kernel .
for example , drawbridge @xcite presents a windows 7 personality , so as to run windows 7 applications under newer versions of windows .
similarly , the original exokernel operating system @xcite provided additional operating system services beyond those of a small underlying operating system kernel , and this was argued to often be more efficient that a larger kernel directly providing those services . -12pt
in order to develop a successful plugin model for checkpointing in the context of eda , one required modularity that enabled the dmtcp team , intel , and mentor graphics to each write their own modular code .
further , the intel and mentor graphics dmtcp - based plugins and other code were of necessity proprietary .
this work has shown how the dmtcp plugin model can be used to provide a flexible model enabling full cooperation , while avoiding the more extreme roadmaps of either fully application - specific code or transparent , system - level checkpointing with no knowledge of the proprietary aspects of the mentor graphics hardware emulator .
j. ansel , k. arya , and g. cooperman , `` dmtcp : transparent checkpointing for cluster computations and the desktop , '' in _ ieee int .
symp . on parallel and
distributed processing ( ipdps)_.1em plus 0.5em minus 0.4em ieee press , 2009 , pp .
112 .
k. arya , r. garg , a. y. polyakov , and g. cooperman , `` design and implementation for checkpointing of distributed resources using process - level virtualization , '' in _ ieee int .
conf . on cluster computing ( cluster16)_.1em plus 0.5em minus 0.4emieee press , 2016 , pp .
402412 .
j. cao , k. arya , r. garg , s. matott , d. k. panda , h. subramoni , j. vienne , and g. cooperman , `` system - level scalable checkpoint - restart for petascale computing , '' in _
22nd ieee int . conf . on parallel and distributed systems ( icpads16)_.1em plus 0.5em minus 0.4emieee press , 2016 , also , technical report available as : arxiv preprint arxiv:1607.07995 .
g. cooperman , j. evans , a. garg , r. garg , n. a. rosenberg , and k. suresh , `` transparently checkpointing software test benches to improve productivity of soc verification in an emulation environment , '' 2017 , ( submitted ) .
i. ljubuncic , r. giri , a. rozenfeld , and a. goldis , `` be kind , rewind checkpoint & restore capability for improving reliability of large - scale semiconductor design , '' in _ 2014 ieee high performance extreme computing conference ( hpec-2014 ) _ , 2014 .
[ online ] .
available : http://www.ieee-hpec.org/2014/cd/index_htm_files/finalpapers/34.pdf a. shina , k. ootsu , t. ohkawa , t. yokota , and t. baba , `` proposal of incremental software simulation for reduction of evaluation time , '' in _ 2012 third international conference on networking and computing ( icnc ) _ , dec 2012 , pp . 311315 .
s. resmerita and w. pree , `` verification of embedded control systems by simulation and program execution control , '' in _ 2012 american control conference ( acc)_.1em plus 0.5em minus 0.4emieee , 2012 , pp .
35813586 .
w. leungwattanakit , c. artho , m. hagiya , y. tanabe , m. yamamoto , and k. takahashi , `` modular software model checking for distributed systems , '' _ ieee trans . on software engineering _ ,
40 , no . 5 ,
483501 , may 2014 .
j. cao , g. kerr , k. arya , and g. cooperman , `` transparent checkpoint - restart over infiniband , '' in _ proc . of the 23rd int
. symp . on high - performance parallel and
distributed computing_.1em plus 0.5em minus 0.4emacm press , 2014 , pp .
1324 .
g. janakiraman , j. santos , d. subhraveti , and y. turner , `` cruz : application - transparent distributed checkpoint - restart on standard operating systems , '' in _ international conference on dependable systems and networks , 2005 .
dsn 2005 . proceedings _ , jun .
2005 , pp . 260269 .
g. vallee , r. lottiaux , d. margery , and c. morin , `` ghost process : a sound basis to implement process duplication , migration and checkpoint / restart in linux clusters , '' in _ proc .
of the the 4th int .
symp . on parallel and
distributed computing _ , ser .
ispdc 05 , 2005 , pp .
97104 .
d. e. porter , s. boyd - wickizer , j. howell , r. olinsky , and g. c. hunt , `` rethinking the library os from the top down , '' in _ proc . of the sixteenth international conference on architectural support for programming languages and operating systems
asplos xvi.1em plus 0.5em minus 0.4emnew york , ny , usa : acm , 2011 , pp . 291304 .
d. r. engler , m. f. kaashoek , and j. otoole , jr .
, `` exokernel : an operating system architecture for application - level resource management , '' in _ proc . of 15th acm symp .
on operating systems principles _ , ser .
sosp 95.1em plus 0.5em minus 0.4emacm , 1995 , pp . 251266 . | checkpoint - restart is now a mature technology .
it allows a user to save and later restore the state of a running process .
the new plugin model for the upcoming version 3.0 of dmtcp ( distributed multithreaded checkpointing ) is described here .
this plugin model allows a target application to disconnect from the hardware emulator at checkpoint time and then re - connect to a possibly different hardware emulator at the time of restart .
the dmtcp plugin model is important in allowing three distinct parties to seamlessly inter - operate .
the three parties are : the eda designer , who is concerned with formal verification of a circuit design ; the dmtcp developers , who are concerned with providing transparent checkpointing during the circuit emulation ; and the hardware emulator vendor , who provides a plugin library that responds to checkpoint , restart , and other events .
the new plugin model is an example of process - level virtualization : virtualization of external abstractions from within a process .
this capability is motivated by scenarios for testing circuit models with the help of a hardware emulator .
the plugin model enables a three - way collaboration : allowing a circuit designer and emulator vendor to each contribute separate proprietary plugins while sharing an open source software framework from the dmtcp developers .
this provides a more flexible platform , where different fault injection models based on plugins can be designed within the dmtcp checkpointing framework .
after initialization , one restarts from a checkpointed state under the control of the desired plugin .
this restart saves the time spent in simulating the initialization phase , while enabling fault injection exactly at the region of interest . upon restart
, one can inject faults or otherwise modify the remainder of the simulation .
the work concludes with a brief survey of the existing approaches to checkpointing and to process - level virtualization . |
the bak sneppen ( bs ) model of biological evolution has been introduced in @xcite to study self - organised critical behaviour in populations with natural selection and spatial interactions . in the model
, a population is spread out over a circle so that each species has exactly two neighbours , and each site is assigned a numerical parameter between zero and one that describes the _ fitness _ of the species at that particular site .
the system evolves according to the following discretetime dynamics : at each time step the species with the smallest fitness parameters becomes extinct and is replaced by a new species whose fitness parameter becomes independent uniformly distributed on the interval @xmath0 $ ] . to take into account interactions between species , e.g. competition for resources or predatorprey relations , the fitness parameters of the two neighbours of the least fit species
are also reset to random values .
this represents the idea that the fitness is not just a fixed property of a species but also depends on its environment . even though this is a very rudimentary model for evolution it does possess some of the characteristic features thought to be associated with evolving biological systems , including longrange dependence and evolutionary activity on all time scales ,
see e.g. @xcite for a discussion in the context of selforganised criticality . despite its apparently easy definition ,
the bs model has withstood most attempts at mathematical analysis in the past .
partial results have been obtained , however , in the context of rankdriven processes and meanfield approximations @xcite .
early on it was conjectured based on simulations that the fitness distribution at a fixed site converges , in the limit of large populations , to a uniform distribution on the interval @xmath1 $ ] , where @xmath2 .
there are only few rigorous results about the original bs model ; in @xcite it is proved that the expected fitness at a fixed site in the stationary regime is bounded away from unity uniformly in the size of the population ; @xcite gives a conditional characterization of the limiting distribution in terms of a set of critical thresholds ( see also * ? ? ?
* ; * ? ? ?
in this paper we suggest a novel approach to determining the stationary fitness distribution in the bs model with a finite number of species .
the method consists of translating the evolutionary dynamics into a set of linear recurrence equations for the coefficients of the densities of the finitetime distributions and analysing the asymptotic properties of these recursive equations .
we apply the method to derive the asymptotic fitness distribution in the bs model with four species , which is given by ( [ theorem - asympdist4 ] ) @xmath3{\boldsymbol{1}}_{[0,1]^4}({\boldsymbol{x}}){\mathrm{d}}^4{\boldsymbol{x}}.\ ] ] it has been suggested that a similar , and potentially easier , analysis can be applied to the anisotropic bak
sneppen ( abs ) model ; in this variation of the model the evolutionary dynamics are modified so that the fitness parameter of only one neighbour , say the right one , is updated at each time step together with the fitness parameter of the least fit species . while many qualitative properties are believed to be shared by the original and the anisotropic bs , analytical considerations and numerical calculations indicate that they fall in different universality classes , and have thus different critical exponents @xcite .
our method can be applied to the abs . in the first nontrivial case of three species the asymptotic fitness distribution follows from computations which are very similar to those leading to [ theorem - asympdist4 ] , and
is given by @xmath4{\boldsymbol{1}}_{[0,1]^3}({\boldsymbol{x}}){\mathrm{d}}^3{\boldsymbol{x}}.\ ] ] the complexity of the calculations for larger populations seems to be increasing at a similar rate as in bs and studying abs does therefore not , for the purpose of understanding the method we propose in the current paper , offer major advantages over the original model .
for this reason we concentrate in the following on the isotropic case . [ [ outline - of - the - paper ] ] outline of the paper + + + + + + + + + + + + + + + + + + + + the paper is organised as follows . in [ section - qualana ]
we present one of several possible , precise definitions of the bs model in terms of a markov chain , and use this representation to derive qualitative properties for arbitrary population sizes .
we then turn to the special case of the fourspecies model in [ section - n4 ] , where we derive the full asymptotic fitness distribution .
finally , [ section - genrec ] is devoted to making further progress towards a similarly explicit solution for larger populations .
in particular , we obtain explicit linear recursions for the coefficients that describe the finitetime distributions in the fivespecies model .
[ [ notation ] ] notation + + + + + + + + throughout the paper we denote by @xmath5 the set of positive integers and write @xmath6 for the number of species in the bs model .
arithmetic operations in indices pertaining to the number of species are always understood to be performed modulo @xmath7 , so as to give a result between @xmath8 and @xmath7 .
the symmetric group on @xmath7 letters is denoted by @xmath9 . for a vector @xmath10^n$ ] and an integer @xmath11 we denote by @xmath12}$ ] and @xmath13\nu[}$ ] the inner and outer @xmath14centred projections , that is @xmath15}:[0,1]^n\to[0,1]^3,\quad & { \boldsymbol{x}}_{[\nu ] } = ( x_{\nu-1},x_\nu , x_{\nu+1});\\ ( \cdot)_{]\nu[}:[0,1]^n\to[0,1]^{n-3},\quad & { \boldsymbol{x}}_{]\nu [ } = ( x_{\nu+2},\ldots , x_n , x_1,\ldots , x_{\nu-2}).\end{aligned}\ ] ] the @xmath16th components of the vectors @xmath17}$ ] and @xmath18\nu[}$ ] are denoted by @xmath17,i}$ ] and @xmath18\nu[,i}$ ] , respectively . given another vector @xmath19 ^ 3 $ ] we extend the notation of outer projections by setting @xmath20\nu[_{{\boldsymbol{\xi}}}}:[0,1]^n\to[0,1]^n,\quad { \boldsymbol{x}}_{]\nu[_{{\boldsymbol{\xi } } } } = ( \xi_1,\xi_2,\xi_3,x_{\nu+2},\ldots , x_n , x_1,\ldots , x_{\nu-2}).\ ] ] we use the multiindex notation @xmath21 , @xmath10^n$ ] , @xmath22 , as a shorthand for the monomial @xmath23 .
we further let @xmath24 denote probability and write @xmath25 for the borel sets of a topological space @xmath26 .
finally , we denote by @xmath27 the indicator of an expression @xmath28 , which is defined to be one if @xmath28 is true , and zero else ; for a set @xmath29 , we also write @xmath30 instead of @xmath31 .
we interpret the bak sneppen model with @xmath7 species as a markov chain @xmath32 on the hypercube @xmath0^n$ ] , see , e.g. @xcite for a general introduction to this topic . to avoid the trivial case in which all fitness parameters are updated at each time step , we assume that @xmath33 . for each time @xmath34 ,
the values of @xmath35 denote the fitnesses of the different species in the population . in order to formalise the heuristic description of the evolutionary processes described in the introduction we first observe that the fitness landscape @xmath36 at time @xmath37 can be determined easily if one knows the fitness values at time @xmath38 and , in particular , which species was the least fit at that time .
in fact , if at time @xmath38 the @xmath14th species is the least fit , then the fitness values of the species at the @xmath39th , @xmath14th , and @xmath40th site are independent uniformly distributed at time @xmath37 , while all other fitness values remain unchanged . for this argument to be valid ,
it is necessary that @xmath41 has a unique minimal value .
in fact the bs model , as we have described it , is not welldefined without a rule for breaking ties .
one natural possibility is to randomly select one of the sites with minimal fitness value .
we will , however , not have to concern ourselves with this complication because we will assume that the initial fitness distribution at time @xmath42 is absolutely continuous , in which case ties occur with probability zero .
further formalising the argument , we can partition the event @xmath43 , @xmath44\in\mathscr{b}([0,1]^n)$ ] , as @xmath45 and obtain , @xmath46\nu[}\in a_{]\nu [ } \text { and } \min{\boldsymbol{f}}_{n , k}=f_{n , k,\nu}\right){\mathbb{p}}\left((u_1,u_2,u_3)\in a_{[\nu]}\right)},\end{aligned}\ ] ] where @xmath13\nu[}$ ] and @xmath12}$ ] denote the inner and outer projections , respectively .
the random variables @xmath47 are independent and uniformly distributed on @xmath0 $ ] , and thus the last probability equals @xmath48}\right)$ ] .
put differently , the transition kernels @xmath49 of the markov chain @xmath50 , which are characterized by the equations @xmath51^n,\quad a\in\mathscr{b}([0,1]^n),\ ] ] are given by @xmath52 here , @xmath53^n\times[0,1]^n$ ] , and @xmath54 denotes the dirac measure located at @xmath55 .
this quite explicit expression is enough to prove that the markov chains @xmath50 are ergodic in a strong sense , which is crucial for our approach to determining their stationary distributions . for the definition of uniform convergence
we refer the reader to ( * ? ? ?
* definition ( 16.6 ) ) .
[ prop - ergodicity ] for every positive integer @xmath7 , the markov chain @xmath50 is uniformly ergodic .
it is straightforward , albeit tedious , to show that the multistep transition kernels @xmath56 , which are defined recursively by @xmath57^n}{\mathbb{p}}_{n,{\boldsymbol{x}}}^1\left({\mathrm{d}}^n{\boldsymbol{y}}\right){\mathbb{p}}_{n,{\boldsymbol{y}}}^{m-1}\left({\mathrm{d}}^n{\boldsymbol{\xi}}\right),\quad m{\geqslant}2,\ ] ] dominate , for each @xmath58 , a constant times lebesgue measure restricted to @xmath0^n$ ] .
the claim is then a consequence of @xcite[theorem b , ii ) ] . since a similar result has been obtained in @xcite , we omit the technical details
. implies in particular that there exists a unique invariant distribution @xmath59 satisfying @xmath60^n}{{\mathbb{p}}_{n,{\boldsymbol{x}}}^1({\mathrm{d}}^n{\boldsymbol{\xi}})\mu_n({\mathrm{d}}^n{\boldsymbol{x}})},\quad \text{for all } a\in\mathscr{b}\left([0,1]^n\right);\ ] ] moreover , the finitetime distributions , irrespective of the initial distribution at time @xmath42 , converge to @xmath59 in the total variation norm .
one way to obtain results about the stationary distribution @xmath59 is thus to fix a convenient initial distribution for @xmath61 , to determine the finitetime distributions , and to analyse their asymptotic behaviour .
the most convenient choice of initial distribution turns out to be the uniform distribution @xmath62}^{\otimes n}$ ] .
we now proceed to give a qualitative description of the finitetime distributions .
[ prop - finitetimedistqual ] for every @xmath33 , the finitetime distributions of the markov chain @xmath50 are absolutely continuous with piecewise polynomial lebesgue densities .
more precisely , for every @xmath34 and @xmath63 , there exist finite index sets @xmath64 and polynomials @xmath65 $ ] of the form @xmath66^{n-3},\ ] ] such that @xmath67}^{\otimes n}\right.\right)\notag\\ = & \left[\sum_{\nu=1}^n\sum_{\sigma\in{\mathcal{s}}_{n-3}}{q_{n , k,\sigma}\left({\boldsymbol{x}}_{]\nu[}\right){\boldsymbol{1}}_{\{0{\leqslant}x_{]\nu[,\sigma(1)}{\leqslant}\ldots{\leqslant}x_{]\nu[,\sigma(n-3)}{\leqslant}1\}}}\right]{\mathrm{d}}^n{\boldsymbol{x}}.\end{aligned}\ ] ] the finitetime distributions are absolutely continuous because the initial distribution at time @xmath42 is , and the transition kernel [ eq - transkern ] preserves absolute continuity .
a reformulation of [ eq - recfnkaabstract ] shows that the densities @xmath68 satisfy the recursion @xmath69 ^ 3 } { \boldsymbol{1}}_{\left\{\xi_2<\min(\xi_1,\xi_3,{\boldsymbol{x}}_{]\nu[})\right\}}g_{n , k}\left({\boldsymbol{x}}_{]\nu[_{{\boldsymbol{\xi}}}}\right){\mathrm{d}}^3{\boldsymbol{\xi}}.\ ] ] the claim thus follows by induction on @xmath38 , using the fact that integrals of polynomial expressions over bounded convex polytopes are again polynomial expressions .
more precisely , we assume that , for some @xmath70 , the function @xmath71 is of the form asserted in [ eq - defgnk ] .
it then follows that @xmath72 ^ 3}{\boldsymbol{1}}_{\left\{\xi_2<\min(\xi_1,\xi_3,{\boldsymbol{x}}_{]\nu[})\right\}}{\boldsymbol{1}}_{\left\{\left({\boldsymbol{x}}_{]\nu[_{{\boldsymbol{\xi}}}}\right)_{]\mu[}\sim\sigma\right\}}\left[\left({\boldsymbol{x}}_{]\nu[_{{\boldsymbol{\xi}}}}\right)_{]\mu[}\right]^{{\boldsymbol{i}}}{\mathrm{d}}^3{\boldsymbol{\xi}},\end{aligned}\ ] ] where , for @xmath73^{n-3}$ ] and @xmath63 , the expression @xmath74 is a shorthand for @xmath75 .
since , for each @xmath76 , the integrand is equal to @xmath77\nu[}\right)$ ] for some polynomial function @xmath78 $ ] , the claim follows . for the special cases
@xmath79 we will see later how [ eq - recgnk ] can be translated into explicit recursions for the coefficients @xmath80 . before we do that , however , we compute , for general @xmath7 , the density @xmath81 of the fitness distribution after the first evolution step . for each @xmath33 , the density @xmath81 is given by @xmath82^n}({\boldsymbol{x}})\sum_{\nu=1}^n{p(\min{\boldsymbol{x}}_{]\nu[})},\quad p(x ) = x\left(\frac13x^2-x+1\right).\ ] ]
the claim follows from [ eq - recgnk ] and the observation that an initial distribution of @xmath62}^{\otimes n}$ ] corresponds to @xmath83 .
alternatively , one can use the fact that @xmath84.\qedhere\ ] ]
in this section we expand on the results from the last section to compute the stationary distribution in the bak sneppen model with four species , @xmath85 .
this is much easier than the general situation and we are able to obtain a complete picture of the joint asymptotic behaviour of the four fitnessparameters .
partial generalisations to larger populations are discussed in [ section - genrec ] . for notational convenience
we suppress the index @xmath85 in this section .
[ prop - finitetimedistqual ] then implies that there are rational numbers @xmath86 and positive integers @xmath87 such that the finitetime distributions of @xmath88 are given by @xmath89}^{\otimes 4}\right.\right ) = \left[\sum_{\nu=1}^4{q_k(x_\nu)}\right]1_{[0,1]^4}({\boldsymbol{x}}){\mathrm{d}}^4{\boldsymbol{x}},\quad q_k(x ) = \sum_{i=0}^{d_k}{\alpha_{k , i}x^i},\ ] ] so that @xmath90 .
the chapman kolmogorovtype equation [ eq - recgnk ] implies that , for @xmath70 , @xmath91 ^ 3 } { \boldsymbol{1}}_{\{\xi_2<\min(\xi_1,\xi_3,x_\nu)\}}\left[x_\nu^i+\xi_2^i+\xi_1^i+\xi_3^i\right]{\mathrm{d}}^3{\boldsymbol{\xi}}.\ ] ] equating each of the four @xmath14summands separately and evaluating the integrals on the right side , one sees that , for @xmath92 , @xmath93 ^ 3 } { \boldsymbol{1}}_{\{\xi_2<\min(\xi_1,\xi_3,x_\nu)\}}{\mathrm{d}}^3{\boldsymbol{\xi}}+\int_{[0,1]^3 } { \boldsymbol{1}}_{\{\xi_2<\min(\xi_1,\xi_3,x_\nu)\}}\xi_2^i{\mathrm{d}}^3{\boldsymbol{\xi}}+2\int_{[0,1]^3 } { \boldsymbol{1}}_{\{\xi_2<\min(\xi_1,\xi_3,x_\nu)\}}\xi_1^i{\mathrm{d}}^3{\boldsymbol{\xi}}\right]\notag\\ = & \sum_i\alpha_{k , i}x_\nu^{i+1}\left [ \left(\frac13 x_\nu^2-x_\nu+1\right ) + \left(\frac{x_\nu^2}{i+3}-\frac{2x_\nu}{i+2}+\frac{1}{i+1}\right ) + \frac{2x_\nu}{i+1}\left(\frac{x_\nu}{i+3}-\frac{1}{i+2}\right)\right]+x_\nu(2-x_\nu)\sum_i\frac{\alpha_{k , i}}{i+1}.\end{aligned}\ ] ] comparing coefficients of equal powers of @xmath94 on both sides of this equation , one arrives at the following recursion for the coefficients @xmath86 .
[ prop - rec - alpha4 ] for each @xmath95 , the constant term @xmath96 vanishes .
the remaining coefficients @xmath86 satisfy the recursion @xmath97 in particular , @xmath86 is equal to zero if @xmath98 .
only the last statement requires further justification . by the last line of the recursion [ eq - rec - alpha4 ] ,
the coefficient @xmath99 can be written , for any @xmath100 , as a linear combination of the coefficients @xmath101 , @xmath102 .
choosing @xmath103 and observing that @xmath104 is zero for @xmath105 , it follows that @xmath106 vanishes for @xmath107 .
the next step consists in finding as explicit a solution to this recursive system as possible .
fortunately , since we are mainly interested in the asymptotic behaviour of the bak
sneppen model , it suffices to concentrate on large values of @xmath38 .
[ lemma - beta ] for every @xmath108 , there exists a rational number @xmath109 such that @xmath110 for every @xmath111 .
in particular , @xmath112 .
the claim is true for @xmath113 ; this follows from the fact that @xmath114 is a probability density function and thus integrates to one ; by symmetry , @xmath115 then integrates to @xmath116 . observing
that @xmath117 equals @xmath118 , it follows from [ eq - rec - alpha4 ] that , for all @xmath119 , @xmath120 similarly , @xmath121 , @xmath122 , and @xmath123 , @xmath124 , and so the claim is true for @xmath125 as well . for general @xmath126
, the claim follows by induction .
[ prop - limitq4 ] let @xmath127 be the numbers from [ lemma - beta ] .
the functions @xmath115 converge uniformly on @xmath0 $ ] , i.e. in the supremum norm , to @xmath128 , which is given by @xmath129 it follows from [ prop - ergodicity ] that the functions @xmath115 converge uniformly to some limit , and because @xmath86 converges , for each @xmath16 , to @xmath109 by [ lemma - beta ] , this limit must be @xmath130 . using [ prop - rec - alpha4 ]
one sees that the sequence @xmath131 satisfies the homogeneous four - term recursion @xmath132 by the theory of homogeneous linear recurrence equations , @xmath133 is given by @xmath134 where @xmath135 are the roots of the characteristic polynomial @xmath136 , and @xmath137 are constants determined by the initial conditions . in the present case , as the proof of [ lemma - beta ] shows , these initial values are given by @xmath138 they can also easily be read off [ table - alpha4 ] , where we have recorded the functions @xmath115 for @xmath139 .
the constants @xmath140 are thus the solution of the vandermonde system @xmath141 that is @xmath142 it is then easy to compute @xmath143 as @xmath144 the form of the function @xmath143 with all horner coefficients being equal to @xmath145 is very interesting .
we do not have an intuitive explanation for this peculiarity .
.values of the coefficients @xmath86 determining the finitetime distributions of the fourspecies bak sneppen model .
values of @xmath109 , @xmath146 , emphasised .
empty cells indicate zeros . [ cols="^,^,^,^,^,^,^,^,^,^,^,^,^,^,^,^ " , ] we have thus proved the following result .
[ theorem - asympdist4 ] let @xmath143 be as in [ prop - limitq4 ] .
the stationary distribution @xmath147 of the bak sneppen model with four species is given by @xmath148{\boldsymbol{1}}_{[0,1]^4}({\boldsymbol{x}}){\mathrm{d}}^4{\boldsymbol{x}}.\ ] ] in particular , the onedimensional marginal of @xmath149 , which is the asymptotic fitness distribution at a fixed site , is absolutely continuous with density @xmath150}(x).\ ] ] in [ fig - margdens ] we have plotted the densities of the onedimensional marginal distributions of @xmath151 for @xmath152 , together with their limit as given by [ eq - asympmargdens ] .
the convergence asserted by [ prop - limitq4 ] is clearly visible . for @xmath152 ( dashed ) , together with their limit ( solid ) as given by [ eq - asympmargdens ] ]
for general values of @xmath7 , a similarly complete analysis of the asymptotic fitness distribution in the bs model as we presented for @xmath85 in the previous section seems difficult to obtain .
here we collect some formulae that serve as the starting point if one wants to carry through the same approach as in [ section - n4 ] . similarly to [ eq - comparecoeff4 ] , it holds for every @xmath153 that @xmath154\nu[}^{{\boldsymbol{i}}}{\boldsymbol{1}}_{\{0{\leqslant}{\boldsymbol{x}}_{]\nu[,\sigma(1)}{\leqslant}\ldots{\leqslant}{\boldsymbol{x}}_{]\nu[,\sigma(n-3)}\}}\notag\\ = & \sum_{\mu,{\boldsymbol{i}},\sigma}\alpha_{n , k,\sigma,{\boldsymbol{i}}}\int_{[0,1]^3 } { \boldsymbol{1}}_{\left\{\xi_2<\min(\xi_1,\xi_3,{\boldsymbol{x}}_{]\nu[})\right\}}\left[\left({\boldsymbol{x}}_{]\nu[_{{\boldsymbol{\xi}}}}\right)_{]\mu[}\right]^{{\boldsymbol{i}}}{\boldsymbol{1}}_{\left\{0{\leqslant}\left({\boldsymbol{x}}_{]\nu[_{{\boldsymbol{\xi}}}}\right)_{]\mu[,\sigma(1)}{\leqslant}\ldots{\leqslant}\left({\boldsymbol{x}}_{]\nu[_{{\boldsymbol{\xi}}}}\right)_{]\mu[,\sigma(n-3)}{\leqslant}1\right\}}{\mathrm{d}}^3{\boldsymbol{\xi}}.\end{aligned}\ ] ] the integrals that appear on the right hand side of this equation fall in four categories , depending on how many of the three @xmath155 s are hit by the outer projection @xmath13\mu[}$ ] .
[ [ case-1-mu2 . ] ] case 1 : @xmath156 .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + in this case , all three @xmath155 s are hit by the outer projection and the integral on the right side of [ eq - comparecoeffn ] simplifies to @xmath157 ^ 3 } { \boldsymbol{1}}_{\left\{\xi_2<\min(\xi_1,\xi_3,{\boldsymbol{x}}_{]\nu[})\right\}}\left({\boldsymbol{x}}_{]\nu[}\right)^{{\boldsymbol{i}}}{\boldsymbol{1}}_{\left\{0{\leqslant}{\boldsymbol{x}}_{]\nu[,\sigma(1)}{\leqslant}\ldots{\leqslant}{\boldsymbol{x}}_{]\nu[,\sigma(n-3)}{\leqslant}1\right\}}{\mathrm{d}}^3{\boldsymbol{\xi}}\\ = & \left({\boldsymbol{x}}_{]\nu[}\right)^{{\boldsymbol{i}}}{\boldsymbol{x}}_{]\nu[,\sigma(1)}\left[\frac13{\boldsymbol{x}}_{]\nu[,\sigma(1)}^2 - { \boldsymbol{x}}_{]\nu[,\sigma(1)}^2 + 1\right]{\boldsymbol{1}}_{\left\{0{\leqslant}{\boldsymbol{x}}_{]\nu[,\sigma(1)}{\leqslant}\ldots{\leqslant}{\boldsymbol{x}}_{]\nu[,\sigma(n-3)}{\leqslant}1\right\}}. \end{aligned}\ ] ] [ [ case-2-muin13 . ] ] case 2 : @xmath158 .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + in this case , either @xmath159 ( if @xmath160 ) or @xmath161 ( if @xmath162 ) is not hit by the outer @xmath149projection . for definiteness , we assume that @xmath160 , the other subcase being analogous . the sequential application of the outer projections @xmath13\nu[_{{\boldsymbol{\xi}}}}$ ] and @xmath133[}$ ] to @xmath163 results in the vector @xmath164 , contributing a factor @xmath165 to the integral . with the definitions @xmath166 ,
@xmath167 the integral in [ eq - comparecoeffn ] thus becomes @xmath168 ^ 3 } { \boldsymbol{1}}_{\left\{\xi_2<\min(\xi_1,\xi_3,{\boldsymbol{x}}_{]\nu[})\right\}}{\boldsymbol{1}}_{\left\{{\boldsymbol{y}}_{\sigma\left(\sigma^{-1}(n-3)-1\right)}{\leqslant}\xi_1{\leqslant}{\boldsymbol{y}}_{\sigma\left(\sigma^{-1}(n-3)+1\right)}\right\}}\xi_1^{i_{n-3}}{\mathrm{d}}^3{\boldsymbol{\xi}}\right]{\boldsymbol{1}}^{(*)}_{\left\{0{\leqslant}{\boldsymbol{y}}_{\sigma(1)}{\leqslant}\ldots{\leqslant}{\boldsymbol{y}}_{\sigma(n-3)}{\leqslant}1\right\}},\end{aligned}\ ] ] where @xmath169 , @xmath170 , and the superscript @xmath171 indicates that only the ordering of the components of @xmath172 that pertain to @xmath163 , in contrast to @xmath173 , is restricted .
integration of this expression is standard and left to the reader .
[ [ case-3-muinn4 . ] ] case 3 : @xmath174 .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + in this case , both @xmath175 and either @xmath161 ( if @xmath176 ) or @xmath159 ( if @xmath177 ) are hit by the outer @xmath149projection . again , for definiteness , we assume that @xmath176 , and observe that the other subcase can be treated in an analogous way . in this case @xmath178\nu[_{{\boldsymbol{\xi}}}}\right)_{]4 [ } = { \boldsymbol{y}}\coloneqq\left(x_{\nu+4},\ldots , x_n , x_1,\ldots , x_{\nu-2},\xi_1,\xi_2\right),\ ] ] resulting in a factor @xmath179 in the integrand ; the integral in [ eq - comparecoeffn ] is clearly zero unless @xmath180 , so as not contradict the first indicator function that specifies @xmath175 to be less than @xmath159 and @xmath18\nu[}$ ] . with the obvious definitions of @xmath181 and @xmath182 ,
the integral then becomes @xmath183 ^ 3 } { \boldsymbol{1}}_{\left\{\xi_2<\min(\xi_1,\xi_3,{\boldsymbol{x}}_{]\nu[})\right\}}{\boldsymbol{1}}_{\left\{{\boldsymbol{y}}_{\sigma\left(\sigma^{-1}(n-4)-1\right)}{\leqslant}\xi_1{\leqslant}{\boldsymbol{y}}_{\sigma\left(\sigma^{-1}(n-4)+1\right)}\right\}}\xi_1^{i_{n-4}}\xi_2^{i_{n-3}}{\mathrm{d}}^3{\boldsymbol{\xi}}\right]{\boldsymbol{1}}^{(*)}_{\left\{0{\leqslant}{\boldsymbol{y}}_{\sigma(2)}{\leqslant}\ldots{\leqslant}{\boldsymbol{y}}_{\sigma(n-3)}{\leqslant}1\right\ } } , \ ] ] the evaluation of which we leave to the reader .
[ [ case-4-all - other - mu . ] ] case 4 : all other @xmath149 .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + in this last case , all three @xmath155 s remain unaffected by the @xmath149projection and contribute to making the integral a little more complicated than before .
similar to the previous cases one obtains @xmath184\nu[_{{\boldsymbol{\xi}}}}\right)_{]\mu [ } = \left(x_{\mu+2},\ldots , x_n , x_1,\ldots , x_{\nu-2},\xi_1,\xi_2,\xi_3,x_{\nu+2},\ldots , x_{\mu-2}\right),\ ] ] and , again , the integral vanishes if @xmath185 .
consequently , the integral in [ eq - comparecoeffn ] becomes @xmath186 ^ 3 } { \boldsymbol{1}}_{\left\{\xi_2<\min(\xi_1,\xi_3,{\boldsymbol{x}}_{]\nu[})\right\}}{\boldsymbol{1}}_{\left\{{\boldsymbol{y}}_{\sigma\left(\sigma^{-1}(j_1)-1\right)}{\leqslant}\xi_1{\leqslant}{\boldsymbol{y}}_{\sigma\left(\sigma^{-1}(j_1)+1\right)}\right\}}{\boldsymbol{1}}_{\left\{{\boldsymbol{y}}_{\sigma\left(\sigma^{-1}(\nu+1)-1\right)}{\leqslant}\xi_3{\leqslant}{\boldsymbol{y}}_{\sigma\left(\sigma^{-1}(\nu+1)+1\right)}\right\}}\right.\notag\\ & \qquad\qquad\times\left.\xi_1^{i_{j_1}}\xi_2^{i_{j_2}}\xi_3^{i_{j_3}}{\mathrm{d}}^3{\boldsymbol{\xi}}\right]{\boldsymbol{1}}^{(*)}_{\left\{0{\leqslant}{\boldsymbol{y}}_{\sigma(2)}{\leqslant}\ldots{\leqslant}{\boldsymbol{y}}_{\sigma(n-3)}{\leqslant}1\right\}},\end{aligned}\ ] ] where @xmath187 , @xmath188 , @xmath189 , and @xmath190 .
according to whether or not @xmath159 and @xmath161 are separated by some @xmath191 in @xmath192 , there are a number of subcases to consider , which , however , do not present any difficulty .
closedform expressions of the integrals [ eq - case2 ] , [ eq - case3 ] , and [ eq - case4 ] have been obtained , but are omitted for lack of space and because they do not add much insight at the current stage . by equating equal powers of @xmath163 , [ eq - comparecoeffn ] ,
can thus , in principle , be used to derive an explicit recursion for the coefficients @xmath80 . here
, we only illustrate this potential by giving the recursion for @xmath193 , leaving the problem of working out the combinatorics of the general case for future research .
an easy symmetry argument shows that @xmath194 is equal to @xmath195 so that it is enough to consider the identity permutation @xmath196 .
[ prop - rec - alpha5 ] for @xmath193 , the coefficients @xmath197 have the following properties : a. @xmath198 for all @xmath119 and @xmath199
. b. for @xmath119 and @xmath200 , they satisfy the recursion @xmath201\alpha_{k , i-2,j}+\left[\frac{1}{3}+\frac{1}{i(i-2)}\right]\alpha_{k , i-3,j } , & i{\geqslant}3 .
\end{cases}\ ] ] c. for @xmath119 and @xmath202 , they satisfy @xmath203 as well as the recursion @xmath204\alpha_{k , i-2,0}+\left[\frac{1}{3}+\frac{1}{(i-2)i}\right]\alpha_{k , i-3,0}\\ \quad+\left(1+\frac{2}{i}\right)\sum_{p=0}^{3k+1}\frac{\alpha_{k , i-1,p}}{p+1}-\left(\frac{1}{2}+\frac{2}{i}\right)\sum_{p=0}^{3k+1}\frac{\alpha_{k , i-2,p}}{p+1}\\ \quad-\left(1+\frac{2}{i}\right)\sum_{p=0}^{i-2}\frac{\alpha_{k , i-2-p , p}}{p+1}+\left(\frac{1}{2}+\frac{2}{i}\right)\sum_{p=0}^{i-3}\frac{\alpha_{k , i-3-p , p}}{p+1}\\ \quad+\sum_{p=0}^{i-2}\frac{\alpha_{k , i-2-p , p}}{i - p } -\frac{1}{2 } \sum_{p=0}^{i-3}\frac{\alpha_{k , i-3-p , p}}{i - p } , & i{\geqslant}3
. \end{cases}\ ] ] unfortunately , it does not seem possible to find an explicit solution to this recursion .
in fact , the algorithm hyper for linear recurrence relations with polynomial coefficients @xcite indicates that the last line of [ eq - recn5jgeq1 ] does not have a hypergeometric solution .
our results can , however , be used to obtain exact expression for the finitetime fitness distributions in the fivespecies bs model without having to evaluate any integrals .
calculations based on [ eq - recn5jgeq1 ] and [ eq - recn5j0 ] give rise to the following conjecture .
the coefficients @xmath205 have the following properties : a. they form a stair case pattern in the sense that , for @xmath199 , @xmath206 in particular , @xmath207 for @xmath208 .
b. they become constant at time @xmath209 , that is there exist numbers @xmath210 such that @xmath211 for all @xmath212 .
in fact , we conjecture that , for each @xmath7 , there exist @xmath213 and @xmath214 such that @xmath215 for all @xmath216 .
i thank two anonymous referees for carefully reading the manuscript and making helpful suggestions to improve its presentation .
i also thank them for pointing out the applicability of our method to the anisotropic bak
sneppen model . | we suggest a new method to compute the asymptotic fitness distribution in the bak sneppen model of biological evolution . as applications
we derive the full asymptotic distribution in the fourspecies model , and give an explicit linear recurrence relation for a set of coefficients determining the asymptotic distribution in the fivespecies model . |
understanding all the consequences of the interplay between band structure effects and electron - electron interactions remains one of the present - day goals of theoretical solid - state physics .
one of the simplest model that contains the essence of this problem is the hubbard model . in the more than thirty years@xcite@xcite
since this model was formulated , much progress has been accomplished . in one dimension,@xcite@xcite various techniques such as diagrammatic resummations,@xcite bosonization,@xcite renormalization group@xcite@xcite and conformal approaches@xcite@xcite
have lead to a very detailed understanding of correlation functions , from weak to strong coupling . similarly , in infinite dimensions a dynamical mean - field theory@xcite leads to an essentially exact solution of the model , although many results must be obtained by numerically solving self - consistent integral equations .
detailed comparisons with experimental results on transition - metal oxides have shown that three - dimensional materials can be well described by the infinite - dimensional self - consistent mean - field approach.@xcite other methods , such as slave - boson@xcite or slave - fermion@xcite approaches , have also allowed one to gain insights into the hubbard model through various mean - field theories corrected for fluctuations . in this context however , the mean - field theories are not based on a variational principle . instead , they are generally based on expansions in the inverse of a degeneracy parameter,@xcite such as the number of fermion flavors @xmath2 , where @xmath2 is taken to be large despite the fact that the physical limit corresponds to a small value of this parameter , say @xmath3 . hence these theories must be used in conjunction with other approaches to estimate their limits of validity.@xcite expansions around solvable limits have also been explored.@xcite finally , numerical solutions,@xcite with proper account of finite - size effects , can often provide a way to test the range of validity of approximation methods independently of experiments on materials that are generally described by much more complicated hamiltonians . despite all this progress
, we are still lacking reliable theoretical methods that work in arbitrary space dimension . in two dimensions in particular
, it is believed that the hubbard model may hold the key to understanding normal state properties of high - temperature superconductors .
but even the simpler goal of understanding the magnetic phase diagram of the hubbard model in two dimensions is a challenge .
traditional mean - field techniques , or even slave - boson mean - field approaches , for studying magnetic instabilities of interacting electrons fail in two dimensions .
the random phase approximation ( rpa ) for example does not satisfy the pauli principle , and furthermore it predicts finite temperature antiferromagnetic or spin density wave ( sdw ) transitions while this is forbidden by the mermin - wagner theorem .
even though one can study universal critical behavior using various forms of renormalization group treatments@xcite@xcite@xcite@xcite@xcite or through the self - consistent - renormalized approach of moriya@xcite which all satisfy the mermin - wagner theorem in two dimensions , cutoff - dependent scales are left undetermined by these approaches .
this means that the range of interactions or fillings for which a given type of ground - state magnetic order may appear is left undetermined . amongst the recently developed theoretical methods for understanding both collective and single - particle properties of the hubbard model , one should note the fluctuation exchange approximation@xcite ( flex ) and the pseudo - potential parquet approach.@xcite the first one , flex , is based on the idea of conserving approximations proposed by baym and kadanoff.@xcite@xcite this approach starts with a set of skeleton diagrams for the luttinger - ward functional@xcite to generate a self - energy that is computed self - consistently .
the choice of initial diagrams however is arbitrary and left to physical intuition . in the pseudo - potential parquet approach , one parameterizes response functions in all channels , and then one iterates crossing - symmetric many - body integral equations .
while the latter approach partially satisfies the pauli principle , it violates conservation laws .
the opposite is true for flex . in this paper , we present the formal aspects of a new approach that we have recently developed for the hubbard model @xcite@xcite .
the approach is based on enforcing sum rules and conservation laws , rather than on diagrammatic perturbative methods that are not valid for interaction @xmath4 larger than hopping @xmath5 .
we first start from a luttinger - ward functional that is parameterized by two irreducible vertices @xmath6 and @xmath7 that are local in space - time .
this generates rpa - like equations for spin and charge fluctuations that are conserving .
the local - moment sum rule , local charge sum rule , and the constraint imposed by the pauli principle , @xmath8 then allow us to find the vertices as a function of double occupancy @xmath9 ( see eqs.([charge ] ) and ( [ spin ] ) ) . since @xmath10 is a local quantity it depends very little on the size of the system and , in principle , it could be obtained reliably using numerical methods , such as for example monte carlo simulations . here , however , we adopt another approach and find @xmath10 self - consistently@xcite without any input from outside the present theory .
this is done by using an _ ansatz _ eq .
( [ usp ] ) for the double - occupancy @xmath10 that has been inspired by ideas from the local field approach of singwi et al.@xcite .
once we have the spin and charge fluctuations , the next step is to use them to compute a new approximation , eq.([param ] ) , for the single - particle self - energy .
this approach to the calculation of the effect of collective modes on single - particle properties@xcite is similar in spirit to paramagnon theories.@xcite contrary to these approaches however , we _ do _ include vertex corrections in such a way that , if @xmath11 is our new approximation for the self - energy while @xmath12 is the initial green s function used in the calculation of the collective modes , and @xmath9 is the value obtained from spin and charge susceptibilities , then @xmath13 = u\left\langle n_{\uparrow } n_{\downarrow } \right\rangle $ ] is satisfied exactly .
the extent to which @xmath14 $ ] ( computed with @xmath15 instead of @xmath16 ) differs from @xmath17 can then be used both as an internal accuracy check and as a way to improve the vertex corrections .
if one is interested only in two - particle properties , namely spin and charge fluctuations , then this approach has the simple physical appeal of rpa but it satisfies key constraints that are always violated by rpa , namely the mermin - wagner theorem and the pauli principle . to contrast it with usual rpa , that has a self - consistency only at the single - particle level ,
we call it the two - particle self - consistent approach ( tpsc).@xcite@xcite@xcite the tpsc gives a _ quantitative _ description of the hubbard model not only far from phase transitions , but also upon entering the critical regime .
indeed we have shown quantitative agreement with monte carlo simulations of the nearest - neighbor@xcite and next - nearest neighbor@xcite hubbard model in two dimensions .
quantitative agreement is also obtained as one enters the narrow critical regime accessible in monte carlo simulations .
we also have shown@xcite in full generality that the tpsc approach gives the @xmath18 limit of the @xmath19 model , while @xmath20 is the physically correct ( heisenberg ) limit . in two dimensions ,
we then recover both quantum - critical@xcite and renormalized classical@xcite regimes to leading order in @xmath21 .
since there is no arbitrariness in cutoff , given a microscopic hubbard model no parameter is left undetermined .
this allows us to go with the same theory from the non - critical to the beginning of the critical regime , thus providing quantitative estimates for the magnetic phase diagram of the hubbard model , not only in two dimensions but also in higher dimensions@xcite .
the main limitation of the approach presented in this paper is that it is valid only from weak to intermediate coupling .
the strong - coupling case can not be treated with frequency - independent irreducible vertices , as will become clear later . however
, a suitable ansatz for these irreducible vertices in a luttinger - ward functional might allow us to apply our general scheme to this limit as well .
our approach predicts@xcite that in two dimensions , fermi liquid quasiparticles disappear in the renormalized classical regime @xmath22 , which always precedes the zero - temperature phase transition in two - dimensions . in this regime the antiferromagnetic correlation length becomes larger than the single - particle thermal de broglie wave length @xmath23 leading to the destruction of fermi liquid quasiparticles with a concomitant appearance of precursors of antiferromagnetic bands ( `` shadow bands '' ) with no quasi - particle peak between them .
we stress the crucial role of the classical thermal spin fluctuations and low dimensionality for the existence of this effect and contrast our results with the earlier results of kampf and schrieffer@xcite who used a susceptibility separable in momentum and frequency @xmath24 .
the latter form of @xmath24 leads to an artifact that dispersive precursors of antiferromagnetic bands can exist at @xmath25 ( for details see @xcite ) .
we also contrast our results with those obtained in the fluctuation exchange approximation ( flex ) , which includes self - consistency in the single particle propagators but neglects the corresponding vertex corrections .
the latter approach predicts only the so - called `` shadow feature '' @xcite which is an enhancement in the incoherent background of the spectral function due to antiferromagnetic fluctuations .
however , it does not predict@xcite the existence of `` shadow bands '' in the renormalized classical regime .
these bands occur when the condition @xmath26 is satisfied .
flex also predicts no pseudogap in the spectral function @xmath27 at half - filling @xcite . by analyzing temperature and size dependence of the monte carlo data and comparing them with the theoretical calculations
, we argue that the monte carlo data supports our conclusion that the precursors of antiferromagnetic bands and the pseudogap do appear in the renormalized classical regime .
we believe that the reason for which the flex approximation fails to reproduce this effect is essentially the same reason for which it fails to reproduce hubbard bands in the strong coupling limit . more specifically , the failure is due to an inconsistent treatment of vertex corrections in the self - energy _
ansatz_. contrary to the electron - phonon case , these vertex corrections have a strong tendency to cancel the effects of using dressed propagators in the expression for the self - energy .
recently , there have been very exciting developments in photoemission studies of the high-@xmath28 materials@xcite that show the opening of the pseudogap in single particle spectra above the superconducting phase transition . at present
, there is an intense debate about the physical origin of this phenomena and , in particular , whether it is of magnetic or of pairing origin . from the theoretical point of view
there are a lot of formal similarities in the description of antiferromagnetism in repulsive models and superconductivity in attractive models . in sec.([secdestruction ] ) we use this formal analogy to obtain a simple analytical expressions for the self - energy in the regime dominated by critical pairing fluctuations .
we then point out on the similarities and differences in the spectral function in the case of magnetic and pairing pseudogaps .
our approach has been described in simple physical terms in refs.@xcite and @xcite .
the plan of the present paper is as follows . after recalling the model and the notation
, we present our theory in sec.([secformal ] ) .
there we point out which exact requirements of many - body theory are satisfied , and which are violated . before sec.([secformal ] ) , the reader is urged to read appendix ( [ secsumrules ] ) that contains a summary of sum rules , conservation laws and other exact constraints .
although this discussion contains many original results , it is not in the main text since the more expert reader can refer to the appendix as need be .
we also illustrate in this appendix how an inconsistent treatment of the self - energy and vertex corrections can lead to the violation of a number of sum rules and inhibit the appearance of the hubbard bands , a subject also treated in sec.([secflex ] ) .
section ( [ secmontecarlo ] ) compares the results of our approach and of other approaches to monte carlo simulations .
we study in more details in sec.([secdestruction ] ) the renormalized classical regime at half - filling where , in two dimensions , fermi liquid quasiparticles are destroyed and replaced by precursors of antiferromagnetic bands well before the @xmath25 phase transition .
we also consider in this section the analogous phenomenon of pairing pseudogap which can appear in two dimensions when the pairing fluctuations become critical .
the following section sec.([secflex ] ) explains other attempts to obtain precursors of antiferromagnetic bands and points out why approaches such as flex fail to see the effect .
we conclude in sec.([secvalidity ] ) with a discussion of the domain of validity of our approach and in sec.([seccomparisons ] ) with a critical comparison with flex and pseudo - potential parquet approaches , listing the weaknesses and strengths of our approach compared with these .
a more systematic description and critique of various many - body approaches , as well as proofs of some of our results , appear in appendices .
we first present the model and various definitions .
the hubbard model is given by the hamiltonian , @xmath29 in this expression , the operator @xmath30 destroys an electron of spin @xmath31 at site @xmath32 .
its adjoint @xmath33 creates an electron and the number operator is defined by @xmath34 @xmath35 .
the symmetric hopping matrix @xmath36 determines the band structure , which here can be arbitrary .
double occupation of a site costs an energy @xmath4 due to the screened coulomb interaction .
we work in units where @xmath37 , @xmath38 and the lattice spacing is also unity , @xmath39 . as an example that occurs later
, the dispersion relation in the @xmath40-dimensional nearest - neighbor model is given by @xmath41 we will use a `` four''-vector notation @xmath42 for momentum - frequency space , and @xmath43 for position - imaginary time .
for example , the definition of the single - particle green s function can be written as @xmath44 where the brackets @xmath45 represent a thermal average in the grand canonical ensemble , @xmath46 is the time - ordering operator , and @xmath47 is imaginary time . in zero external field and in the absence of the symmetry breaking @xmath48 and
the fourier - matsubara transforms of the green s function are @xmath49 @xmath50 as usual , experimentally observable retarded quantities are obtained from the matsubara ones by analytical continuation @xmath51 in particular , the single - particle spectral weight @xmath52 is related to the single - particle propagator by @xmath53 @xmath54 the self - energy obeys dyson s equation , leading to @xmath55 it is convenient to use the following notation for real and imaginary parts of the analytically continued retarded self - energy @xmath56 causality and positivity of the spectral weight imply that @xmath57 finally , let us point out that for nearest - neighbor hopping , the hamiltonian is particle - hole symmetric at half - filling , @xmath58 with @xmath59 implying that @xmath60 and that , @xmath61 @xmath62 = -\left [ \sigma \left ( { \bf k+q},-ik_n\right ) -\frac u2\right ] .
\label{phs}\ ] ] we shall be primarily concerned with spin and charge fluctuations , which are the most important collective modes in the repulsive hubbard model .
let the charge and @xmath63 components of the spin operators at site @xmath32 be given respectively by @xmath64 @xmath65 the time evolution here is again that of the heisenberg representation in imaginary time .
the charge and spin susceptibilities in imaginary time are the responses to perturbations applied in imaginary - time .
for example , the linear response of the spin to an external field that couples linearly to the @xmath63 component @xmath66 is given by @xmath67 in an analogous manner , for charge we have @xmath68 here @xmath69 is the filling so that the disconnected piece is denoted @xmath70 .
it is well known that when analytically continued , these susceptibilities give physical retarded and advanced response functions .
in fact , the above two expressions are the imaginary - time version of the fluctuation - dissipation theorem .
the expansion of the above functions in matsubara frequencies uses even frequencies . defining the subscript @xmath71 to mean either charge or spin , we have @xmath72 @xmath73 \right\rangle \quad ; \quad
\chi _ { sp}^{\prime \prime } \left ( { \bf q,}% t\right ) = \frac 12\left\langle \left [ s_{{\bf q}}^z\left ( t\right ) , s_{-{\bf % q}}^z\left ( 0\right ) \right ] \right\rangle .\ ] ] the fact that @xmath74is real and odd in frequency in turn means that @xmath75 is real @xmath76 a convenient feature for numerical calculations .
the high - frequency expansion has @xmath77 as a leading term so that there is no discontinuity in @xmath78as @xmath79 contrary to the single - particle case .
to understand how to satisfy as well as possible the requirements imposed on many - body theory by exact results , such as those in appendix ( [ secsumrules ] ) , it is necessary to start from a general non - perturbative formulation of the many - body problem .
we thus first present a general approach to many - body theory that is set in the framework introduced by martin and schwinger@xcite , luttinger and ward@xcite and kadanoff and baym@xcite@xcite .
this allows one to see clearly the structure of the general theory expressed in terms of the one - particle irreducible self - energy and of the particle - hole irreducible vertices .
these quantities represent projected propagators and there is a great advantage in doing approximations for these quantities rather than directly on propagators . our own approximation to the hubbard model
is then described in the subsection that follows the formalism . in our approach ,
the irreducible quantities are determined from various consistency requirements .
the reader who is interested primarily in the results rather than in formal aspects of the theory can skip the next subsection and refer back later as needed . following kadanoff and baym,@xcite we introduce the generating function for the green s function @xmath80 = \ln \left\langle t_\tau e^{-c_{\overline{\sigma } % } ^{\dagger } \left ( \overline{1}\right ) c_{\overline{\sigma } } \left ( \overline{2}\right ) \phi _ { \overline{\sigma } } \left ( \overline{1},\overline{2% } \right ) } \right\rangle \label{z}\ ] ] where , as above , a bar over a number means summation over position and imaginary time and , similarly , a bar over a spin index means a sum over that spin index .
the quantity @xmath81 is a functional of @xmath82 , the position and imaginary - time dependent field .
@xmath81 reduces to the usual partition function when the field @xmath82 vanishes .
the one - particle green s function in the presence of this external field is given by @xmath83 \right ) = -\frac{\delta \ln z\left [ \phi \right ] } { \delta \phi _ \sigma \left ( 2,1\right ) } \label{gdephi}\ ] ] and , as shown by kadanoff and baym , the inverse green s function is related to the self - energy through @xmath84 the self - energy in this expression is a functional of @xmath85 .
performing a legendre transform on the generating functional @xmath86 $ ] in eq.([z ] ) with the help of the last two equations , one can find a functional @xmath87 $ ] of @xmath88 that acts as a generating function for the self - energy @xmath89 \right ) = \frac{\delta \phi \left [ g\right ] } { \delta g_\sigma \left ( 2,1\right ) } .
\label{selflutt}\ ] ] the quantity @xmath87 $ ] is the luttinger - ward functional.@xcite formally , it is expressed as the sum of all connected skeleton diagrams , with appropriate counting factors . conserving approximations start from a subset of all possible connected diagrams for @xmath87 $ ] to generate both the self - energy and the irreducible vertices entering the integral equation obeyed by response functions .
these response functions are then guaranteed to satisfy the conservation laws .
they obey integral equations containing as irreducible vertices @xmath90 \right ) } { \delta g_{\sigma ^{\prime } } \left ( 3,4\right ) } = \frac{\delta ^2\phi \left [ g\right ] } { \delta g_\sigma \left ( 2,1\right ) \delta g_{\sigma ^{\prime } } \left ( 3,4\right ) } = \gamma _ { \sigma ^{\prime } \sigma } ^{ir}(4,3;2,1 ) \label{irrlutt}\ ] ] a complete and exact picture of one- and two - particle properties is obtained then as follows .
first , the generalized susceptibilities @xmath91 are calculated by taking the functional derivative of @xmath92 and using the dyson equation ( [ dyson3 ] ) to compute @xmath93 .
one obtains@xcite @xmath94 where one recognizes the bethe - salpeter equation for the three - point susceptibility in the particle - hole channel .
the second equation that we need is automatically satisfied in an exact theory .
it relates the self - energy to the response function just discussed through the equation @xmath95 which is proven in appendix ( [ proofs ] ) . the diagrammatic representation of these two equations eqs.([1a])([2a ] ) appearing in fig.([figkadanoffbaym ] ) may make them look more familiar . despite this diagrammatic representation ,
we stress that this is only for illustrative purposes
. the rest of our discussion will not be diagrammatic .
because of the spin - rotational symmetry the above equations eqs.([1a ] ) and ( [ 2a ] ) can be decoupled into symmetric ( charge ) and antisymmetric ( spin ) parts , by introducing spin and charge irreducible vertices and generalized susceptibilities : @xmath96 @xmath97 the usual two - point susceptibilities are obtained from the generalized ones as @xmath98 .
the equation eq.([1a ] ) for the generalized spin susceptibility leads to @xmath99 and similarly for charge , but with the plus sign in front of the second term .
finally , one can write the exact equation eq.([2a ] ) for the self - energy in terms of the response functions as @xmath100 g_\sigma \left ( 1,\overline{2}\right ) . \label{2}\ ] ] our two key equations are thus those for the three - point susceptibilities , eq.([1 ] ) , and for the self - energy , eq.([2 ] ) .
it is clear from the derivation in appendix ( [ proofs ] ) that these equations are intimately related . in formulating approximation methods for the many - body problem , it is preferable to confine our ignorance to high - order correlation functions whose detailed momentum and frequency dependence is not singular and whose influence on the low energy physics comes only through averages over momentum and frequency .
we do this here by parameterizing the luttinger - ward functional by two constants @xmath101 and @xmath102 .
they play the role of particle - hole irreducible vertices that are eventually determined by enforcing sum rules and a self - consistency requirement at the two - particle level . in the present context , this functional can be also considered as the interacting part of a landau functional .
ansatz _ is , @xmath103 = \frac 12g_{\overline{\sigma } } \left ( \overline{1},% \overline{1}^{+}\right ) \gamma _ { \overline{\sigma } \overline{\sigma } % } ^{ir}g_{\overline{\sigma } } \left ( \overline{1},\overline{1}^{+}\right ) + % \frac 12g_{\overline{\sigma } } \left ( \overline{1},\overline{1}^{+}\right ) \gamma _ { \overline{\sigma } -\overline{\sigma } } ^{ir}g_{-\overline{\sigma } % } \left ( \overline{1},\overline{1}^{+}\right ) .
\label{trial}\ ] ] as in every conserving approximation , the self - energy and irreducible vertices are obtained from functional derivatives as in eq.([selflutt ] ) and eq.([irrlutt ] ) and then the collective modes are computed from the bethe - salpeter equation eq.([1 ] ) .
the above luttinger - ward functional gives a momentum and frequency independent self - energy@xcite , that can be absorbed in a chemical potential shift .
from the luttinger - ward functional , one also obtains two local particle - hole irreducible vertices @xmath104 and @xmath105 @xmath106 we denote the corresponding local spin and charge irreducible vertices as @xmath107 notice now that there are only two equal - time , equal - point ( _ i.e. _ local ) two - particle correlation functions in this problem , namely @xmath108 and @xmath109 the last one is completely determined by the pauli principle and by the known filling factor , while @xmath110 is the expectation value of the interaction term in the hamiltonian
. only one of these two correlators , namely @xmath110 , is unknown .
assume for the moment that it is known@xmath111 then , we can use the two sum rules eqs.([suschargesum ] ) and ( [ susspinsum ] ) that follow from the fluctuation - dissipation theorem and from the pauli principle to determine the two trial irreducible vertices from the known value of this one key local correlation functions . in the present notation ,
these two sum rules are @xmath112 @xmath113 and since the spin and charge susceptibilities entering these equations are obtained by solving the bethe - salpeter eq.([1 ] ) with the constant irreducible vertices eqs.([irreducible])([defvertex ] ) we have one equation for each of the irreducible vertices @xmath114 @xmath115 we used our usual short - hand notation for wave vector and matsubara frequency @xmath116 . since the self - energy corresponding to our trial luttinger - ward functional is constant , the irreducible susceptibilities take their non - interacting value @xmath117 the local pauli principle @xmath118 leads to the following important sum - rule @xmath119 = 2n - n^2 , \label{spinchargepauli0}\ ] ] which can be obtained by adding eqs.([spin]),([charge ] ) .
this sum - rule implies that effective interactions for spin @xmath6 and charge @xmath120 channels must be different from one another and hence that ordinary rpa is inconsistent with the pauli principle ( for details see appendix ( [ secconsttwopart ] ) ) .
eqs.([charge ] ) and ( [ spin ] ) determine @xmath6 and @xmath7 as a function of double occupancy @xmath121 .
since double occupancy is a local quantity it depends little on the size of the system .
it could be obtained reliably from a number of approaches , such as for example monte carlo simulations .
however , there is a way to obtain double - occupancy self - consistently@xcite without input from outside of the present theory .
it suffices to add to the above set of equations the relation @xmath122 eqs.([spin ] ) and ( [ usp ] ) then define a set of self - consistent equations for @xmath6 that involve only two - particle quantities .
ansatz _ is motivated by a similar approximation suggested by singwi et al.@xcite in the electron gas , which proved to be quite successful in that case . on a lattice
we will use it for @xmath123 .
the case @xmath124 can be mapped on the latter case using particle - hole transformation . in the context of the hubbard model with on - site repulsion ,
the physical meaning of eq .
( [ usp ] ) is that the effective interaction in the most singular spin channel , is reduced by the probability of having two electrons with the opposite spins on the same site .
consequently , the _ ansatz _ reproduces the kanamori - brueckner screening that inhibits ferromagnetism in the weak to intermediate coupling regime ( see also below ) .
we want to stress , however , that this _ ansatz _ is not a rigorous result like sum rules described above .
the plausible derivation of this _ ansatz _ can be found in refs.@xcite , @xcite as well as , in the present notation , in appendix ( [ ansatz ] ) .
we have called this approach two - particle self - consistent to contrast it with other conserving approximations like hartree - fock or fluctuation exchange approximation ( flex)@xcite that are self - consistent at the one - particle level , but not at the two - particle level .
this approach@xcite to the calculation of spin and charge fluctuations satisfies the pauli principle @xmath125 by construction , and it also satisfies the mermin - wagner theorem in two dimensions . to demonstrate that this theorem is satisfied , it suffices to show that @xmath126 does not grow indefinitely .
( this guarantees that the constant @xmath127 appearing in eq.([integlorentz = c ] ) is finite . )
to see how this occurs , write the self - consistency condition eq.([spin ] ) in the form @xmath128 consider increasing @xmath121 on the right - hand side of this equation .
this leads to a decrease of the same quantity on the left - hand side .
there is thus negative feedback in this equation that will make the self - consistent solution finite .
a more direct proof by contradiction has been given in ref.@xcite : suppose that there is a phase transition , in other words suppose that @xmath129 then the zero - matsubara frequency contribution to the right - hand side of eq.([spin2 ] ) becomes infinite and positive in two dimensions as one can see from phase - space arguments ( see eq.([integlorentz = c ] ) ) .
this implies that @xmath108 on the left - hand side must become negative and infinite , but that contradicts the starting hypothesis since @xmath130 means that @xmath131 is positive .
although there is no finite - temperature phase transition , our theory shows that sufficiently close to half - filling ( see sec.([secpd ] ) ) there is a crossover temperature @xmath132 below which the system enters the so - called renormalized classical regime , where antiferromagnetic correlations grow exponentially .
this will be discussed in detail in sec.([secrc ] ) .
kanamori - brueckner screening is also included as we already mentioned above . to see how the screening occurs , consider a case away from half - filling , where one is far from a phase transition . in this case , the denominator in the self - consistency condition can be expanded to linear order in @xmath4 and one obtains @xmath133 where @xmath134 clearly , quantum fluctuations contribute to the sum appearing above and hence to the renormalization of @xmath135 @xmath136 the value of @xmath137 is found to be near @xmath138 as in explicit numerical calculations of the maximally crossed kanamori - brueckner diagrams.@xcite at large @xmath4 , the value of @xmath139 saturates to a value of the order of the inverse bandwidth which corresponds to the energy cost for creating a node in the two - body wave function , in agreement with the physics described by kanamori.@xcite to illustrate the dependence of @xmath140 on bare @xmath4 we give in fig.([figuspuch ] ) a plot of these quantities at half - filling where the correlation effects are strongest .
the temperature for this plot is chosen to be above the crossover temperature @xmath132 to the renormalized classical regime , in which case the dependence of @xmath6 and @xmath7 on temperature is not significant . as one can see , @xmath6 rapidly saturates to a fraction of the bandwidth , while @xmath7 rapidly increases with @xmath4 , reflecting the tendency to the mott transition .
we have also shown previously in fig.(2 ) of ref.@xcite that @xmath6 depends only weakly on filling .
since @xmath6 saturates as a function of @xmath4 due to kanamori - brueckner screening , the crossover temperature @xmath132 also saturates as a function of @xmath4 .
this is illustrated in fig.([figtxu ] ) along with the mean - field transition temperature that , by contrast , increases rapidly with @xmath141 quantitative agreement with monte carlo simulations on the nearest - neighbor@xcite and next - nearest - neighbor models@xcite is obtained@xcite for all fillings and temperatures in the weak to intermediate coupling regime @xmath142 .
this is discussed further below in sec.([secmontecarlo ] ) .
we have also shown that the above approach reproduces both quantum - critical and renormalized - classical regimes in two dimensions to leading order in the @xmath21 expansion ( spherical model)@xcite . as judged by comparisons with monte carlo simulations@xcite ,
the particle - particle channel in the repulsive two - dimensional hubbard model is relatively well described by more standard perturbative approaches .
although our approach can be extended to this channel as well , we do not consider it directly in this paper .
it manifests itself only indirectly through the renormalization of @xmath6 and @xmath7 that it produces . as in any implementation of conserving approximations , the initial guess for the self - energy , @xmath143 , obtained from the trial luttinger - ward functional
is inconsistent with the exact self - energy formula eq.([2 ] ) .
the latter formula takes into account the feedback of the spin and charge collective modes actually calculated from the conserving approximation . in our approach
, we use this self - energy formula eq.([2 ] ) in an iterative manner to improve on our initial guess of the self - energy .
the resulting formula for an improved self - energy @xmath11 has the simple physical interpretation of paramagnon theories.@xcite as another way of physically explaining this point of view , consider the following : the bosonic collective modes are weakly dependent on the precise form of the single - particle excitations , as long as they have a quasiparticle structure .
in other words , zero - sound or paramagnons exist , whether the bethe - salpeter equation is solved with non - interacting particles or with quasiparticles .
the details of the single - particle self - energy by contrast can be strongly influenced by scattering from collective modes because these bosonic modes are low - lying excitations .
hence , we first compute the two - particle propagators with hartree - fock single - particle green s functions , and then we improve on the self - energy by including the effect of collective modes on single - particle properties . the fact that collective modes can be calculated first and
self - energy afterwards is reminiscent of renormalization group approaches,@xcite@xcite where collective modes are obtained at one - loop order while the non - trivial self - energy comes out only at two - loop order .
the derivation of the general self - energy formula eq.([2 ] ) given in appendix ( [ proofs ] ) shows that it basically comes from the definition of the self - energy and from the equation for the collective modes eq.([1 ] ) .
this also stands out clearly from the diagrammatic representation in fig.([figkadanoffbaym ] ) . by construction , these two equations eqs.([1 ] ) and ( [ 2 ] ) satisfy the consistency requirement @xmath144 ( see appendix ( [ proofs ] ) ) , which in momentum and frequency space can be written as @xmath145 the importance of the latter sum rule , or consistency requirement , for approximate theories should be clear from the appearance of the correlation function @xmath121 that played such an important role in determining the irreducible vertices and in obtaining the collective modes . using the fluctuation dissipation theorem eqs.([sumspin]),([sumcharge ] )
this sum - rule can be written in form that explicitly shows the relation between the self - energy and the spin and charge susceptibilities @xmath146 g_\sigma ( k)=% \frac u4\frac tn\sum_q\left [ \chi _ { ch}(q)-\chi _ { sp}(q)\right ] .
\label{sumsigma1}\ ] ] to keep as much as possible of this consistency , we use on the right - hand side of the self - energy expression eq.([2 ] ) the same irreducible vertices and green s functions as those that appear in the collective - mode calculation eq.([1 ] ) .
let us call @xmath12 the initial green s function corresponding to the initial luttinger - ward self - energy @xmath147 our new approximation for the self - energy @xmath148 then takes the form @xmath149 g_\sigma ^{\left ( 0\right ) } ( k+q ) .
\label{param}\ ] ] note that @xmath150 satisfies particle - hole symmetry eq.([phs ] ) where appropriate .
this self - energy expression ( [ param ] ) is physically appealing since , as expected from general skeleton diagrams , one of the vertices is the bare one @xmath4 , while the other vertex is dressed and given by @xmath6 or @xmath7 depending on the type of fluctuation being exchanged .
it is because migdal s theorem does not apply for this problem that @xmath6 and @xmath7 are different from the bare @xmath4 at one of the vertices . @xmath6 and @xmath7 here take care of vertex corrections.@xcite the use of the full @xmath151 instead of @xmath152 in the above expression eq.([param ] ) would be inconsistent with frequency - independent irreducible vertices . for the collective mode eq.([1 ] ) this is well known to lead to the violation of the conservation laws as was discussed in detail in the previous subsection . here
we insist that the same is true in the calculation of the effect of _ electronic _ collective modes on the single - particle properties .
formally , this is suggested by the similarity between the equation for the susceptibility eq.([1 ] ) and that for the self - energy eq.([2 ] ) in terms of irreducible vertices .
more importantly , two physical effects would be absent if one were to use full @xmath88 and frequency independent irreducible vertices .
first , upper and lower hubbard bands would not appear because the @xmath153 high - frequency behavior in eq.([selfhaut ] ) that is necessary to obtain the hubbard bands would set in too late , as we discuss in sec.([subsecconstsingle ] ) and in sec.([hubbardbands ] ) .
this result is also apparent from the fact that flex calculations in infinite dimension do not find upper and lower hubbard bands@xcite where the exact numerical solution does .
the other physical effect that would be absent is precursors of antiferromagnetic bands , sec.([secdestruction ] ) and the pseudogap in @xmath154 , that would not appear for reasons discussed in sec.([secflex ] ) .
we also will see in sec.([secmontecarlo ] ) below that flex calculations of the single - particle green s function , significantly disagree with monte carlo data , even away from half - filling , as was already shown in fig.1 of ref.@xcite .
the chemical potential for interacting electrons @xmath155 is found from the usual condition on particle number @xmath156 this chemical potential @xmath155 is , of course , different from @xmath157 but the luttinger sum rule @xmath158 is satisfied to a high accuracy ( about few percent ) for all fillings and temperatures @xmath159 . as usual
this occurs because the change in @xmath160 is compensated by the self - energy shift on the fermi surface @xmath161 . for @xmath162
there is some deviation from the luttinger sum rule which is due to the appearance of the precursors of the antiferromagnetic bands below @xmath132 ( sec .
[ secdestruction ] ) which develop into true sdw bands at @xmath25 .
it is important to realize that @xmath163 on the right hand side of the equation for the self - energy @xmath164 can not be calculated as @xmath165 , because otherwise it would not reduce to zero - temperature perturbation theory when it is appropriate . as was pointed out by luttinger ,
( see also section [ lat ] ) the `` non - interacting '' green s function used in the calculation for @xmath166 should be calculated as @xmath167 , where @xmath168 is calculated on the same level of accuracy as @xmath169 , i.e. from eq.([mu ] ) with @xmath170 . in our calculation
below , we approximate @xmath171 by @xmath157 because for the coupling strength and temperatures considered in this paper ( @xmath172 , @xmath159 ) the luttinger theorem is satisfied to high accuracy and the change of the fermi surface shape is insignificant . in addition , at half - filling the condition @xmath173 is satisfied exactly at any @xmath4 and @xmath174 because of particle - hole symmetry . for somewhat larger coupling strengths and away from half - filling , one may try to improve the theory by using @xmath175 , with @xmath176 and @xmath155 found self - consistently .
however , the domain of validity of our approach is limited to the weak - to - intermediate coupling regime since the strong - coupling regime requires frequency - dependent pseudopotentials ( see below ) . finally , let us note that , in the same spirit as landau theory , the only vertices entering our theory are of the type @xmath177 and @xmath178 or , through eq.([defvertex ] ) , @xmath179 and @xmath7 . in other words , we look at the problem from the longitudinal spin and charge particle - hole channel . consequently , in the contact pseudopotential approximation the exact equation for the self - energy eq.([2 ] ) reduces to our expression eq .
( [ param ] ) which does _ not _ have the factor @xmath180 in the front of the spin susceptibility .
this is different from some paramagnon theories , in which such factor was introduced to take care of rotational invariance .
however , we show in appendix ( [ paramagnon theories ] ) that these paramagnon theories are inconsistent with the sum - rule eq.([sumsigma1 ] ) which relates one and two - particle properties . in our approach ,
questions about transverse spin fluctuations are answered by invoking rotational invariance @xmath181 . in particular , one can write the expression for the self - energy eq.([param ] ) in an explicitly rotationally invariant form by replacing @xmath182 by @xmath183 $ ] .
if calculations had been done in the transverse channel , it would have been crucial to do them while simultaneously enforcing the pauli principle in that channel . in functional integration methods , it is well known that methods that enforce rotational invariance without enforcing the pauli principle at the same time give unphysical answers , such as the wrong factor @xmath184 in the rpa susceptibility@xcite @xmath185 or wrong hartree - fock ground state.@xcite the quantitative accuracy of the theory will be discussed in detail when we compare with monte carlo calculations in the next section . here
we show that we can use the consistency requirement between one- and two - particle properties eq.([consistency ] ) to gauge the accuracy of the theory from within the theory itself . the important advantage of the expression for the self - energy @xmath186 given by eq.([param ] )
is that , as shown in appendix ( [ proofs ] ) , it satisfies the consistency requirement between one- and two - particle properties eq.([consistency ] ) , in the following sense @xmath187 let @xmath188 be defined by @xmath189 ^{-1}\equiv g_0^{-1}-\sigma ^{\left ( 1\right ) } .$ ] we can use the fact that in an exact theory we should have @xmath190 $ ] in the above expression instead of @xmath191 $ ] to check the accuracy of the theory .
it suffices to compute by how much @xmath192 $ ] differs from @xmath190 .$ ] in the parameter range @xmath193 and @xmath194,@xmath174 arbitrary but not too deep in the , soon to be described , renormalized - classical regime , we find that @xmath195 $ ] differs from @xmath196 $ ] by at most @xmath197 another way to check the accuracy of our approach is to evaluate the right - hand side of the @xmath198sum rule eqs.([f2 ] ) with @xmath199 and to compare with the result that had been obtained with @xmath200 .
again we find the same @xmath201 disagreement , at worse , in the same parameter range .
as one can expect , this deviation is maximal at half - filling and becomes smaller away from it .
eq.([param ] ) for the self - energy @xmath202 already gives good agreement with monte carlo data but the accuracy can be improved even further by using the general consistency condition eq.([consistency ] ) on @xmath196 $ ] to improve on the approximation for vertex corrections .
to do so we replace @xmath6 and @xmath7 on the right - hand side of eq.([param ] ) by @xmath203 and @xmath204 with @xmath205 determined self - consistently in such a way that eq.([sums ] ) is satisfied with @xmath206 replaced by @xmath207 . for @xmath208
, we have @xmath209 . the slight difference between the irreducible vertices entering the collective modes and the vertex corrections entering the self - energy formula
can be understood from the fact that the replacement of irreducible vertices by constants is , in a way , justified by the mean - value theorem for integrals .
since the averages are not taken over the same variables , it is clear that the vertex corrections in the self - energy formula and irreducible vertices in the collective modes do not need to be strictly identical when they are approximated by constants .
before we move on to comparisons with monte carlo simulations , we stress that @xmath202 given by eq .
( [ param ] ) can not be substituted back into the calculation of @xmath210 by simply replacing @xmath211 with the dressed bubble @xmath212 . indeed
, this would violate conservation of spin and charge and @xmath213-sum rule .
in particular , the condition @xmath214 that follows from the ward identity ( [ ward ] ) would be violated as we see in eq.([wardtriste ] ) . in the next order
, one is forced to work with frequency - dependent irreducible vertices that offset the unphysical behavior of @xmath215 at non - zero frequencies .
in this section , we present a few numerical results and comparisons with monte carlo simulations .
we divide this section in two parts . in the first one we discuss data sufficiently far from half - filling , or at high enough temperature , where size effects are unimportant for systems sizes available in monte carlo simulations . in the second part
, we discuss data at half - filling . there ,
size effects become important below the crossover temperature @xmath132 where correlations start to grow exponentially ( sec.([secdestruction ] ) ) .
all single - particle properties are calculated with our approximation eq.([param ] ) for the self - energy using the vertex renormalization @xmath205 explained in the previous section .
the results would differ at worse by @xmath201 if we had used @xmath216 we have shown previously in fig.4a - d of ref.@xcite and in figs.2 - 4 and fig.6 of ref.@xcite that both spin and charge structure factor sufficiently away from the crossover temperature @xmath132 are in quantitative agreement with monte carlo data for values of @xmath4 as large as the bandwidth . on the qualitative level , the decrease in charge fluctuations as one approaches half - filling has been explained@xcite as a consequence of the pauli principle embodied in the calculation of the irreducible vertex @xmath217@xcite here we present on fig.([figchiomega ] ) and fig.([figchit ] ) comparisons with a dynamical quantity , namely the spin susceptibility .
similar comparisons , but with a phenomenological value of @xmath6 , have been done by bulut _
_ ref.@xcite .
fig.([figchiomega ] ) shows the staggered spin susceptibility as a function of matsubara frequencies for @xmath218 @xmath219 and @xmath220 .
the effect of interactions is already quite large for the zero - frequency susceptibility .
it is enhanced by a factor of over @xmath221 compared with the non - interacting value .
nevertheless , one can see that the theory and monte carlo simulations are in good agreement .
fig.([figchit ] ) shows the temperature dependence of the zero - frequency staggered spin susceptibility for the same filling and interaction as in the previous figure .
symbols represent monte carlo simulations from refs.@xcite and @xcite , the solid line is for our theory while dotted and dashed lines are for two versions of flex . surprisingly , the fully conserving flex theory , ( dashed line ) compares worse with monte carlo data than the non - conserving version of this theory that neglects the so - called aslamasov - larkin diagrams ( dotted line ) .
by contrast , our theory is in better agreement with the monte carlo data than flex for the staggered susceptibility @xmath222 , and at the same time it agrees exactly with the conservation law that states that @xmath223 .
finally , fig.([figgupdown ] ) shows the double occupancy @xmath224 as a function of filling for various values of @xmath4 .
the symbols again represent monte carlo data for @xmath225 , and the lines are the results of our theory . everywhere
the agreement is very good , except for @xmath226 in the latter case , the system is already below the crossover temperature @xmath132 to the renormalized classical regime . as explained in sec.([secvalidity ] ) , the appropriate procedure for calculating double occupancy in this case is to take for @xmath227 its value ( dotted line ) at @xmath132 instead of using the _ ansatz _ eq.([usp ] ) . in any case , the difference is not large .
fig.1(a ) of ref.@xcite shows @xmath228 for filling @xmath229 , temperature @xmath219 and @xmath220 for the wave vector on the @xmath230 lattice which is closest to the fermi surface , namely @xmath231 .
our theory is in agreement with monte carlo data and with the parquet approach@xcite but in this regime second - order perturbation theory for the self - energy gives the same result .
surprisingly , flex is the only theory that disagrees significantly with monte carlo data .
the good performance of perturbation theory ( see also @xcite ) can be explained in part by compensation between the renormalized vertices and susceptibilities ( @xmath232 , @xmath233 ; @xmath234 , @xmath235 ) .
we have also calculated re@xmath236 and compared with the monte carlo data in fig.2a of ref.@xcite obtained at @xmath237 , @xmath220 , @xmath238 our approach agrees with monte carlo data for all frequencies , but again second - order perturbation theory gives similar results .
the occurrence of the crossover temperature @xmath132 at half - filling is perhaps best illustrated in the upper part of fig.([figjoint ] ) by the behavior of the static structure factor @xmath239 for @xmath220 as a function of temperature .
when the correlation length becomes comparable to the size of the system used in monte carlo simulations,@xcite the static structure factor starts to increase rapidly , saturating to a value that increases with system size .
the solid line is calculated from our theory for an infinite lattice .
the monte carlo data follow our theoretical curve ( solid line ) until they saturate to a size - dependent value .
the theory correctly describes the static structure factor not only above @xmath132 but also as we enter the renormalized classical regime at @xmath132 .
analytical results for this regime are given in sec.([secrc ] ) .
note that the rpa mean - field transition temperature for this value of @xmath4 is more than three times larger than @xmath240 .
the size - dependence of monte carlo data for @xmath241 at all other values of @xmath242 available in simulations is negligible and our calculation for infinite system reproduces this data ( not shown ) .
equal - time ( frequency integrated ) single - particle properties are much less sensitive to precursor effects than dynamical quantities as we now proceed to show . for example
, @xmath243 is a sum of @xmath244 over all matsubara frequencies .
we have verified ( figure not shown ) that @xmath245 obtained from monte carlo simulations@xcite is given quite accurately by either second - order perturbation theory or by our theory .
this has very important consequences since , for this quantity , the non - interacting value differs from second - order perturbation theory by at most @xmath197 this means that the numerical value of the right - hand side of the @xmath213 sum - rule eq.([f2 ] ) is quite close to that obtained from the left - hand side using our expression for the spin and charge susceptibility .
one can also look in more details at @xmath246 itself instead of focusing on a sum rule .
fig.([fign(k ) ] ) shows a comparison of our theory and of second order perturbation theory with monte carlo data for @xmath246 obtained for a set of lattice sizes from @xmath247 to @xmath248 at @xmath249 @xmath250 , @xmath251 size effects appear unimportant for this quantity at this temperature .
these monte carlo data have been used in the past@xcite to extract a gap by comparison with mean field sdw theory . our theory for the same set of lattice sizes is in excellent agreement with monte carlo data and predicts a pseudogap at this temperature , as we will discuss below . however , for available values of @xmath252 on finite lattices , second order perturbation theory is also in reasonable agreement with monte carlo data for @xmath246 . since second order perturbation theory does not predict a pseudogap , this means that @xmath246 is not really sensitive to the opening of a pseudogap .
this is so both because of the finite temperature and because the wave vectors closest to the fermi surface are actually quite far on the appropriate scale .
for this filling , the value of @xmath253 is fixed to @xmath254 on the fermi surface itself .
it is thus necessary to find a dynamical quantity defined on the fermi surface whose temperature dependence will allow us to unambiguously identify the pseudogap regime in both theory and in monte carlo data .
the most dramatic effect is illustrated in the lower part of fig.([figjoint ] ) where we plot the quantity @xmath255 defined by@xcite@xcite @xmath256 the physical meaning of this quantity @xmath255 is that it is an average of the single - particle spectral weight @xmath257 within @xmath258 of the fermi level ( @xmath259 ) . when quasiparticles exist , this is the best estimate of the usual zero - temperature quasiparticle renormalization factor @xmath260 that can be obtained directly from imaginary - time monte carlo data . for non - interacting particles @xmath255 is unity . for a normal fermi liquid
it becomes equal to a constant less than unity as the temperature decreases since the width of the quasiparticle peak scales as @xmath261 and hence lies within @xmath174 of the fermi level .
however , contrary to the usual @xmath262 this quantity gives an estimate of the spectral weight @xmath263 around the fermi level , even if quasiparticles disappear and a pseudogap forms , as in the present case , ( see sec.([secdestruction ] ) ) .
one can clearly see from the lower part of fig.([figjoint ] ) that while second - order perturbation theory exhibits typical fermi - liquid behavior for @xmath264 , both monte carlo data@xcite and a numerical evaluation of our expression for the self - energy lead to a rapid fall - off of @xmath255 below @xmath132 ( for @xmath220 , @xmath265@xcite ) .
the rapid decrease of @xmath255 clearly suggests non fermi - liquid behavior .
we checked also that our theory reproduces the monte carlo size - dependence .
this dependence is explained analytically in sec.([subseceffectonqp ] ) . in ref.@xcite
we have shown that at half - filling , our theory gives better agreement with monte carlo data@xcite for @xmath266 than flex , parquet or second order perturbation theory . to gain a qualitative insight into the meaning of this drop in @xmath267
, we use the analytical results of the next section to plot in fig.([figspectralweight ] ) the value of @xmath268 this plot is obtained by retaining only the contribution of classical fluctuations eq.([somega ] ) to the self - energy .
one sees that above @xmath132 , there is a quasiparticle but that at @xmath269 a minimum instead of a maximum starts to develop at the fermi surface @xmath270 below @xmath132 , the quasiparticle maximum is replaced by two peaks that are the precursors of antiferromagnetic bands .
this is discussed in detail in much of the rest of this paper .
the main features predicted by our approach for the magnetic phase diagram of the nearest - neighbor hopping model have been given in ref.@xcite .
needless to say , all our considerations apply in the weak to intermediate coupling regime .
note also that both quantum critical and renormalized classical properties of this model have been studied in another publication@xcite .
the shape of the phase diagram that we find is illustrated in fig.([figjacksonphasediagram ] ) for @xmath271 and @xmath220 .
at zero temperature and small filling , the system is a paramagnetic fermi liquid , whatever the value of the interaction @xmath4 @xmath272 .
then , as one moves closer to half - filling , one hits a quantum critical point at a value of filling @xmath273 .
since , @xmath6 in our theory saturates with increasing @xmath4 , the value of @xmath273 is necessarily larger than about @xmath274 . at this point ,
incommensurate order sets in at a wave vector @xmath275 or at symmetry - related points . whatever the value of @xmath4 , the value of @xmath276 is contained@xcite in the interval @xmath277 , increasing monotonously towards @xmath278 as @xmath4 increases . since our approach applies only in the paramagnetic phase , at zero temperature we can not move closer to half - filling .
starting from finite - temperature then , the existence of long - range order at low temperature is signaled by the existence of a crossover temperature @xmath279 below which correlations start to grow exponentially .
we have already discussed the meaning of @xmath280 at half - filling .
this crossover temperature becomes smaller and smaller as one moves away from half - filling , until it reaches the quantum - critical point that we just discussed .
the correlations that start to grow at @xmath281 when @xmath282 are at the antiferromagnetic wave vector , and they stay at this wave vector for a range of fillings @xmath194 .
finally , at some filling , the correlations that start to grow at @xmath280 are at an incommensurate value until the quantum - critical point is reached .
note that the above phase diagram is quite different from the predictions of hartree - fock theory mostly because of the strong renormalization of @xmath6 .
this quantitative change leads to qualitative changes in the hartree - fock phase diagram since , for example , stoner ferromagnetism never occurs in our picture . while the existence of ferromagnetism in the _ strong coupling _
limit has been proven only recently@xcite , the absence of stoner ferromagnetism in the hubbard model was already suggested by kanamori@xcite a long time ago and was verified by more recent studies.@xcite@xcite@xcite more relevant to the present debate though , is the fact that sdw order persists away from half - filling for a finite range of dopings . while this is in agreement with slave - boson approaches@xcite and studies@xcite using the infinite - dimension
methodology,@xcite it is in clear disagreement with monte carlo simulations@xcite .
our approach certainly fails sufficiently below @xmath132 , but given the successes described above , we believe that it can correctly predict the exponential growth of fluctuations at @xmath132 .
it would be difficult to imagine how one could modify the theory in such a way that the growth of magnetic fluctuations does not occur even at incommensurate wave vectors . also , such an approach would also need to stop the growth of fluctuations that we find as we approach the quantum critical point along the zero temperature axis , from the low - filling , paramagnetic side , where @xmath283
. it could be that monte carlo simulations@xcite fail to see long - range order at zero temperature away from half - filling because at zero temperature , in the nearest - neighbor model , this order has a tendency to being incommensurate everywhere except at @xmath282 .
furthermore , as we saw above , this incommensuration is in general far from one of the available wave vectors on an @xmath230 lattice .
it comes close to @xmath284 only for the largest values of @xmath4 available by monte carlo .
hence , incommensurate order on small lattices is violently frustrated not only by the boundary conditions , but also by the fact that there is no wave vector on what would be the fermi surface of the infinite system .
this means that the electron - electron interaction scatters the electrons at wave - vectors that are not those where the instability would show up , rendering these scatterings not singular .
this is clearly an open problem .
one of the most striking consequences of the results discussed in the context of monte carlo simulations is the fall of the spectral weight below the temperature @xmath132 where antiferromagnetic fluctuations start to grow exponentially in two dimensions .
we have already shown in a previous publication@xcite that this corresponds to the disappearance of fermi liquid quasiparticles at the fermi surface , well above the zero temperature phase transition .
we also found that , simultaneously , precursors of the antiferromagnetic bands develop in the single - particle spectrum .
given the simplicity of our approach , it is possible to demonstrate this phenomenon analytically .
this is particularly important here because size effects and statistical errors make numerical continuation of the monte carlo data to real frequencies particularly difficult .
such analytic continuations using the maximum entropy method@xcite have , in the past , lead to a conclusion different from the one obtained later using singular value decomposition@xcite . in this section then , we will consider the conditions for which fermi liquid quasiparticles can be destroyed and replaced by a pseudogap in two dimensions .
the major part of this section will be concerned with the single particle pseudogap and the precursors of antiferromagnetic bands in the vicinity of the zero temperature antiferromagnetic phase transition in the positive @xmath4 hubbard model .
however , it is well known that the problem of superconductivity is formally related to the problem of antiferromagnetism , in particular at half - filling where the nearest - neighbor hopping positive @xmath4 hubbard model maps exactly onto the nearest - neighbor negative @xmath4 hubbard model .
the corresponding canonical transformation maps the @xmath286 spin correlations of the repulsive model onto the @xmath287 pairing and @xmath288 charge correlations of the attractive model while the single - particle green s functions of both models are identical .
thus all our results below concerning the opening of the pseudogap in @xmath289 in the repulsive @xmath4 half - filled hubbard model are directly applicable to the attractive @xmath4 model at half - filling , the only difference being in the physical interpretation . while in the case of repulsive interaction the pseudogap is due to the critical thermal spin fluctuation , in the case of attractive interactions it is , obviously , due to the critical thermal pairing and charge fluctuations .
away from half - filling the mapping between two models is more complicated and the single particle spectra in the pairing pseudogap regime @xmath289 have important qualitative differences with the single particle spectra in the magnetic pseudogap regime .
however , even in this case there are very useful formal similarities between two problems so that in subsec.([pairing ] ) we will give some simple analytical results for the self - energy in the regime dominated by critical pairing fluctuations .
the problem of precursor effects in the repulsive hubbard model has been first studied by kampf and schrieffer@xcite .
their analysis however was done at zero temperature and although the precursor effect that they found , called `` shadow bands '' , looks similar to what we find , there are a number of important differences .
for example , they find a quasiparticle between the precursors of antiferromagnetic bands , while we do not .
also , one does not obtain precursors at _ zero _ temperature when one uses our more standard expression for the dynamical susceptibility instead of the phenomenological form @xmath290 that they use .
the physical reason why a function that is separable in both momentum and frequency , such as @xmath291 , leads to qualitatively different results than the conventional one has been explained in ref.@xcite .
the microscopic justification for @xmath291 is unclear .
we comment below on this problem as well as on some of the large related literature that has appeared lately . repeating some of the arguments of ref.@xcite , we first show by general phase space arguments that the feedback of antiferromagnetic fluctuations on quasiparticles has the potential of being strong enough to destroy the fermi liquid only in low enough dimension , the upper critical dimension being three .
then we go into more detailed analysis to give explicit analytic expressions for the quasi - singular part of the self - energy , first in matsubara frequency .
the analysis of the self - energy expression directly in real - frequencies is in appendix ( [ apprealfrequ ] ) .
the latter analysis is useful to exhibit in the same formalism both the fermi liquid limit and the non - fermi liquid limit . for simplicity
we give asymptotics for @xmath282 at the fermi wave vector , where @xmath292 , but similar results apply for @xmath293 as long as there is long - range order at @xmath25 and one is below @xmath132 .
this case is also discussed briefly , but for more details the reader is referred to ref.@xcite . before describing the effect of spin fluctuations on quasiparticles , we first describe the so - called renormalized classical regime of spin fluctuations that precedes the zero - temperature phase transition in two dimensions .
the spin susceptibility @xmath294 below @xmath132 is almost singular at the antiferromagnetic wave vector @xmath295 because the energy scale @xmath296 ( @xmath297 ) associated with the proximity to the sdw instability becomes exponentially small.@xcite this small energy scale , @xmath298 , leads to the so - called renormalized classical regime for the fluctuations.@xcite in this regime , the main contribution to the sum over matsubara frequencies entering the local - moment sum rule eq.([spin ] ) comes from @xmath299 and wave vectors @xmath300 near @xmath301 . approximating @xmath302 by its asymptotic form @xmath303 where @xmath304 , @xmath305 and @xmath306 we obtain , in @xmath40 dimensions @xmath307 where @xmath308 is the left - hand side of eq.([spin ] ) minus corrections @xmath309
that come from the sum over non - zero matsubara frequencies ( quantum effects ) and from @xmath310 .
there is an upper cutoff to the integral which is less than or of the order of the brillouin zone size .
the important point is that the left - hand side of the above equation eq.([ksi1 ] ) is bounded and weakly dependent on temperature .
this implies , as discussed in detail in ref.@xcite , that the above equation leads to critical exponents for the correlation length that are in the spherical model ( @xmath311 universality class .
for our purposes , it suffices to notice that the integral converges even when @xmath312 in more than two dimensions .
this leads to a finite transition temperature . in two dimensions ,
the transition temperature is pushed down to zero temperature and , doing the integral , one is left with a correlation length @xmath313 that grows exponentially below @xmath132@xmath314
the important consequence of this is that , below @xmath132 , the correlation length quickly becomes larger than the single - particle thermal de broglie wave length @xmath315 .
this has dramatic consequences on quasiparticles in two dimensions .
when the classical fluctuations ( @xmath316 ) become critical , they also give , in two dimensions , a dominant contribution to the self - energy _ at low frequency_. to illustrate what we mean by the classical frequency contribution , neglect the contribution of charge fluctuations and single out the zero matsubara frequency component from eq.([param ] ) to obtain @xmath317 @xmath318 here , @xmath319 is measured relative to the chemical potential .
the last term is the contribution from quantum fluctuations . in this last term
, the sum over matsubara frequencies @xmath320 must be done before the analytical continuation of @xmath321 to real frequencies otherwise this analytical continuation would involve going through complex plane poles of the other terms entering the full sum over @xmath322 the contribution from classical fluctuations , @xmath323 does not have this problem and furthermore it has the correct asymptotic behavior at @xmath324 .
hence the contribution of classical fluctuations to the retarded self - energy @xmath325 can obtained from the @xmath316 term by trivial analytical continuation @xmath326 note also that the chemical potential entering @xmath16 in the self - energy formula is @xmath327 at half - filling .
doing the same substitution as above for the asymptotic form of the spin susceptibility eq.([chias ] ) in the equation for the self - energy eq.([param ] ) one obtains the following contribution to @xmath164 from classical fluctuations @xmath328 where we have expanded @xmath329 . in the case that we consider , namely half - filling and @xmath330 , we have @xmath331 and @xmath332 the key point is again that in two dimensions the integral in this equation eq.([sigm ] ) is divergent at small @xmath333 for @xmath334 . in a fermi liquid
, the imaginary part of the self - energy at the fermi surface @xmath335 behaves as @xmath336 . here instead , we find a singular contribution @xmath337 that is proportional to @xmath338 in @xmath339 and hence is very large @xmath340 when the condition @xmath341 is realized .
by contrast , 0for @xmath342 , @xmath343 , so that the fermi liquid is destroyed only in a very narrow temperature range close the nel temperature @xmath344 .
dimensional analysis again suffices to show that in four dimensions the classical critical fluctuations do not lead to any singular behavior .
three dimensions then is the upper critical dimension . as usual , logarithmic corrections exist at the upper critical dimension . the effect will be very small in three dimensions not only because it is logarithmic , but also because the fluctuation regime is very small , extending only in a narrow temperature range around the nel temperature .
by contrast , in two dimensions the effect extends all the way from the crossover temperature , @xmath132 , which is of the order of the mean - field transition temperature , to zero temperature where the transition is .
wave vectors near van hove singularities are even more sensitive to classical thermal fluctuations .
indeed , near this point the expansion should be of the type @xmath345 .
this leads , in two dimensions , to even stronger divergence in @xmath346 @xmath347 ^{-1 } $ ] @xcite .
even if the logarithmic divergence is cutoff the prefactor is larger by a factor of @xmath338 compared with points far from the van hove singularities .
let us analyze in more details the consequences of this singular contribution of critical fluctuations to the self - energy in two dimensions .
the integral appearing in the two - dimensional version of the expression for the self - energy , eq.([sigm ] ) , can be performed exactly@xcite
@xmath348 here @xmath349 is a regular part . as a first application
, we can use this expression to understand qualitatively both the temperature and size dependence of the monte carlo data for @xmath350appearing in fig.(2 ) of ref.@xcite or in the lower panel of fig.([figjoint ] ) .
indeed , @xmath350can be written as the alternating series @xmath351 . even though the series converges slowly , in the beginning of the renormalized classical regime and for qualitative purposes it suffices to use the first term of this series .
then , using the expressions for the correlation length eqs .
( [ ksi ] ) and for the self - energy ( [ sigmam ] ) .
one finds , @xmath352 on the infinite lattice , @xmath338 starts growing exponentially below @xmath132 , quickly becoming much larger than @xmath353 .
this implies @xmath354 . on finite lattices @xmath355 , which explains the size effect observed in monte carlo _
i.e. _ smaller @xmath356 for smaller size @xmath2 , ( @xmath357 for fig.([figjoint ] ) )
. the analytic continuation of @xmath358 in eq.([sigmam ] ) is @xmath359 + { \cal r}. \label{somega}\ ] ] for the wave vectors @xmath360 away from the fermi surface the anomalous contribution due to the classical fluctuation has a similar form but with @xmath361 replaced by @xmath362 .
when @xmath363 , the correlation length @xmath338 becomes of order unity and , as we will show in appendix ( [ apprealfrequ ] ) , the regular part @xmath349 dominates so that one recovers standard fermi liquid behavior .
furthermore , even for large correlation length the regular part can not be neglected when @xmath364 since the term exhibited here becomes small .
hence we concentrate on small frequencies and on @xmath365 where the regular part @xmath349 can be neglected .
exactly at the fermi level @xmath335 we recover the result of the previous section , namely that the imaginary part of the self - energy for @xmath341 increases exponentially when the temperature decreases , @xmath366 .
the above analysis shows by contradiction that in the paramagnetic state below @xmath367 there is no fermi - liquid quasiparticle at @xmath368 yet the symmetry of the system remains unbroken at any finite @xmath174 .
indeed , starting from quasiparticles @xmath369 we found that as temperature decreases , @xmath370 _ increases _ indefinitely instead of _ decreasing _ , in direct contradiction with the starting hypothesis .
by contrast , a self - consistent treatment where we use in eq.([param ] ) the full @xmath371 with a large @xmath372 shows that , for @xmath373 , @xmath372 remains large in @xmath339 and does not vanish as @xmath374 , again confirming that the system is not a fermi liquid in this regime ( see however sec.([subsecwhyflex ] ) below ) .
strong modifications to the usual fermi liquid picture also persist away from half - filling as long as @xmath375 , as we discuss later .
one can check that the large @xmath376 in two dimensions ( for @xmath377 leads to a pseudogap in the infinite lattice , contrary to the conclusion reached in ref.@xcite .
indeed , instead of a quasiparticle peak , the spectral weight @xmath378im@xmath379 has a _ minimum _ at the fermi level @xmath259 and two symmetrically located maxima away from it .
more specifically , for @xmath380 we have @xmath381 ^ 2+[ut/(8\xi _ 0 ^
2)]^2}. \label{spectralweight}\ ] ] the maxima are located at @xmath382 .
these two maxima away from zero frequency correspond to precursors of the zero - temperature antiferromagnetic ( or sdw ) bands ( shadow bands@xcite).there is no quasiparticle peak between these two maxima when @xmath383 .
this remains true in the case of no perfect nesting as well@xcite ( see also sec.([subsecaway ] ) ) .
we note that this is different from the results of the zero - temperature @xmath384 calculations of kampf and schrieffer@xcite that were based on a phenomenological susceptibility separable in momentum and frequency @xmath385 as was explained in ref.@xcite , the existence of precursors of antiferromagnetic bands ( shadow bands in the terminology of ref.@xcite ) _ at zero temperature _ is an artifact of the separable form of the susceptibility .
the third peak between the two precursors of antiferromagnetic bands that was found in ref.@xcite is due to the fact that at zero temperature the imaginary part of the self - energy @xmath386 is strictly zero at all @xmath360 . in our calculations , precursor bands appear only at finite temperature when the system is moving towards a zero - temperature phase transition . in this case , the imaginary part of the self - energy goes to infinity for @xmath360 on the `` shadow fermi surface '' @xmath387 and to zero at all other wave vectors .
this is consistent with the sdw result which we should recover at @xmath25 .
indeed , the latter result can be described by the self - energy @xmath388 which implies that the imaginary is a delta function @xmath389 instead of zero at all @xmath360 as in a fermi liquid .
we note also that analyticity and the zero value of @xmath390 in ref.@xcite automatically implies that the slope of the real part of the self - energy @xmath391 is negative .
by contrast , in our case @xmath392 ) is at the bottom left corner of fig.([akom ] ) with the corresponding spectral function @xmath289 @xmath393shown above it . in fig.([figspectralweight ] ) we have already shown the evolution of the spectral function @xmath289 with temperature .
the positions of the precursors of antiferromagnetic bands scale like @xmath394 which itself , at small coupling in two dimensions , scales like the mean field sdw transition temperature or gap ( see appendix b of ref.@xcite ) . as @xmath4 increases , the predicted positions of the maxima obtained from the asymptotic form eq.([spectralweight ] )
will be less accurate since they will be at intermediate frequencies and the regular quantum contribution to the self - energy will affect more and more the position of the peaks .
we have predicted@xcite that the exponential growth of the magnetic correlation length @xmath338 below @xmath132 will be accompanied by the appearance of precursors of sdw bands in @xmath289 with no quasiparticle peak between them .
by contrast with isotropic materials , in quasi - two - dimensional materials this effect should exist in a wide temperature range , from @xmath132 @xmath395 to the nel temperature @xmath344 ( @xmath396 ) .
although there are some formal similarities between the precursors of antiferromagnetic bands and the hubbard bands ( see sec.[secflex ] ) we would like to stress that these are two different physical phenomena .
a clear illustration of this is when a four peak structure exists in the spectral function @xmath397 two peaks being precursors of antiferromagnetic bands , and two peaks being upper and lower hubbard bands .
the main differences between these bands are in the @xmath360_@xmath398dependence _ of the self - energy @xmath399 and in the conditions for which these bands develop .
precursors of antiferromagnetic bands appear even for small @xmath4 in the renormalized classical regime @xmath365 , and their dispersion has the quasi - periodicity of the magnetic brillouin zone . in contrast ,
upper and lower hubbard bands are high - frequency features that appear only for sufficiently large @xmath400 and @xmath401 and have the periodicity of the whole brillouin zone in the paramagnetic state . furthermore , the existence of hubbard bands is not sensitive to dimensionality so they exist even in infinite dimension where the self - energy does not depends on momentum @xmath360 at all .
in contrast , the upper critical dimension for the precursors of antiferromagnetic bands is three ( see sec.([subsecabsin3d ] ) ) . in our theory
the precursors of antiferromagnetic bands come from the almost singular behavior of the zero matsubara frequency susceptibility @xmath402 , which leads to the characteristic behavior of @xmath403 with @xmath404 . on another hand ,
the hubbard bands appear in our theory because the high - frequency asymptotics @xmath405 has already set in for @xmath406 , and this leads to the bands at @xmath407 for @xmath408 .
( see for more details sec.([secflex ] ) ) .
the coefficient @xmath409 is determined by the sum over all matsubara frequencies and @xmath410 : @xmath411 $ ] .
it was noticed in monte carlo simulations@xcite@xcite that for intermediate @xmath4 , the spectral weight has four maxima .
we think that peaks at @xmath412 are hubbard bands , while the peaks closer to @xmath413 are precursors of antiferromagnetic bands .
if this interpretation is correct , then the latter peaks should disappear with increasing temperature when @xmath338 becomes smaller than @xmath353 , while the hubbard bands should exist as long as @xmath401 . while the location of the precursors of antiferromagnetic bands should be accurate in our theory
, the same will not be true for the location of the upper and lower hubbard bands .
this is because our theory is tuned to the low frequency behavior of the irreducible vertices and does not have the right numerical coefficient in the high - frequency expansion of the self - energy , as shown in eq.([og ] ) below .
nevertheless , our analytical approach to date is the only one that agrees at least qualitatively with the finding that precursors of antiferromagnetic bands as well as upper and lower hubbard bands can occur simultaneously .
note however that a four peak structure at @xmath282 was also obtained in ref.@xcite but the physical difference between hubbard bands and precursors of antiferromagnetic bands was not clearly spelled out .
we comment on recent findings of the flex approach in sec.([secflex])@xcite@xcite@xcite . in two dimensions ,
the finite - temperature phase is disordered , but the zero - temperature one is ordered and has a finite gap , except at the quantum critical point away from half - filling .
hence , precursors of antiferromagnetic bands that appear in the paramagnetic state do so with a finite pseudogap which appears consistent with the finite zero - temperature gap towards which the system is evolving .
by contrast , in higher dimensions the gap opens - up with a zero value at the transition temperature . based on this simple argument
, one does not expect precursors of antiferromagnetic bands in dimensions larger than two ( see , however , below ) . here
, we will also show that there is no phase space reasons for the existence of precursors of the antiferromagnetic bands when @xmath414 .
we have already shown that in three dimensions the quasiparticle at the fermi level at half - filling will have an imaginary part of the self - energy that grows like @xmath415 an effect that is much weaker than @xmath416 found in two dimensions . despite this small effect ,
in three dimensions the classical fluctuations do not affect the self - energy for energies larger than @xmath417 .
indeed , consider the contribution of classical thermal fluctuations to the self - energy eq.([sigm ] ) . in two dimensions
, we have for @xmath418@xmath419 \cong \frac{ut}{2\xi _ 0 ^ 2}\int \frac{d^2q}{(2\pi ) ^2}\frac 1{q^2+\xi ^{-2}}% \frac 1\omega .
\label{sigm2}\ ] ] which allows us to recover the approximate formula for the spectral weight given in eq.([spectralweight ] ) above . in three dimensions
however , this approximation can not be done because the integral is _ not _ dominated by small values of @xmath333 . to see this explicitly in three dimensions ,
consider the contribution of classical thermal fluctuations @xmath420
@xmath421 \frac 1{\omega + i\eta + v_fq_{\vert } } .\ ] ] as long as @xmath418 , the logarithmic singularity that develops at @xmath422 when @xmath423 is integrable and gives no singular contribution to the self - energy .
hence , unusual effects of classical thermal fluctuations are confined to the range of frequencies @xmath424 at higher frequencies , @xmath418 , all bosonic matsubara frequencies in eq.([param ] ) need to be taken into account and from phase space considerations alone there is no reason for the existence of precursors of antiferromagnetic bands in the @xmath425 case .
however , the existence of such bands in @xmath425 can not be completely excluded based on dimensional arguments alone because they occur at finite frequencies and strictly speaking they are non - universal . in particular , as discussed in ref.@xcite , one expects to see precursors that look like @xmath426 antiferromagnetic bands ( shadow bands ) in the vicinity of the finite temperature phase transition in strongly anisotropic quasi - two - dimensional material . on the other hand
, such bands do not generically exist in the almost isotropic @xmath425 case , because even in @xmath426 the conditions for such bands are quite stringent .
the difference between shadow bands and hubbard bands has been discussed in the previous subsection and the discussion of non - analyticities sometimes encountered in fermi liquid theory can be found in appendix ( [ apprealfrequ ] ) . close to half - filling , in the nearest - neighbor hopping model , one can enter a renormalized classical regime with large antiferromagnetic correlation length , even though the zero - temperature fermi surface properties may favor incommensurate correlations .
this renormalized - classical regime with large @xmath427 correlations occurs when @xmath428 . by arguments similar to those above
, one finds that in this regime one still has precursors of antiferromagnetic bands .
however , the chemical potential is in or near the lower precursor band and the system remains metallic .
the high - frequency precursor appears only below @xmath367 at @xmath429 . with second - neighbor hopping , the points of the fermi surface that intersect the magnetic brillouin zone ( hot spots )
behave as does the whole fermi surface of the nearest - neighbor ( nested ) case discussed above .
these questions were discussed in detail in ref.@xcite .
as we have already pointed out above , the results for the single particle spectra obtained for the half - filled nearest - neighbor hopping repulsive hubbard model can be directly applied to the corresponding attractive hubbard model , in which case the pseudogap opens up in the renormalized classical regime of pairing and charge fluctuations .
away from half - filling , the symmetry between charge and pair correlations is lost and pair fluctuations dominate , becoming infinite at the kosterlitz - thouless transition temperature .
this temperature is below the temperature at which the magnitude of the pair order parameter acquires rigidity despite the randomness of its phase .
one expects then that a pseudogap will also open in this case when the correlation length for pairing fluctuations becomes larger than the single - particle thermal de broglie wavelength @xmath430 .
this should occur below the crossover temperature to the renormalized classical regime of pairing fluctuations but above the kosterlitz - thouless transition temperature .
the quantitative microscopic theory for the negative @xmath4 hubbard model will be considered in a separate publication .
by contrast with all other sections of this paper , our considerations here will be more phenomenological .
nevertheless , they will allow us to present some analytical results for the self - energy obtained in the critical regime dominated by pairing fluctuations .
details of the model should not be very important since we are in a regime where everything is dominated by long wave length fluctuations .
the derivation of @xmath399 in the pairing case is a straightforward extension of what we did in the antiferromagnetic case ( see subsecs.([subseceffectonqp ] ) , ( [ subsecprecursors ] ) and ref.@xcite ) . in particular , in complete analogy with the magnetic case , the main contribution to the self - energy in the critical regime comes from the classical thermal fluctuations @xmath299 .
assuming some effective coupling constant @xmath431 between quasiparticles and pairing fluctuations , which in general can be momentum dependent , one can write in the one loop approximation @xmath432 here @xmath319 is the electron dispersion relative to the chemical potential , and all factors in front of integral are reabsorbed into the coupling constant @xmath431 .
this expression is similar to the expression eq.([sigm ] ) in the magnetic case but there are two important differences : _
i ) _ instead of @xmath433 we have now @xmath434 _ ii ) _ there is no minus sign in front of @xmath435 .
the first difference is due to the fact that superconductivity usually occurs with zero center of mass momentum for the pair , and hence the pairing susceptibility in the normal state @xmath436 must be peaked near @xmath437 , ( the integration variable @xmath410 in eq .
( [ sigm ] ) was measured relative to @xmath438 ) .
the second difference comes from the fact that we are now considering the contribution to @xmath164 coming from the particle - particle channel instead of the particle - hole channel . taking the integrals over @xmath410 and using the fact that small @xmath410 only will contribute we neglect the @xmath410 dependence of the coupling constant and obtain for the imaginary part of @xmath439 the following expression @xmath440 in the renormalized classical regime the pairing correlation length @xmath441 increases faster with decreasing temperature than @xmath442 .
consequently , @xmath443 tends to diverge with decreasing temperature and a pairing pseudogap in the spectral function @xmath444 opens up over the complete fermi surface , except maybe at a few points where @xmath445 .
this is different from the antiferromagnetic case , where the pseudogap in @xmath446 opens up only when , so called , `` hot spots '' ( @xmath447 ) exist in a given model@xcite .
the antiferromagnetic pseudogap opens everywhere on the fermi surface only in the case of perfect nesting , where all points on the fermi surface are `` hot spots '' .
the real part of the self - energy can be obtained from eq.([sigi ] ) using the kramers - kronig relation and has the form : @xmath448 to understand how precursors of the superconducting bands develop , let us look at @xmath449 at frequencies @xmath450 . in this case , using inversion symmetry @xmath451 , one can obtain from eq.([sigr ] ) the following asymptotic form @xmath452 when , @xmath453 ( see , more general case below ) this form of the self - energy leads to the usual bcs result @xmath454 with the gap @xmath455 . on the other hand , the imaginary part @xmath456 eq.([sigi ] ) , vanishes everywhere in the @xmath25 limit , except when @xmath457 where it becomes infinite .
the results for @xmath458 and @xmath459 can thus be combined to write for the corresponding limit of the retarded self - energy @xmath460 .
this limit leads to the standard bcs expression for the normal green s function when substituted back into the dyson equation @xmath461 . above the transition temperature ,
the anomalous green s function remains zero since there is no broken symmetry . the qualitative picture for the development of the pairing pseudogap and of the precursors of superconducting bands at @xmath462
is illustrated in fig.([figspectralweight ] ) and in the left part of fig.([akom ] ) .
while in the case of magnetic critical fluctuations these figures describe the precursor effect in @xmath27 for perfect - nesting or for the `` hot spots '' ( when such points exist ) , in the case of pairing fluctuations they describe the spectra for all @xmath463 and for all fillings where the ground state is superconducting .
we need to comment on a subtle difference between the antiferromagnetic and the pairing precursor effects in the single particle spectra . while the magnetic order parameter has three components and can order only at zero temperature in the two - dimensional repulsive model , away from half - filling in the attractive model the pairing order parameter becomes the only relevant order parameter at low temperature .
since it has only two components , a finite temperature kosterlitz - thouless phase transition is then allowed in two dimensions . the critical behavior in vicinity of this transition
is given by @xmath464 $ ] instead of @xmath465 as in the magnetic case .
to take this properly into account one would need a treatment of the problem that is more sophisticated than that given above .
in particular , one would have to take into account corrections to the simple form that we used for the pairing susceptibility @xmath466 this lorentzian form of the susceptibility in the critical regime is strictly valid only in the @xmath467 limit ( @xmath194 is the number of the components of the order parameter ) and is , clearly , a less accurate approximation in the case of pairing fluctuations @xmath468 than in the case of the antiferromagnetic fluctuations @xmath469 .
nevertheless , we believe that qualitatively the picture given above is correct for two reasons . first , because in the kosterlitz - thouless picture the magnitude of the order parameter is locally non - zero starting below a crossover temperature @xmath367 that is larger than the transition temperature @xmath470 .
it is only the phase that is globally decorrelated above @xmath470 .
this means that locally the quasiparticles are basically in a superconducting state even above @xmath471 a second reason to believe in the precursor effects is that the superfluid density and the gap are _ finite _ as @xmath472 and , hence , the two peak structure in @xmath473 exists even as the phase transition point is approached from the low - temperature side . by analogy with the antiferromagnetic case ,
this two peak structure should not immediately disappear when one increases the temperature slightly above @xmath470 .
finally , we point out that the precursor phenomenon described above has to be distinguished from , so - called , pre - formed pairs considered first by p. nozires and s. schmitt - rink@xcite ( see also @xcite ) .
these pre - formed pairs exist in any dimension when the coupling strength is sufficiently large , while the precursor effect considered above can be caused by arbitrarily small attractive interactions but only in two dimensions .
we think that recent monte carlo data on the negative @xmath474 hubbard model illustrates the opening of the single - particle pseudogap due to critical fluctuations , rather than a strong - coupling effect . in these simulations ,
the drop in the density of states at the fermi level should be accompanied by a simultaneous rapid increase of the pairing structure factor @xmath475 .
the latter must be exponential in the infinite 2d lattice and a size analysis of monte carlo data similar to the one shown on fig.([figjoint ] ) would be extremely helpful to clarify this issue .
in this section , we explain why the theories that use self - consistent propagators but neglect the corresponding frequency - dependent vertex corrections _
fail _ to see two important physical effects : namely upper and lower hubbard bands , as well as the precursors of antiferromagnetic bands that we just discussed .
the failure of this type of self - consistent schemes to correctly predict upper and lower hubbard bands has been realized a long time ago in the context of calculations in infinite dimension@xcite@xcite .
while one may brush aside this failure by claiming that high - energy phenomena are not so relevant to low - energy physics , we show that in fact these schemes also fail to reproduce the _ low - energy _ pseudogap and the precursors of antiferromagnetic bands for essentially the same reasons that they fail to see hubbard bands .
it is thus useful to start by a discussion of the better understood phenomenon of upper and lower hubbard bands and then to move to precursors of antiferromagnetic bands .
we first note that ordinary perturbation theory satisfies the correct high - frequency behavior eq.([selfhaut ] ) for the self - energy namely , for @xmath476@xmath477 it is the latter property that guarantees the existence of the hubbard bands for @xmath400 . to see this , consider the half - filled case . in this case
, @xmath478 , @xmath60 and one finds for the spectral weight @xmath479 which has pronounced maxima at the upper and lower hubbard bands , namely @xmath480has long as @xmath481is not too large .
since these results are obtained using high - frequency asymptotics , they are valid only when the asymptotic eq.([selfhaut ] ) has already set in when @xmath482 in the exact theory and in ordinary perturbation theory in terms of _ bare _ green functions @xmath12 , eq.([selfhaut ] ) is valid for @xmath483and the hubbard bands appears as soon as @xmath4become larger than @xmath484 . the fact that this simple high - frequency behavior sets in at the energy scale given by @xmath484rather than @xmath4 , even when @xmath485 , is a non - trivial consequence of the pauli principle .
to see this we first recall the exact result for the self - energy @xmath486 in the atomic limit@xcite @xmath487 formally , the atomic limit means that hopping is the smallest of all energy scales in the problem , including the temperature , @xmath488 , which is not a very interesting case .
however , the same arguments that have been used to derive the expression ( [ exact ] ) in the atomic limit can be used to show that eq.([exact ] ) is valid at any @xmath489when @xmath490 .
indeed , in the equations of motion for two - particle correlators@xcite one can neglect hopping terms when @xmath490 .
this is where the asymptotic behavior ( [ exact ] ) sets in since the equations of motion then immediately lend themselves to a solution without any additional approximation for the interacting term .
this solution is possible because the pauli principle @xmath491allows us to collapse three - particle correlation function which enters equation of motion to the two - particle one @xmath492 .
hence , the expression for atomic limit eq.([exact ] ) is also a general result for the self - energy that is valid for @xmath493at half - filling @xmath478 , @xmath60and the asymptotic ( [ selfhaut ] ) sets in at @xmath494 , as was pointed out above .
away from half - filling , as long as @xmath495and @xmath496are both much smaller than @xmath497(they both vanish at half - filling ) , the asymptotic behavior will also start at @xmath498 .
the situation is qualitatively different when one uses dressed green functions , but does not take into account the frequency dependence of the vertex , as it is done in flex ( see eq.[berksfl ] ) ) or for second - order perturbation theory with dressed @xmath88 .
for example , the second - order expression for @xmath486 in terms of full @xmath499does satisfy the asymptotics eq.([selfhaut ] ) , but it sets in too late , namely for @xmath500 , instead of @xmath490 .
indeed , when @xmath490 , the equation for the self - energy at half - filling in this type of theories reduces to @xmath501 where @xmath502with @xmath503a constant of proportionality involving the sum over all wave vectors and matsubara frequencies of the self - consistent dynamical susceptibilities . in a given theory
the value of @xmath503 may differ from its value @xmath504obtained from the exact result eq .
( [ exact ] ) , but its always of order unity .
the solution of eq.([assigflex ] ) @xmath505 has the analytically continued form @xmath506 @xmath507 from this one can immediately see that a @xmath153regime exists for re@xmath508only when @xmath509@xmath510 .
this means that such regime sets in too late to give the hubbard bands described by eq.([hlhubbard ] ) , because the hubbard bands occur at @xmath511 and for such @xmath512the asymptotic form @xmath513is not valid yet in flex and similar theories .
consequently , instead of well defined peaks at @xmath514 in the half - filled case , one obtains only long tails in the spectral function @xmath515 no matter how large @xmath4 is.@xcite(see also following subsection ) .
this explains why there is no hubbard bands in any theory that uses self - consistent green s functions , but neglects the frequency dependence of the vertex .
this is an explicit example that illustrates what seems to be a more general phenomenon when there is no migdal theorem for vertex corrections : a calculation with dressed green s functions but no frequency dependent vertex correction often gives worse results than a calculation done with bare green s functions and a frequency independent vertex . in this subsection
we describe the qualitative differences between our results and the results of flex approximations given by eq.([berksfl ] ) with regards to the `` shadow bands '' and explain why we believe that the failure of the flex to reproduce these bands is an artifact of that approximation . to avoid any confusion , we first clarify the terminology , because the term `` shadow bands '' has been used previously to describe different physical effects .
( see for details ref .
we note that the so - called shadow features discussed in @xcite , @xcite as well as the pseudogap in the _ total _ density of states @xmath516exist in both theories and we will not discuss them here . instead , we concentrate on the precursors of antiferromagnetic bands in the spectral function @xmath257which correspond to two new solutions of the quasi - particle equation @xmath517 we start by recalling a simple physical argument why the precursors of antiferromagnetic bands must exist at finite temperatures in the vicinity of the zero - temperature phase transition in two dimensions . this can be best understood by contrasting this case with isotropic 3d case where such precursor effect are highly unlikely .
( for a discussion of the strongly anisotropic case see sec.([subsecabsin3d ] ) ) .
indeed , in three dimensions there is a _
finite _ temperature phase transition and the gap is equal to zero at this temperature @xmath518 . consequently at @xmath344there is only one peak in the @xmath289at @xmath259which starts to split into two peaks only below @xmath344 .
based on this simple physical picture , one would not expect to see precursors of antiferromagnetic bands above @xmath344 in this case .
the situation is qualitatively different in two - dimensions where classical thermal fluctuations suppress long - range order at any finite temperature while at the @xmath25 phase transition the system goes directly into the ordered state with a _
finite _ gap .
clearly , the two peak structure in @xmath257 at @xmath25 can not disappear as soon as we raise the temperature . for simplicity
we again consider half - filling .
as we have seen in section ( [ subsecprecursors ] ) two new quasi - particle peaks do appear in the renormalized classical regime @xmath365 in our theory .
we have also found a pseudogap with the minimum at @xmath259 in this regime .
in contrast , the numerical solution of the flex equations@xcite found a spectral function with a single maximum in @xmath289at @xmath413even when @xmath519becomes strongly peaked at @xmath520 .
with decreasing temperature this central maximum becomes anomalously broad , but the two peak structure does not appear .
the clear deviation from the fermi liquid is signaled by the positive sign of @xmath521 .
however the value of @xmath522does not become larger than unity .
the latter would unavoidably lead to the existence of two new quasi - particle peaks away from @xmath259as is clear from the graphical solution of the quasiparticle equation eq.([qpeq ] ) shown on the bottom left panel of figs.([akom ] ) .
we now explain analytically the origin of these qualitative differences in the two theories . in our theory
@xmath523and hence it quickly becomes larger than unity in the renormalized classical regime @xmath524 .
in addition , for @xmath525the real part of the self - energy has the same behavior as in the ordered state @xmath526with @xmath527 .
the important point is that this asymptotic behavior @xmath526of the self - energy already sets in for @xmath528 .
it is this property that leads to the appearance of the precursors of antiferromagnetic bands at @xmath529 in a manner analogous to the appearance of the hubbard bands in the strong coupling limit that is discussed in the previous subsection .
let s now try to understand analytically what happens in the flex approximation .
as in our theory , the main contribution to the self - energy in the strongly fluctuating regime comes from the zero - frequency term in the matsubara sum in the equation for the self - energy ( eq.([classicalcontribution ] ) in our theory and eq.([berksfl ] ) in flex ) .
an upper bound of the effect of the critical spin fluctuations can be obtained by approximating @xmath530 .
then one immediately obtains the same expression for the self - energy as the one obtained in flex in the context of hubbard bands eq.([assigflex ] ) .
( the only difference is that the parameter @xmath531 is now defined by the zero - frequency matsubara contribution of @xmath519 , rather than by the sum over all matsubara frequencies . ) as we have already discussed in the context of hubbard bands , such a form for @xmath164 does not lead to the appearance of two new quasiparticle solution away from @xmath259 because the characteristic behavior @xmath532 sets in too late , namely for @xmath533 . in addition
, the slope of @xmath534 at @xmath535 does not diverge with decreasing temperature as in our theory but instead saturates to its value given by the analog of eq.([sigmaflexreal ] ) , _ i.e. _ @xmath536 . as we mentioned above , a value larger than unity
@xmath537 would guarantee the existence of two new solutions of the quasiparticle equation eq.([qpeq ] ) away from @xmath259 .
the right - hand side of fig.([akom ] ) illustrates clearly what happens in a flex - like approach such as eq.([assigflex ] ) . the contribution of classical fluctuations to the spectral weight does not lead to a fermi liquid since @xmath263 saturates to a finite width as temperature decreases , but nevertheless precursors of shadow bands do not occur because @xmath538 is bounded below unity .
( note that the spectral weight would not vanish so steeply at large frequencies if we had taken into account the quantum contribution of the spin fluctuations , as in full flex calculations . )
we just saw that the self - consistency in the propagators without corresponding self - consistency in the vertices inhibits the existence of the shadow bands in essentially the same way as it inhibits the existence of the hubbard bands .
it thus seems to us very likely that the absence of the precursors of antiferromagnetic bands below @xmath132 in flex is an artifact .
this conclusion can be reliably verified by comparison with monte carlo data despite the fact that the latter is done for finite lattices and in the matsubara formalism .
this was discussed in more detail in sec.([subsecprecursors ] ) .
here we just note that the temperature dependence of matsubara quantities such as @xmath539 and @xmath540 have a very characteristic form in the pseudogap regime . for example , @xmath541 in the pseudogap regime , while in flex we would expect a much weaker temperature dependence of this quantity ( the upper bound being given by the analog of eq.([sigmaflex ] ) ) .
we also would like to comment on the 1d model@xcite which describes the interaction of electrons with _ static _
spin fluctuations characterized by the susceptibility @xmath542 $ ] .
the nice thing about this model is that it has an exact solution which shows the development of shadow bands and of the pseudogap in @xmath289 .
a treatment similar to ours which uses non - interacting green s functions in the one - loop approximation also reproduces this feature@xcite .
however , the analogous approximation with dressed green s functions leads to eq.([assigflex ] ) and hence inhibits the existence of the `` shadow bands '' and of the pseudogap in @xmath289 .
in closing we comment on semantics and on the physical interpretation of some results obtained in the flex approximation .
the expression `` conserving approximation '' has been widely used to describe flex calculations of the single particle properties and , in particular , in the context of the shadow bands and of the failure of luttinger s theorem @xcite . the conserving aspect has been emphasized , but in fact the only desirable feature in the calculation of the single - particle properties is that the self - energy @xmath164 is obtained from a functional derivative of the luttinger - ward functional @xmath543 and hence it is guaranteed to satisfy luttinger s theorem whenever appropriate . only on the next level
does this scheme lead to a calculation of the `` true '' susceptibilities@xcite and of collective modes that satisfy conservation laws ( ward identities ) .
however , these `` true '' susceptibilities are never substituted back in the calculation of the self - energy and the effect of `` true '' collective modes on the single - particle spectrum is an open question in flex .
in fact , the rpa propagators @xmath544 appearing in the self - energy expression are different from susceptibilities from which collective modes should be computed and further they explicitly break conservation laws , as can be seen from the fact that rpa - like expressions @xmath545 with a _ dressed _
bubble @xmath215 have the unphysical properties that are mentioned in eqs.([wardtriste ] ) and ( [ ftriste ] ) of appendix ( [ secsumrules ] ) .
the fact that there are in effect two susceptibilities in the flex approximation leads , in our opinion , to some confusion and incorrect physical interpretation of the results in the literature .
in particular , it was argued that the non - fermi - liquid behavior and deviations from luttinger theorem found in flex @xcite are not due to critical thermal fluctuation in the vicinity of the phase transition but are rather the result of large @xmath4 .
the reasoning for such claim was that although the rpa susceptibilities @xmath546 is very strongly peaked at @xmath520 , the `` true '' flex susceptibility is not . in our opinion
, such claim could be justified only if one would substitute the `` true '' susceptibility back in the calculation of @xmath547 ( for example using the exact eq.([2 ] ) ) and found that the deviation from the luttinger theorem and other qualitative changes in @xmath548 increase with decreasing temperature without almost divergent behavior of the _ conserving _ susceptibility @xmath549 and of the static structure factor @xmath550 the monte carlo data on fig.([figjoint ] ) are also instructive since they clearly show that qualitative changes in the single - particle spectra occur when the system enters the renormalized classical regime with rapidly growing @xmath551 .
the fact that the flex `` true '' susceptibility does not show such behavior at half - filling@xcite tells us that it even more drastically disagrees with the monte carlo data than the rpa - like @xmath552 which enters the expression for self - energy . moreover , even away from half - filling the `` true '' susceptibility in flex at @xmath520 significantly underestimates the strength of the spin fluctuations , as is clear from the comparison with monte carlo data in fig.([figchit ] ) . in our opinion
the , so - called , `` true susceptibility '' in flex is the key element in the confusion surrounding the interpretation of flex results for the self - energy because the `` true susceptibility '' never comes in the calculation of the self - energy . for all practical purposes
these calculations of the self - energy should be considered as consistent with luttinger s theorem at @xmath553 but based on a non - conserving susceptibility .
consistency with conservation laws and consistency with luttinger s theorem are not identical requirements because to satisfy rigorously luttinger s theorem one needs that @xmath543 , while to have conserving susceptibilities one needs that the irreducible vertices used in bethe - salpeter equation eq.([1a ] ) should be obtained from @xmath554 .
our approach is not valid beyond intermediate coupling .
that is perhaps best illustrated by fig.([figtxu ] ) that shows that the crossover temperature first increases with @xmath4 and then saturates instead of decreasing .
the decrease is expected on general grounds from the fact that at strong coupling the tendency to antiferromagnetism should decrease roughly as @xmath555 the reason for this failure of our approach is clear . as we know from studies in infinite dimension@xcite , to account for strong - coupling effects
it is necessary to include at least a frequency dependence to the self - energy and to the corresponding irreducible vertices .
our theory also fails at half - filling deep in the renormalized classical regime , i.e. @xmath556 mainly for two reasons .
first , the _ ansatz _ @xmath557 , eq.([usp ] ) , fails in the sense that @xmath558 eventually reaches zero at @xmath559 because of the @xmath560 divergence in the irreducible susceptibility @xmath561 due to perfect nesting .
the physically appropriate choice for @xmath558 in the renormalized classical regime is to keep its value fixed to its crossover - temperature value ( see fig.([figgupdown ] ) and sec.([secmontecarlo ] ) ) .
the more serious reason why our approach fails for @xmath556 is that , as we just saw , critical fluctuations destroy completely the fermi liquid quasiparticles and lead to a pseudogap .
this invalidates our starting point .
it is likely that in a more self - consistent theory , the logarithmic divergence of the appropriate irreducible susceptibility will be cutoff by the pseudogap .
however , just a simple dressing of the green s function is not the correct solution to the problem because it would make the theory non - conserving , as we discussed in sec.([secconsttwopart ] ) .
one needs to take into account wave vector and energy dependent vertex corrections similar to those discussed by schrieffer@xcite@xcite .
in appendix ( [ discussionotherapproaches ] ) , we discuss in detail various theories , pointing out limitations and advantages based on the criteria established in appendix([subsecconstsingle ] ) and ( [ secconsttwopart ] ) .
more specifically , we include in our list of desirable properties , the local pauli principle @xmath562 the mermin - wagner theorem eq.([susspinsum ] ) , the ward identities eq.([ward ] ) , and @xmath198sum rule eq.([f2 ] ) , one - particle _ vs _ two - particle consistency @xmath563 eq.([consistency ] ) , luttinger s theorem , and the large frequency asymptotic for the self - energy eq.([selfhaut ] ) , which is important for the existence of the hubbard bands . comments on the forward scattering sum - rule eq.([forward ] ) appear in appendix ( [ lat ] ) . in the present section , we only state without proof where each theory has strengths and weaknesses . in standard paramagnon theories,@xcite@xcite
the spin and charge fluctuations are computed by rpa , using either bare or dressed green s functions .
then the fluctuations are feedback in the self - energy .
when rpa with bare green s functions are used for the collective modes , these satisfy the @xmath198sum rule , but that is the only one of our requirements that is satisfied by such theories . in conserving approximation
schemes@xcite,@xcite the mermin - wagner theorem , the luttinger theorem and conservation laws are satisfied , but none of the other above requirements are fulfilled . in the parquet approach,@xcite@xcite one enforces complete antisymmetry of the four point function by writing down fully crossing - symmetric equations for these . however , in actual calculations , the local pauli principle , the mermin wagner theorem , and the consistency between one and two particle properties are only approximately satisfied , while nothing enforces the other requirements . in our approach , the high - frequency asymptotics and luttinger s theorem are satisfied to a very good degree of approximation while all other properties in our list are exactly enforced .
let us specify the level of approximation .
luttinger s theorem is trivially satisfied with our initial approximation for the self - energy @xmath564 but at the next level of approximation , @xmath565 , one needs a new chemical potential to keep the electron density @xmath566 $ ] fixed . with this new chemical potential
the fermi surface volume is preserved to a very high accuracy .
finally , consider the high - frequency asymptotics .
since we use bare propagators , the high - frequency asymptotics comes in at the appropriate frequency scale , namely @xmath567 , which is crucial for the existence of the hubbard bands .
however , the coefficient of the @xmath568 term in the high - frequency expansion eq.([selfhaut ] ) is incorrect because our irreducible vertices @xmath6 and @xmath7 are tuned to the low frequencies . if one would take into account the frequency dependence of @xmath6 and @xmath120 and assume that at high frequency they become equal to the bare interaction @xmath4 , then one would recover the exact result , provided the pauli principle in the form of eq.([spinchargepauli0 ] ) is satisfied .
the difficulty with such a procedure is that frequency dependent irreducible vertices requires frequency dependent self - energy in the calculation of collective modes and that would make the theory much more complicated .
yet it is , probably , the only way to extend the theory to strong coupling .
we have presented a new simple approach@xcite@xcite to the repulsive single - band hubbard model .
we have also critically compared competing approaches , such as paramagnon , fluctuation exchange approximation , and pseudo - potential parquet approaches .
our approach is applicable for arbitrary band structure@xcite and gives us not only a _ quantitative _ description of the hubbard model , but also provide us with some _ qualitatively _ new results .
let us summarize our theory again .
we first obtain spin and charge fluctuations by a self - consistent parameterization of the two - particle effective interactions ( irreducible vertices ) that satisfies a number of exact constraints usually not fulfilled by standard diagrammatic approaches to the many - body problem
. then the influence of collective modes on single - particle properties is taken into account in such a way that single - particle properties are consistent with two - particle correlators , which describe these collective modes .
more specifically , our approach satisfies the following constraints : 1 . spin and charge susceptibilities , through the fluctuation - dissipation theorem ,
satisfy the pauli principle in the form @xmath0 as well as the local moment sum - rule , conservations laws and consistency with the equations of motion in a local - field - like approximation .
2 . in two dimensions ,
the spin fluctuations satisfy the mermin - wagner theorem .
the effect of collective modes on single - particle properties is obtained by a paramagnon - like formula that is consistent with the two - particle properties in the sense that the potential energy obtained from @xmath569 $ ] is identical to that obtained from applying the fluctuation - dissipation theorem to spin and charge susceptibilities .
vertex corrections are included not only in spin and charge susceptibilities @xmath570 but also in the self - energy formula . in the latter case , this takes into account the fact that there is no migdal theorem controlling the effect of spin and charge fluctuations on the self - energy .
the results for both single - particle and two - particle properties are in _ quantitative _ agreement with monte carlo simulations for all fillings , as long as @xmath4 is less than the bandwidth and @xmath174 is not much smaller than the crossover temperature @xmath132 where renormalized - classical behavior sets in . both quantum - critical and
renormalized - classical behavior can occur in certain parameter ranges but the critical behavior of our approach is that of the @xmath19 model with @xmath18.@xcite the main predictions of physical significance are as follows : 1 .
the theory predicts a magnetic phase diagram where magnetic order persists away from half - filling but with completely suppressed ferromagnetism .
2 . in the renormalized classical regime above the zero - temperature phase transition
, precursors of antiferromagnetic bands ( shadow bands ) appear in @xmath571 .
these precursors occur when @xmath383 ( or @xmath572)@xmath111 between these precursors of antiferromagnetic bands a pseudogap appears at half - filling , so that the fermi liquid quasiparticles are completely destroyed in a wide temperature range above the zero - temperature phase transition @xmath573 .
the upper critical dimension for this phenomenon is three .
we stress the qualitative difference between the hubbard bands and the precursors of antiferromagnetic bands and we predict that in two dimensions one may see both sets of bands simultaneously in certain parameter ranges .
this prediction is consistent with the results of numerical simulations.@xcite@xcite .
we know of only one other analytic approach@xcite which leads to similar four peak structure in the spectral function .
the zero temperature magnetic phase diagram is partly an open question because , despite the qualitative agreement with other analytical approaches , there is still an apparent contradiction with monte carlo simulations@xcite .
our prediction of precursors of antiferromagnetic bands on the other hand is in agreement with monte carlo simulations .
neither this effect nor upper and lower hubbard bands are observed in self - consistent schemes such as flex .
this is because of inconsistent treatment of the vertex and self - energy corrections in this approximation , as we have explained in sec.([secflex ] ) . however , if there was a migdal theorem for spin fluctuations , it would be justifiable to neglect the vertex corrections and keep only the self - energy effects as is done in the flex approximation .
the presence of precursors of antiferromagnetic bands in two - dimensions is then a clear case of qualitatively new physics that would _ not _ appear if there was a migdal theorem for spin fluctuations .
the same is true for the hubbard bands for large @xmath400 in any dimension .
we would like to state again clearly the nature of our critique of approximation schemes which are based on using migdal s theorem for systems with electron - electron interactions .
we do _ not _ imply that one does not need at all to take into account the feedback of the single - particle spectra on collective modes .
the only point that we want to make here is that , based on sum rules and comparison with monte carlo data , we see that frequency and momentum dependent corrections to the self - energy and to the vertex often tend to cancel one another and that ignoring this leads to qualitatively incorrect results , in particular , with regards to the pseudogap . in this paper
we were able to look only at the beginning of the renormalized classical regime when the pseudogap starts to form .
the truly self - consistent treatment of the one - particle and two - particle properties in the pseudogap regime remains an open and very challenging problem .
we hope that by extending our approach to the ordered state and looking at how the pseudogap starts to disappear as the temperature is raised , one can better understand how to develop a more self - consistent theory in the pseudogap regime .
we now point out how our approach can be extended in other directions . as we mentioned in sec.([secdestruction ] ) , the pseudogap and precursors of antiferromagnetic bands in the two - dimensional repulsive hubbard model have interesting analogs in the attractive hubbard model . in that model ,
one expects a pairing pseudogap and precursors of superconducting quasiparticle bands above @xmath28 . at half - filling
the negative and positive hubbard models are mapped onto one another by a canonical transformation and the present theory is directly applicable to the attractive case .
however , away from half - filling the mapping between the two models is more complicated and the microscopic theory requires additional sum - rule for pairing susceptibilities to find self - consistently the effective pairing interaction .
this work is now in progress .
the present approach can be also extended to stronger coupling @xmath400 .
again the key idea would be to parameterize the irreducible vertices , which have now to be frequency dependent , and then use the most important sum - rules to find the parameterization coefficients .
this will , of course , require solving much more complicated self - consistent equations than in the present approach , but we believe that the problem still can be made tractable . finally , we would like to make two comments about the magnetic and the pairing pseudogap in the context of high-@xmath28 superconductors , based on the results of our studies .
first , as was stressed in ref.@xcite , to understand clearly the physics of the single - particle pseudogap phenomena it is important to distinguish _
static _ short - range order from _ dynamical _ short - range order .
the former is defined by a nearly lorentzian form of the corresponding static structure factor @xmath574 ( @xmath438 in magnetic case , @xmath575 in the case of pairing ) , while the latter means only that the corresponding susceptibility @xmath576 has such a lorentzian form .
a condition for the existence of the single particle pseudogap in the vicinity of a given phase transition is that the corresponding short - range order is quasi - static ( _ i.e._@xmath577)@xcite .
experimentally , one can measure directly the dynamical spin structure factor @xmath578 , and then obtain the static structure factor through the integral @xmath579 .
even if the zero - frequency dynamical structure factor @xmath580 is very strongly peaked at @xmath581 it is possible that the static structure factor @xmath582 is only weakly momentum dependent@xcite .
thus in order to know whether one should expect to see the precursors of the antiferromagnetic bands and the corresponding pseudogap at a given doping and temperature it is necessary to obtain the _
static _ spin structure factor from the experimentally determined dynamical structure factor and then analyze its momentum dependence to see both if it is peaked and if it is quasi - two - dimensional .
the second comment that we would like to make is that both the pairing and the magnetic single - particle pseudogap discussed above are an effect of low dimensionality and hence they exist as long as there is a large two - dimensional fluctuating regime before the real three - dimensional phase transition . in this context
, a pairing pseudogap could exist on either side of optimal doping.@xcite the much larger temperature range over which a pseudogap appears in the underdoped compounds suggests that , in addition to pairing fluctuations , other thermal fluctuations ( charge , spin ... ) prohibit finite - temperature ordering.@xcite an example of this occurs in the attractive hubbard model where charge fluctuations push the kosterlitz - thouless temperature to zero at half - filling , precisely where the crossover temperature to the pseudogap regime is largest .
we thank a.e .
ruckenstein , liang chen and r. ct for helpful discussions and a. georges for pointing out useful references .
this work was partially supported by the natural sciences and engineering research council of canada ( nserc ) , the fonds pour la formation de chercheurs et laide la recherche from the government of qubec ( fcar ) , and ( for y.m.v . ) the national science foundation ( grant no .
nsf - dmr-91 - 20000 ) through the science and technology center for superconductivity and ( for a .- m.s.t . )
the killam foundation as well as the canadian institute for advanced research ( ciar ) .
a .- m.s.t . would like to thank the institute for theoretical physics , santa barbara for hospitality during the writing of this work .
partial support there was provided by the national science foundation under grant no .
phy94 - 07194 .
in this appendix , we recall well known constraints on many - body theory that follow from sum - rules and conservation laws and comment , wherever possible , on their physical meaning and on where commonly used approaches fail to satisfy these constraints . although we come back on a detailed discussion of various theories in a later appendix , we find it useful to include some of this discussion here to motivate our approach .
we consider in turn various results that would be satisfied by any exact solution of the many - body problem .
they are all consequences of either anticommutation relations alone ( pauli principle ) or of anticommutation relations and the heisenberg equations of motion .
we describe in turn : 1 .
the relation between self - energy and two - body correlation functions that embodies the details of the hamiltonian ; 2 .
sum rules for one - particle properties ; 3 . sum rules and constraints on two - particle properties , in particular _
f_-sum rule and ward identities that express conservation laws ; 4 . a few relations that are crucial in fermi liquid theory , namely luttinger s theorem and the forward scattering sum rule .
the self - energy ( we always mean one - particle irreducible self - energy ) is related to the potential energy , and hence to two - particle correlations through the expression eq.([consistency ] ) , which in the kadanoff and baym notation can be written as @xmath583 here , the index with an overbar , @xmath584 means that there is a sum over corresponding lattice positions and an integral over imaginary time .
the notation @xmath585 means that the imaginary time implicit in @xmath586 is @xmath587 where @xmath588 is a positive infinitesimal number .
eq.([consistency1 ] ) is an important consistency requirement between self - energy and double occupancy in the hubbard model that can easily be proven as follows . from the equations of motion for the single - particle green s function eq.([defg ] ) one finds , @xmath589 g_\sigma \left ( { \bf r}_\ell -{\bf r}_j,\tau _ i-\tau _ j\right)\ ] ] @xmath590 using the short - hand notation in eqs.([defg ] ) and ( [ gfur ] ) and the definition of self - energy ( dyson s equation ) the above equation is also written in the form , @xmath591 comparing the last two equations , the well known relation eq.([consistency1 ] ) ( or eq.([consistency ] ) ) between self - energy , green s function and potential energy follows .
so - called conserving@xcite approaches to the many - body problem violate the above consistency requirement eq.([consistency ] ) in the following sense .
the right - hand side can be computed from the collective modes using the fluctuation - dissipation theorem . in conserving approximations , this gives a result that is different from what is computed directly from the left - hand side of the equation , namely from the self - energy and from the green s function .
in fact , all many - body approaches satisfy the above consistency requirement at best in an approximate way .
however , it is a very important requirement and eq.([consistency ] ) plays a key role in our discussion . seen in matsubara frequency ,
it is a sum rule , or an integral constraint that involves all frequencies , large and small .
the spectral weight @xmath592 can be interpreted as a probability of having an electron in a state @xmath593 and it satisfies the normalization sum rule @xmath594 formally this is a consequence of the jump in the green s function at @xmath595 , as can be seen from calculating @xmath596 to do perturbation theory directly for the green s function to any _ finite
_ order would require that the interaction @xmath4 be small not only in comparison with the bandwidth @xmath484 but also in comparison with the smallest matsubara frequency @xmath597 .
also , the direct perturbation series for the green s function gives , after analytical continuation , poles of arbitrary high order located at the unperturbed energies .
these high - order poles are inconsistent with the simple pole ( or branch cut ) structure of the green s function predicted by the spectral representation .
furthermore , the high - order poles lead to a spectral weight that can be negative.@xcite the common way to get around these difficulties is to make approximations for the self - energy @xmath164 instead and then calculate the green s function using dyson s equation eq.([gself ] ) .
it is interesting to note that to satisfy the constraint eq.([a1 ] ) , it suffices that @xmath598 , defined by eq.([gself ] ) , has a finite limit as @xmath599 more constraints on approximations for the self - energy may be found by continuing this line of thought .
a systematic way of doing this is to do a high - frequency expansion for both the matsubara green s function and the self - energy and to find coefficients using sum - rules .
the sum - rules that we need then are@xcite @xmath600 , c_{{\bf k}\sigma } ^{\dagger } \right\ } \right\rangle = \epsilon _ { { \bf k}}-\mu + un_{-\sigma } \label{a2}\ ] ] @xmath601 where @xmath602 since we are in the paramagnetic state . using the spectral representation eq.([spectralg ] ) one can easily see that the above sum rules give the coefficients of the high - frequency expansion of the matsubara green s function @xmath603 the self - energy has the same analytic properties as the green s function . using its high frequency expansion in the expression for the green s function eq.([gself ] ) , one finds that the first term in eq.([highg ] ) , leads to the requirement that the self - energy has a finite limit at @xmath604 .
the second term fixes the value of this constant to the hartree - fock result , and the last and second - term combine to give the leading term in @xmath605of the self - energy high - frequency expansion . in short ,
we find the result quoted in eq.([selfhaut ] ) , namely @xmath606 the kramers - kronig relation for the self - energy @xmath607 = { \cal p}\int \frac{d\omega ^{\prime } } \pi \frac{% % tcimacro{\func{im } } % beginexpansion \mathop{\rm im}% % endexpansion \left [ \sigma _ \sigma ^r\left ( { \bf k,}\omega ^{\prime } \right ) \right ] } { % \omega ^{\prime } -\omega } . \label{kk - re_self}\ ] ] and the high - frequency result eq.([selfhaut2 ] ) imply the following sum - rule for the imaginary part of the self - energy @xmath608 = u^2n_{-\sigma } \left ( 1-n_{-\sigma } \right)\ ] ] important consequences of this equation are that for a given @xmath4 the integrated imaginary part of the self - energy is independent of temperature and is increasing towards half - filling .
the right - hand side of this equation is also a measure the width of the single - particle excitation spectrum , as can be seen from the spectral weight moments eqs.([a2 ] ) and ( [ a3 ] ) , @xmath609 ^2=u^2n_{-\sigma } \left ( 1-n_{-\sigma } \right)\ ] ] an important physical point is that the asymptotic behavior eq.([selfhaut2 ] ) is a necessary condition for the existence of upper and lower hubbard bands , as has been explained in sec.[hubbardbands ] .
however , it is important to realize that it is not a sufficient condition .
indeed , the following paradox has been noticed in explicit calculations in infinite dimensions@xcite,@xcite . while ordinary second - order perturbation theory with _ bare _ green functions @xmath610reproduces correctly the appearance of the hubbard bands with increasing @xmath611 , the perturbation theory with dressed green function @xmath612^{-1}$ ] does not .
the reason for this is that although the second - order expression for @xmath486 in terms of full @xmath88 does satisfy the asymptotics eq.([selfhaut2 ] ) , it sets in too late , namely for @xmath500 , instead of @xmath490 .
the fact that the asymptotics should start at @xmath613 even when @xmath400 is a non - trivial consequence of the pauli principle , as explained in sec.([hubbardbands ] ) .
thus there are no hubbard bands in any theory that uses self - consistent green functions but neglects the frequency dependence of the vertex .
this is an explicit example that illustrates what seems to be a more general phenomena : a calculation with dressed green s functions but no frequency dependent vertex correction often gives worse results that the one done with bare green s functions and a frequency independent vertex .
we will see in the next subsection that this also happens in the calculation of the two - particle properties .
also , as we have argued in sec.([secflex ] ) , a similar situation occurs with the precursors of antiferromagnetic bands in the renormalized classical regime in two - dimensions . finally , we quote two more well known sum - rules that we will need .
they involve the fermi function @xmath614 and the spectral weight .
the first one follows from definition of @xmath615 and the spectral representation @xmath616 the quantity @xmath617 is the distribution function .
it is equal to the fermi function only when the self - energy is frequency independent .
the next result , that follows simply from the equations of motion , @xmath618 is useful to show to what extent certain dressed - propagator approaches fail to satisfy the @xmath198sum rule .
for any one - band model , independently of the hamiltonian , the pauli principle ( anticommutation relations ) @xmath619 implies the following two simple identities : @xmath620 the correlation functions on the left - hand side are equal - time and equal - position spin and charge correlation functions .
the susceptibilities @xmath621 @xmath622 in eqs.([suscharge ] ) and ( [ susspin ] ) are response functions for arbitrary @xmath623 so they must reduce to the above equal - time equal - position correlation functions when @xmath624 and @xmath625 .
this is one special case of the imaginary - time version of the fluctuation - dissipation theorem eqs.([susspin])([suscharge ] ) .
this translates into local - moment and local - charge sum - rules for the susceptibilities @xmath626 @xmath627 where we have removed the @xmath32 dependence of @xmath628 using translational invariance .
the right - hand side of the local - moment sum - rule is equal to @xmath629while that of the local - charge sum rule is equal to @xmath630 if arbitrary sets of diagrams are summed , nothing can prevent the right - hand side from taking unphysical values . for example , the pauli principle may be violated , _
i.e. _ @xmath631 to see this , notice that when the pauli principle is satisfied , our two sum rules eqs.([susspinsum ] ) and ( [ suschargesum ] ) lead to @xmath119 = 2n - n^2 .
\label{spinchargepauli}\ ] ] it is easy to check that well known approaches to the many - body problem , such as rpa , violate this basic requirement .
indeed , the ordinary rpa expressions for spin and charge are @xmath632 @xmath633 where @xmath634 that rpa does not satisfy the sum rule eq.([spinchargepauli ] ) already to second order in @xmath4 can be easily seen by expanding the denominators . to satisfy the mermin - wagner theorem , approximate theories must also prevent @xmath9 from taking unphysical values .
this quantity is positive and bounded by its value for @xmath635 and its value for non - interacting systems , namely @xmath636 .
hence , the right - hand side of the local - moment sum - rule eq.([susspinsum ] ) is contained in the interval @xmath637 .$ ] any theory that prevents the right - hand side of the local - moment sum rule from taking infinite values satisfies the mermin - wagner theorem . proof : : : near a magnetic phase transition , the zero matsubara - frequency component of the spin susceptibility takes the ornstein - zernicke form @xmath638 where @xmath333 is measured with respect to the ordering wave vector @xmath301 and where @xmath639 is the square of the correlation length . near its maximum , the above susceptibility is of order @xmath639 while all finite matsubara - frequency components at the ordering wave vector are at most of order @xmath640 which is much smaller than @xmath639 .
hence , one can keep only the zero - matsubara frequency contribution on the left - hand side of the local - moment sum rule eq.([susspinsum ] ) obtaining @xmath641 where @xmath127 contains non - zero matsubara frequency contributions as well as @xmath642 since @xmath643 is finite , this means that in two dimensions @xmath644 , it is impossible to have @xmath645 on the left - hand side otherwise the integral would diverge logarithmically .
finally , the @xmath213-sum rule on spin and charge susceptibilities follows as usual from the fact that the hamiltonian conserves particle number .
\right\rangle \right| _ { \tau = 0}$ ] and @xmath647 \right\rangle \right| _ { \tau = 0}$]one obtains for either charge or spin @xmath648 as can be seen from the spectral representations of spin and charge susceptibilities , eq.([spectch ] ) , the quantity that obeys the @xmath198sum rule is the coefficient of the leading term in the @xmath77 high - frequency expansion of the susceptibilities .
the single - particle energies @xmath649 entering explicitly the right - hand side of the @xmath198sum rule are independent of interactions , so interactions influence the @xmath198sum rule only very weakly through the @xmath650 .
in fact , in a continuum @xmath651 so @xmath617 enters only in the form @xmath652 . in this case , the right - hand side of the @xmath198sum rule is proportional to @xmath653 and hence is independent of interactions . on a lattice however ,
the energies can not in general be taken out of the sum and interactions influence the value of the right - hand side , but only through the fact that @xmath617 differs from the non - interacting fermi function @xmath654 . at strong - coupling ,
where the self - energy is strongly frequency dependent , this difference between @xmath617 and @xmath654 becomes important .
but from weak to intermediate coupling , calculations where @xmath654 appears on the right - hand side should be good approximations . in the explicit examples that we have treated ,
the @xmath4 dependence of the @xmath198sum rule becomes important only close to half - filling and for @xmath655 , signaling the breakdown of approximations based on frequency - independent self - energies .
while rpa - like theories that use @xmath654 instead of @xmath656 violate only weakly the @xmath198 sum rule in the weak to intermediate coupling regime , self - consistent theories that use frequency - dependent self - energies but no frequency - dependent vertices violate conservations laws in general , and the @xmath198 sum rule in particular , in a much more dramatic way .
the point is that susceptibilities with a dressed bubble , @xmath657 are bad approximations because they have the following properties , for any value of @xmath4 @xmath658 @xmath659 the first of these equations explicitly violates the ward identity , eq.([ward ] ) below , at all frequencies , including small non - zero ones , since at zero wave vector we should have @xmath660 for _ all _ frequencies except zero .
the second equation , eq.([ftriste ] ) violates the @xmath213-sum rule eq.([f2 ] ) at _ all wave _ vectors _ , _ by a constant term @xmath661 which in practical calculations , say at @xmath220 , is of the same order as the first term , which is the only one that should be there according to the @xmath213-sum rule .
proof : : : eqs.([wardtriste ] ) and ( [ ftriste ] ) are proven as follows .
consider the standard rpa expression but with dressed bubbles @xmath662 @xmath663 using the spectral representation for the green s function and inversion symmetry in the brillouin zone one finds , @xmath664 when the bubble is not dressed , the spectral weights are delta functions so that at @xmath437 the susceptibility would vanish for all non - zero values of @xmath665 , as required by the ward identity .
however , here because the spectral weight has a width and because the integrand is even and positive , then the integral will not vanish , resulting in the first anomaly eq.([wardtriste ] ) we mention . to prove the second equation , eq.([ftriste ] )
, it suffices to remember from the spectral representation of the susceptibility eq.([spectch ] ) and the derivation of the @xmath198sum rule eqs.([f2 ] ) that we are looking for the coefficient of the @xmath77 term in the high - frequency expansion .
given the rpa form , eq.([rpapseudo ] ) , only the numerator contributes to this limit .
one obtains , for the coefficient of the @xmath77 term , @xmath666 from which eq.([ftriste ] ) follows using the sum rules for occupation number eq.([occupation ] ) and for energy ( [ energie ] ) .
conservation laws have general consequences not only on equal - time correlation functions , as in the @xmath198sum rule above , but also on time - dependent correlation functions .
for example , from the heisenberg equations of motion and anti - commutation relations , follow the ward identities@xcite @xmath667 @xmath668 where @xmath669 for charge , and @xmath670 for spin .
the @xmath198sum rule above , eq.([f2 ] ) , follows from the above identity by simply taking @xmath671 , summing over @xmath672 and subtracting the two results for @xmath673 and @xmath674 .
we have seen in this section that there are strong cancelations for two - particle properties between the frequency dependence of self - energy and that of the vertex corrections , so that putting a frequency dependence in only one of them is a bad approximation .
we have adopted the kadanoff - baym formalism in the main text since it can be used as a guide to make approximations that satisfy conservation laws . when perturbation theory converges ( no phase transition ) then at zero temperature @xmath25 the imaginary part of the self - energy vanishes , @xmath675 , for all @xmath252 values and the fermi surface defined by @xmath676 encloses a volume that is equal to the volume enclosed by non - interacting particles @xmath677 this is the content of luttinger s theorem.@xcite@xcite .
it implies that there is a strong cancelation between the change of the chemical potential and the change of the self - energy on the fermi surface . in particular ,
when @xmath678 does not depend on @xmath360 or on the direction of @xmath463 ( infinite @xmath679 hubbard model , electron gas ) the change in @xmath680 is exactly canceled by @xmath678 @xmath681 luttinger s theorem is satisfied when @xmath682 any theory that calculates its self - energy from a functional derivative of the luttinger - ward functional @xmath683/\delta g$ ] will satisfy luttinger s theorem@xcite@xcite .
the latter procedure requires self - consistent determination of the self - energy as a function of momentum and frequency @xmath486 and is usually quite computationally involved .
however , even when this procedure to calculate the self - energy is not followed , it turns out to be rather easy to satisfy this theorem to an excellent degree of approximation in the weak to intermediate coupling regime .
the reason for this is that any frequency - independent self - energy will preserve luttinger s theorem and weak frequency dependence will not cause great harm . for the electron gas ,
luttinger@xcite suggests a way to build a perturbation theory in terms of non - interacting green s functions which allows to satisfy luttinger s theorem to very good accuracy .
the trick is that the chemical potential for the interacting electrons @xmath155 should always enter the calculations in the form of the difference with the shift of the self - energy on the fermi surface @xmath684 $ ] .
the `` non - interacting '' green s function @xmath685 in this formalism is the green s function of some effective non - interacting system and , in general , it is different from both @xmath686 and @xmath687 .
however , when @xmath688 luttinger s theorem requires that @xmath689 and one can approximate @xmath685 by the green s function for a non - interacting system of the same density @xmath690 . in practice , one can also have a phase transition ( or crossover ) at a finite temperature @xmath28 ( @xmath132 ) . in these cases
luttinger s theorem is satisfied only approximately since the zero - temperature limit can not be reached without a breakdown of perturbation theory .
then the relevant question is how well it is satisfied at @xmath28 ( @xmath132 ) .
( see also sec.([single - particle ] ) for a discussion of luttinger s theorem in our approach ) when luttinger s theorem holds , one can usually develop a landau fermi liquid theory . in this approach , the pauli principle is implemented only for momentum states near the fermi surface by imposing the forward scattering sum rule .
this sum rule , _ in two dimensions _ , reads @xmath691 = 0 \label{forward}\ ] ] where @xmath692 and @xmath693 are the symmetric and antisymmetric landau parameters expanded on the @xmath694 basis instead of the legendre polynomial basis .
recent renormalization group analysis has however claimed@xcite that the forward scattering sum rule comes from an inaccurate use of crossing symmetry and is not the proper way to enforce the pauli principle .
most approaches to the many - body problem disregard this sum rule anyway , in the same way that they disregard the local pauli principle .
in this appendix , we give the proofs of various relations mentioned in secs.([secformal ] ) and ( [ subsecaccuracy ] ) . 1 . the general expression for the self - energy eq.([2a ] ) can be obtained as follows .
use the equations of motion and the definition of the self - energy eqs.([dyson1])([dyson2 ] ) which in the present notation give @xmath695 \right\rangle\ ] ] @xmath696\ ] ] substituting the equation for the three - point susceptibility ( collective modes ) eq.([1a ] ) in this last equation and multiplying on both sides by @xmath697 proves@xcite the expression eq.([2a ] ) for the self - energy .
we now show that our approach satisfies the consistency requirement between single - particle properties and collective modes in the form of eq.([sums ] ) . using our expression eq.([param ] ) for @xmath698 and the definition of @xmath699 eq.([chi0 ] ) we obtain @xmath700 \frac{\chi _ 0\left ( q\right ) } 2\ ] ] using @xmath701 @xmath702 and the local moment eq.([spin ] ) and local charge eq.([charge ] ) sum rules proves the result .
the result is also obvious if we follow the steps in the first part of this appendix to deduce the self - energy expression eq.([2 ] ) using the collective mode equation eq.([1 ] ) adapted to our approximation .
using the present notation and formalism , we now give a physical derivation of eq.([usp ] ) that is equivalent to the one already given using the equations of motion approach@xcite .
( the latter derivation was inspired by the local field approximation of singwi _
et al._@xcite ) .
since our considerations on collective modes are independent of the precise value of the interaction @xmath4 , we do have to use the equations of motion , or the equivalent , to feed that information back in the definition of irreducible vertices .
the two irreducible vertices that we need are in principle calculable from @xmath704 } { % \delta g_{\sigma ^{\prime } } \left ( 3,4\right ) } \ ] ] the rewriting on the right - hand side has been done to take advantage of the fact that in the hubbard model , the equations of motion ( see eqs.([dyson1])([dyson2 ] ) ) give us the product @xmath705 as the highly local four field correlation function @xmath706 \right\rangle .$ ] ordinary rpa amounts to a hartree - fock factoring of this correlation function .
pursuing the philosophy that the minimum number of approximations should be done on local correlation functions , we do this factoring in such a way that it becomes exact when all points are identical , namely when @xmath707 . in other words , we write @xmath708 \right\rangle \sim u\frac{\langle n_{\uparrow } \left ( 1\right ) n_{\downarrow } \left ( 1\right ) \rangle } { % \langle n_{\uparrow } \left ( 1\right ) \rangle \left\langle n_{\downarrow } \left ( 1\right ) \right\rangle } g_{-\sigma } \left ( 1,1^{+}\right ) g_\sigma \left ( 1,\overline{2}\right)\ ] ] all quantities are evaluated as functionals of @xmath88 up to this point .
we can now evaluate the functional derivative @xmath709 } { \delta g_{\sigma ^{\prime } } \left ( 3,4\right ) } \ ] ] @xmath710 } { \delta g_{\sigma ^{\prime } } \left ( 3,4\right ) } g_{-\sigma } \left ( 1,1^{+}\right ) \delta \left ( 1 - 2\right ) + u\frac{\langle n_{\uparrow } \left ( 1\right ) n_{\downarrow } \left ( 1\right ) \rangle } { \langle n_{\uparrow } \left ( 1\right ) \rangle \left\langle n_{\downarrow } \left ( 1\right ) \right\rangle } \frac{\delta g_{-\sigma } \left ( 1,1^{+}\right ) } { \delta g_{\sigma ^{\prime } } \left ( 3,4\right ) } \delta \left ( 1 - 2\right)\ ] ] the functional derivatives are now evaluated for the actual equilibrium value of @xmath711 hence , we can use rotational invariance to argue that the first term is independent of @xmath31 and @xmath712 whereas the last one is proportional to @xmath713 since @xmath714 , only this last term proportional to @xmath715 contributes to @xmath6 . to obtain this term
, it suffices to note that @xmath716 and we obtain the desired result eq.([usp ] ) for @xmath6 .
it is instructive to recover the two - dimensional result for precursors of antiferromagnetic bands using the real - frequency formalism since it also clarifies the limit in which the fermi liquid result is recovered .
again we neglect the contribution of charge fluctuations . starting from our expression for the self - energy eq.([param ] ) , one uses the spectral representation for the susceptibility and for @xmath12 .
the matsubara frequency sums can be then done and the result is trivially continued to real frequencies.@xcite one obtains , for the contribution of classical and quantum spin fluctuations to the self - energy in @xmath40 dimensions @xmath717 \frac{\chi _ { sp}^{\prime \prime } \left ( { \bf q},\omega ^{\prime } \right ) } { % \omega + i\eta + \omega ^{\prime } -\left ( \varepsilon _ { { \bf k}+{\bf q}}-\mu _ 0\right ) } \label{sigmareel}\ ] ] where @xmath718 at half - filling in the nearest - neighbor model and where @xmath213 is , as usual , the fermi function , while @xmath719 is the bose - einstein distribution . to analyze this result in various limiting cases we need to know more about the frequency dependence of the spin susceptibility .
when the antiferromagnetic correlation length is large , the zero - frequency result eq.([chias ] ) mentioned above can be generalized to @xmath720 \label{chirpa}\ ] ] where , @xmath721 is the characteristic spin relaxation frequency . in the notation of ref.@xcite ,
the microscopic diffusion constant @xmath679 is defined by @xmath722 with the microscopic relaxation time , @xmath723 this relaxation - time is non - zero in models where the fermi surface intersects the magnetic brillouin zone . clearly , the frequency dependence of @xmath724 is on a scale @xmath725 the @xmath726 decrease of @xmath727 at high - frequency is not enough to ensure that the real frequency version of the local - moment sum rule is satisfied and the simplest way to cure this problem is to introduce@xcite a high - frequency cutoff @xmath728 .
the large correlation length makes the characteristic energy of the spin fluctuations @xmath729 a small number ( critical slowing down ) .
we consider in turn two limiting cases.@xcite the fermi - liquid regime appears for @xmath730 and the non - fermi liquid regime in the opposite ( renormalized classical ) regime @xmath731 perhaps the best known characteristic of a fermi liquid is that @xmath732 and @xmath733 . to recover this result in the regime
@xmath730 far from phase transitions , we start from the above expression eq.([sigmareel ] ) for the self - energy to obtain @xmath734 \chi _ { sp}^{\prime \prime } \left ( q_{\perp } , q_{\vert } \left ( q_{\perp } , { \bf k}_f,\omega , \omega ^{\prime } \right ) ; \omega ^{\prime } \right ) \label{imsigmareel}\ ] ] where @xmath735 , the component of @xmath410 parallel to the fermi momentum @xmath463 , is obtained from the solution of the equation @xmath736 the key to understanding the fermi liquid _ vs _
non - fermi liquid regime is in the relative width in frequency of @xmath737 _ vs _ the width of the combined bose and fermi functions . in general
, the function @xmath738 depends on @xmath739 on a scale @xmath740 while far from a phase transition , the _ explicit _ frequency dependence of @xmath741 is on a scale @xmath742 .
hence , in this case we can assume that @xmath743 is a constant in the frequency range over which @xmath744 differs from zero .
also , since @xmath745 depends on wave vector @xmath410 over a scale of order @xmath746 one can neglect the @xmath747 dependence of @xmath735 obtained from eq.([condition_q_par ] ) .
hence , we can approximate our expression eq.([imsigmareel ] ) for @xmath748 by @xmath749 \omega ^{\prime } = -\frac{uu_{sp}}4\frac{a\left ( { \bf k}% _ f\right ) } { 4v_f}\left [ \omega ^2+\left ( \pi t\right ) ^2\right ] \label{imsigmafermiliquid}\ ] ] where the substitution @xmath750allowed the integral to be done exactly and where @xmath751 in general , @xmath752 depends on the orientation of the fermi wave vector , @xmath753 , because it determines the choice of parallel and perpendicular axis @xmath754 .
the above result eq.([imsigmafermiliquid ] ) for @xmath748 is the well known fermi liquid result .
there are known corrections to the fermi liquid self - energy that come from the non - analytic @xmath755 behavior of @xmath745 near the ferromagnetic ( zone center ) wave vector . in three
dimensions@xcite this non - analyticity leads to subdominant @xmath756 corrections , while in two dimensions it leads to the dominant @xmath757 behavior.@xcite@xcite in the case under consideration , the antiferromagnetic contribution has a larger prefactor .
even when it dominates however , it can also lead to non - analyticities in the case of a nested fermi surface .
indeed , we note that @xmath758 in two dimensions , the logarithmic divergence of the density of states @xmath759 at the van hove singularity makes the zero - frequency limit of the microscopic relaxation time eq.([gamma0 ] ) ill - defined , because of the logarithmic divergence at @xmath259 . however
, this leads only to logarithmic corrections .
if we drop logarithmic dependencies then for @xmath760 one has @xmath761 and this @xmath762 dependence of @xmath763 changes the temperature dependence of @xmath764 from @xmath261 to @xmath174 as discussed in the `` nested fermi liquid '' approach.@xcite near an antiferromagnetic phase transition , the spin - fluctuation energy becomes much smaller than temperature .
this is the renormalized classical regime .
the condition @xmath765 means that @xmath766 is peaked over a frequency interval @xmath767 much narrower than the interval @xmath768 over which @xmath769 changes .
this situation is the opposite of that encountered in the fermi liquid regime . to evaluate @xmath770 eq.([imsigmareel ] ) the fermi factor
can now be neglected compared with the classical limit of the bose factor , @xmath771 .
then the dominant contribution to @xmath772 is from classical spin fluctuations @xmath773 as we see below .
more specifically , we take into account that the integral is peaked near @xmath774 and measure wave vector with respect to the zone center . for simplicity
we consider below the half - filled case @xmath775 .
then , with the help of @xmath776 we approximate the equation for @xmath735 eq.([condition_q_par ] ) by @xmath777 this gives us for eq.([imsigmareel ] ) the approximation @xmath778 the dependence of @xmath727 on @xmath739 through @xmath779 may be neglected because @xmath735 appears only in the combination @xmath780 and in the regime @xmath781 we have @xmath782 the latter inequality is generically satisfied when @xmath783 .
using @xmath784 the above equation eq.([sigmareelrc ] ) for @xmath772 reduces precisely to the classical contribution found using imaginary - time formalism eq.([sigm ] ) . as we saw in sec.([secrc ] ) , when the condition @xmath341 is satisfied , then this contribution is dominant and leads to @xmath785
this appendix expands on sec.([seccomparisons ] ) to discuss in detail various theories , explaining the advantages and disadvantages of each in the context of the sets of constraints described in appendices ( [ subsecconstsingle ] ) and ( [ secconsttwopart ] ) . in standard paramagnon theories,@xcite@xcite
the spin and charge fluctuations are computed by rpa , using either bare or dressed green s functions .
then the fluctuations are fed back in the self - energy .
in fact there is a whole variety of paramagnon theories .
they are largely phenomenological .
the reader is referred to ref.@xcite for a review .
we concentrate our discussion on recent versions@xcite of the so - called berk - schrieffer formula@xcite . in this approach ,
infinite subsets of diagrams are summed and bare propagators are used in the calculation of both the susceptibilities and the self - energy , the latter being given by @xmath786 g_\sigma ^0(k+q ) .
\label{berks}\ ] ] the rpa spin and charge susceptibilities have been defined in eqs.([rpasp])([rpach ] ) . comparing with our self - energy formula eq.([param ] ) , it is clear that here there is no vertex correction .
in addition , the factor of three in front of the spin susceptibility in eq.([berks ] ) is supposed to take into account the presence of both longitudinal and transverse spin waves and the subtracted term is to avoid double - counting the term of order @xmath787 .
we can now see the advantages and disadvantages of this approach .
first , note that the susceptibilities entering the berk - schrieffer formula are the rpa ones . as we saw in appendix([secsumrules ] ) , these fail to satisfy both the local pauli principle and the mermin - wagner theorem .
hence , spurious phase transitions will influence the self - energy in uncontrollable ways .
the collective modes do however satisfy conservation laws since they are obtained with bare vertices and green s functions containing a constant self - energy .
the @xmath213-sum rule eqs.([f2 ] ) then is satisfied without renormalization of the distribution function @xmath788 because the zeroth order self - energy is constant .
this is all in agreement with the definition of a conserving approximation for the collective modes .
the high - energy asymptotics of the self - energy sets in at the correct energy scale @xmath789 in this approach , but the second term of the large - frequency asymptotics is incorrect . indeed , at large values of @xmath790 , @xmath791 + \ldots\ ] ] and the sums can be evaluated as follows using the fluctuation - dissipation theorem @xmath792 @xmath793 @xmath794 the correlators on the right - hand side take their rpa value so they do not satisfy the pauli principle , _
i.e. _ @xmath795 taking these results together we have @xmath796 + \ldots\ ] ] this does not gives the correct asymptotic behavior eq.([selfhaut ] ) even if the pauli principle @xmath797 were satisfied , because @xmath227 depends on the interaction @xmath4 . the paramagnon self - energy eq.([berks ] ) also does not satisfy the consistency requirement eq.([sumsigma1 ] ) between self - energy and collective modes imposed by the equations of motion . to see this
we first note that @xmath798 \chi _ 0\left ( q\right ) \label{sumsberk}\ ] ] using this expression in the sum - rule eq.([sumsigma1 ] ) which relates one and two - particle correlators and expanding both sides of this sum - rule in powers of @xmath4 , one finds that it is satisfied only up to order @xmath787 . on the other hand , if one replaces @xmath799 in eq.([berks ] ) by @xmath800 , the sum - rule eq.([sumsigma1 ] ) is satisfied to all orders in @xmath4 . in our opinion ,
the problem of enforcing rotational invariance in approximate theories is highly non - trivial and can not be solved simply by adding factor of @xmath180 in front of @xmath182 and then subtracting @xmath801 to avoid double counting . for more detailed discussions
see ref.@xcite and the comments at the end of sec.([single - particle ] ) .
luttinger s theorem is trivially satisfied if the occupation number is calculated with the initial constant self - energy since it gets absorbed in the chemical potential .
if the occupation number is calculated with the green s function that contains the berk - schrieffer self - energy then luttinger s theorem is in general violated .
it is advisable to use a new chemical potential . in the conserving approximation
schemes@xcite , one takes any physically motivated subset of skeleton diagrams to define a luttinger - ward functional @xmath802 .
skeleton diagrams contain fully dressed green s functions and no self - energy insertions .
this functional is functionally differentiated to generate a self - energy that is then calculated self - consistently since it appears implicitly in the green s functions used in the original set of diagrams .
a further functional differentiation allows one to calculate the irreducible vertices necessary to obtain the collective modes in a way that preserves ward identities .
if one uses for the free energy the formula @xmath803
+ \text{tr}\left ( \sigma g\right ) -\phi \label{zbaym}\ ] ] then one obtains thermodynamic consistency in the sense that thermodynamic quantities obtained by derivatives of the free energy are identical to quantities computed directly from the single - particle green s function . for example
, particle number can be obtained either from a trace of the green s function or from a chemical potential derivative of the free energy . in this scheme ,
luttinger s theorem is satisfied as long as perturbation theory converges since then any initial guess for the luttinger - ward functional will satisfy luttinger s theorem .
flex refers to a particular choice of diagrams for @xmath804 this choice leads to the following self - consistent expression for the self - energy @xmath805 g_\sigma ( k+q ) .
\label{berksfl}\ ] ] this expression for the self - energy does not contain vertex corrections , despite the fact that , contrary to the electron - phonon case , migdal s theorem does not apply here .
we have explained in detail in sec.([subsecwhyflex ] ) why this may lead to qualitatively wrong results , such as the absence of precursors of antiferromagnetic bands and of the pseudogap in @xmath806 in two dimensions .
another drawback of this approach is that it does not satisfy the pauli principle in any form , either local or through crossing symmetry@xcite .
indeed , one would need to include all exchange diagrams to satisfy it . in practice
this is never done . in the same way that there is nothing to constrain the value of @xmath807 obtained by the fluctuation - dissipation theorem to be equal to @xmath808
, there is nothing to explicitly constrain the value of @xmath809 .
nevertheless , the mermin - wagner theorem is believed to be satisfied in flex because the feedback through the self - energy tends to prevent the divergence of fluctuations in low dimension.@xcite,@xcite .
physically however , this seems to be an artificial way of satisfying the mermin - wagner theorem since this theorem should be valid even in localized spin systems where single - particle properties are negligibly influenced by thermal fluctuations .
we also point out that the proof of the mermin - wagner theorem in @xmath18 models implies that the finite temperature phase transition in two dimensions is not simply removed by thermal fluctuations , but that it is replaced by a crossover to the renormalized classical regime with exponentially growing susceptibility .
the fact that the conserving susceptibility in flex does not show such behavior@xcite means that flex is actually inconsistent with the generic phase space arguments responsible for the absence of finite - temperature phase transition in two dimensions .
the case of one dimension also suggests that collective modes by themselves should suffice to guarantee the mermin - wagner theorem without feedback on single - particle properties .
indeed , in one dimension one shows by diagrammatic methods ( parquet summation or renormalization ) that the zero - temperature phase transition is prohibited at the two - particle level even _ without _ self - energy effects@xcite .
although , the second - order diagram is included correctly in flex , it does not have the correct coefficient in the @xmath568 expansion of the self - energy .
more importantly , the high - frequency behavior sets - in too late to give the hubbard bands , as we have explained in sec.([subsecwhyflex ] ) .
we have also seen a case where flex , as judged from comparisons with monte carlo simulations ( fig.(1a ) of ref.@xcite ) , does not reproduce the results of second - order perturbation theory even when it is a good low - energy approximation .
one of the inconsistencies of conserving approximations that is seldom realized , is that the self - energy is inconsistent with the collective modes . in other words , the consistency formula eq.([consistency ] ) is not satisfied in the following sense .
the explicit calculation of @xmath810 leads to an estimate of _
_ @xmath17 that differs from the one obtained by applying the fluctuation - dissipation theorem to the _ conserving _ spin and charge susceptibilities . in the parquet approach ,
one enforces complete antisymmetry of the four point function by writing down fully crossing - symmetric equations for these .
there are three irreducible vertices , namely one for the particle - particle channel , and one for each of the two particle - hole channels .
they obey the so - called parquet equations.@xcite the green s functions are dressed by a self - energy which itself contains the four point function . in this way , self - consistency between one - particle and two - particle quantities is built - in .
solutions are possible for the one - impurity problem@xcite and in one - dimension@xcite .
however , to solve the parquet equations in higher dimension with presently available computing power is impossible
. bickers _
et al._@xcite@xcite have formulated the parquet equations as a systematic improvement over flex and have devised a way to do practical calculations by introducing so - called pseudo - potentials .
since the main computational difficulty is in keeping the full momentum and frequency dependence of the four point functions entering the calculation of the self - energy , this is where the various fluctuations channels are approximated by rpa - like forms eq.([rpapseudo ] ) but with fully dressed propagators and an effective interaction ( pseudo - potential ) instead of @xmath4 .
a different strategy is under development.@xcite the criticism of the present section applies only to the current pseudo - potential parquet approach.@xcite@xcite it can be seen that one drawback of this approach at the physical level is that the use of constant effective interactions with dressed single - particle propagators means that the fluctuations used in the calculation of the self - energy do not satisfy conservation laws , as we just demonstrated in sec([secconsttwopart ] ) .
furthermore , the pseudopotentials are determined by asking that the susceptibilities extracted from the four - point functions in the parquet equations match the corresponding rpa - pseudo - potential susceptibility at only one wave vector and frequency .
the choice of this matching point is arbitrary : should the match be done for the typical , the average , or the maximal value of the susceptibility in the brillouin zone ? as we have seen in sec.([secflex ] ) , even if the expression for the self - energy in this approach explicitly has the second - order perturbation theory diagram in it , this is not sufficient to ensure that the correct high frequency asymptotic behavior starts at the appropriate frequency scale @xmath811 .
nevertheless , in many cases the results of the calculations performed with this approach are not so different from second - order perturbation theory , as can be seen from fig.(1 ) of ref.(@xcite ) .
going rapidly through the rest of our list of properties , we see that the consistency requirement @xmath812 is at least approximately built - in by construction . concerning the local - moment sum - rule and the mermin - wagner theorem ,
it has been shown that the so - called `` basic '' parquet equations should have the same critical behavior as the leading term in the @xmath813 expansion,@xcite and hence should satisfy the mermin - wagner theorem.@xcite the pseudo - potentials should not affect the self - consistency necessary to satisfy the mermin - wagner theorem but the fact that they are matched at a single point might introduce difficulties , especially if the wave - vector at which @xmath182 becomes unstable is unknown from the start .
as far as the pauli principle is concerned , it should be at least approximately satisfied both locally and in momentum space .
nothing however in the approach enforces conservation laws .
if we concentrate on the @xmath437 properties , our spin and charge correlations behave as a special case of the `` local fermi liquid '' defined in ref.@xcite .
a `` local fermi liquid '' is a description of @xmath814 properties that applies when the self - energy , and consequently irreducible vertices , depend only on frequency , not on momentum . in a local fermi liquid
there are only two landau parameters , which in our case are @xmath815 and @xmath816 unitarity and the forward scattering sum rule , if valid , imply that there is no ferromagnetism in the repulsive case@xcite , as we have found .
one can check explicitly that the forward scattering sum rule is satisfied to within about @xmath201 in our usual monte carlo parameter range .
however , as discussed in appendix ( [ lat ] ) , the forward scattering sum - rule refers only to wave vectors on the fermi surface , not to the local version of the pauli principle .
furthermore , the validity of this sum rule has been questioned.@xcite the effective mass at this level of approximation is the bare one , as in a transitionally invariant _ local _ fermi liquid.@xcite recall however that our microscopic calculations are not phenomenological : they explicitly give a value for the landau parameters .
also , our results extend well beyond the @xmath437 quantities usually considered in fermi liquid theory .
the quasi - particle weight @xmath81 calculated with @xmath817 can differ substantially from the initial one .
this means that if we were to calculate the susceptibility with the corresponding frequency and momentum dependent irreducible vertices @xmath818 there would be sizeable compensation between vertices and self - energy because our calculations with @xmath819 ( @xmath820 ) and constant renormalized vertices already gave excellent agreement with monte carlo simulations . finally , consider the high - frequency asymptotics .
since we use bare propagators , the high - frequency asymptotics comes in at the appropriate frequency scale , namely @xmath567 and the hubbard bands do exist in our theory . however , the coefficient of proportionality in front of the asymptotic form @xmath568 is incorrect . using eqs.([param]),([susspinsum]),([suschargesum ] )
we can write the high - frequency asymptotics in the following form @xmath821 + \ldots \label{og}\ ] ] this form is useful to understand what is necessary to obtain the quantitatively correct high - frequency behavior .
indeed , one would recover the exact result eq.([selfhaut ] ) , if one were to take into account that : i ) the irreducible vertices become equal to the bare one @xmath4 at high - frequencies ; ii ) the local pauli principle @xmath822 is satisfied .
contrary to most other approaches , our theory does satisfy the local pauli principle eq.([pauli ] ) exactly .
however , since our irreducible vertices are constant and tuned to describe the low energy physics , we violate the first of the above requirements .
it is thus clear that for a correct quantitative description of both the low energy physics and the hubbard bands one needs to work with frequency - dependent irreducible vertices . for reviews of slave - boson approaches applied to the hubbard model ,
see for example , g. kotliar in _ correlated electrons systems , _ jerusalem winter school for theoretical physics , vol.9 , ed .
emery ( world scientific , singapore , 1993 ) , p.118 and g. kotliar in _ strongly interacting fermions and high _
@xmath28 s_uperconductivity , _ les houches , session lvi , 1991 , eds . b. douot and j. zinn - justin ( elsevier , amsterdam , 1995 ) , p.197 . for a review , see k. s. singwi and m. p. tosi , in _ solid state physics _ , edited by h. ehrenreich , f. seitz , and d. turnbull ( academic , new york , 1981 ) , vol .
* 36 * , p. 177 ; s. ichimaru , rev . mod . phys . * 54 * , 1017 ( 1982 ) . c. bourbonnais and l.g .
caron , int .
b * 5 * , 1033 ( 1991 ) ; c. bourbonnais and l.g .
caron , europhys .
* 5 , * 209 ( 1988 ) ; physica b * 143 , * 451 ( 1986 ) . c. bourbonnais , in _ strongly interacting fermions and high _
@xmath28 s_uperconductivity , _ les houches , session lvi , 1991 , eds .
b. douot and j. zinn - justin ( elsevier , amsterdam , 1995 ) , p.307 .
t. hotta and s. fujimoto , phys .
b * 54 * , 5381 ( 1996 ) have checked within perturbation theory that this decrease of charge fluctuations as one approaches half - filling at fixed temperature can indeed be explained only if vertex corrections are included .
a similar analysis of monte carlo data has been used earlier for the negative @xmath4 hubbard model by mohit randeria , nandini trivedi , adriana moreo and richard t. scalettar , phys .
lett . * 69 * , 2001 ( 1992 ) .
r. frsard and p. wlfle .
matt . * 4 * , 3625 ( 1992 ) .
b. mller , k. doll and r. frsard .
matt . * 5 * , 4847 ( 1993 ) . for
the phase diagram predicted by the gutzwiller variational approach , see m. dzierzawa et r. frsard , z. phys .
matt . * 91 * , 245 ( 1993 ) .
hideki matsumoto and ferdinando mancini , phys .
b * 55 * , 2095 ( 1997 ) . in this work , the crossover to the renormalized classical regime is not clearly identified with the appearance of antiferromagnetic precursor bands .
nevertheless , the new features that appear at low temperature , in addition to hubbard bands , are associated with antiferromagnetic correlations .
we thank h. matsumoto for correspondance on this subject .
blaizot and e. iancu , phys .
* 76 * , 3080 ( 1996 ) .
the problem of doing correctly the many - body problem below @xmath132 is similar to the one encountered in hot three - dimensional quantum electrodynamic plasmas .
there , the photons play the role of fluctuations at their critical point .
the divergences are logarithmic in frequency .
furthermore , since the photons can be taken into account order by order in a systematic manner , an infinite resummation is possible . in our case ,
the spin propagator is not really a different physical entity and , as we saw , it is not clear how to implement a systematic diagrammatic approach . for more details ,
blaizot , e. iancu , phys .
d * 55 , * 973 ( 1997 )
. in - plane infrared conductivity measurements in @xmath824 suggest a pseudogap both above and below optimal doping .
t. startseva , t. timusk , a.v .
puchkov , d.n .
basov , h.a .
mook , t. kimura , k.kishio , cond - mat/9706145 .
once the expression for the self - energy has been written in the form eq.([selfhaut ] ) , it can be interpreted as the leading self - energy of an electron interacting with a bosonic excitation .
numerous studies of this kind of problem have appeared , including some where the bosonic excitation is softening .
for a recent example , see m. crisan and l. tataru , j. of superconductivity , * 8 * , 341 ( 1995 ) ; m. crisan and l. tataru , phys .
b * 54 * , 3597 ( 1996 )
. g. baym and c. pethick , _ landau fermi liquid theory , concepts and applications , _ ( wiley , new york , 1991 ) . for a microscopic calculation ,
see : d.j .
amit , j.w .
kane , and h. wagner , phys .
* 19 * , 425 ( 1967 ) and phys .
rev . * 175 * , 313 ( 1968 ) .
n. e. bickers and d. j. scalapino , phys .
b * 46 * , 8050 ( 1992 ) .
the proof that parquet equations are equivalent to the leading order of the @xmath813 expansion is valid when there is a mode diverging in a single channel.@xcite it is known that in one dimension parquet equations are much more general than the @xmath813 expansion because they contain interference terms between channels that are absent in the classical @xmath813 expansion .
these interference terms are the cause of the vanishing of the transition temperature in one dimension . | a new approach to the single - band hubbard model is described in the general context of many - body theories .
it is based on enforcing conservation laws , the pauli principle and a number of crucial sum - rules .
more specifically , spin and charge susceptibilities are expressed , in a conserving approximation , as a function of two irreducible vertices whose values are found by imposing the local pauli principle @xmath0 as well as the local - moment sum - rule and consistency with the equations of motion in a local - field approximation .
the mermin - wagner theorem in two dimensions is automatically satisfied .
the effect of collective modes on single - particle properties is then obtained by a paramagnon - like formula that is consistent with the two - particle properties in the sense that the potential energy obtained from @xmath1 is identical to that obtained using the fluctuation - dissipation theorem for susceptibilities . since there is no migdal theorem controlling the effect of spin and charge fluctuations on the self - energy , the required vertex corrections
are included .
it is shown that the theory is in quantitative agreement with monte carlo simulations for both single - particle and two - particle properties .
the theory predicts a magnetic phase diagram where magnetic order persists away from half - filling but where ferromagnetism is completely suppressed .
both quantum - critical and renormalized - classical behavior can occur in certain parameter ranges .
it is shown that in the renormalized classical regime , spin fluctuations lead to precursors of antiferromagnetic bands ( shadow bands ) and to the destruction of the fermi - liquid quasiparticles in a wide temperature range above the zero - temperature phase transition .
the upper critical dimension for this phenomenon is three .
the analogous phenomenon of pairing pseudogap can occur in the attractive model in two dimensions when the pairing fluctuations become critical .
simple analytical expressions for the self - energy are derived in both the magnetic and pairing pseudogap regimes .
other approaches , such as paramagnon , self - consistent fluctuation exchange approximation ( flex ) , and pseudo - potential parquet approaches are critically compared . in particular , it is argued that the failure of the flex approximation to reproduce the pseudogap and the precursors afm bands in the weak coupling regime and the hubbard bands in the strong coupling regime is due to inconsistent treatment of vertex corrections in the expression for the self - energy . treating the spin fluctuations as if there was a migdal s theorem can lead not only to quantitatively wrong results but also to qualitatively wrong predictions , in particular with regard to the the single - particle pseudogap . |
below we report the results of an experiment designed to measure the value of @xmath13 using direct lagrangian particle tracking in an elastic turbulence flow in a microfluidic tube . however , our measurements of pair dispersion show no evidence for the predicted exponential growth in time of the mean squared separation . to our great surprise , we have discovered that for a significant fraction of the observation time , the mean relative pair dispersion evolves quadratically to leading order in time @xmath18 , where @xmath19 and @xmath20 .
furthermore , our data admits a scaling collapse , providing a convincing evidence that these observations are well - described by the short time expansion of the relative pair dispersion , the so called _ ballistic _ regime .
we have established an experimental database of trajectories derived from tracers in the flow ; see materials and methods .
one example of a pair realisation can be found in fig.[fig : intro]_a , b_. a sub - sample of pair separation @xmath21 curves is plotted in fig.[fig : intro]_c _ ; here @xmath19 and @xmath20 .
these curves , together with the rest of the pairs which were found in each @xmath22 bin , were collected and the second moments were calculated to construct the datasets of @xmath23 presented in fig.[fig : normed_pair_dispersion ] on a semi - logarithmic scale , where an exponential curve would have appeared linear ( the un - normalised data @xmath24 can be found in fig.[fig : not_normed_pair_dispersion ] ; the sample size data is presented in fig.[fig : sample_size ] ) .
( image ) at ( 0,0 ) ; ( 0.01,0.76 ) node a ; ( 0.01,0.28 ) node b ; ( 0.50,0.76 ) node c ; however , our data shows no supporting evidence for the predicted exponential growth of the second moment of pair separations . the insets present a zoom - in on the temporal sub - intervals where the full range plot may seem to contain linear segments
nevertheless , no single slope , to be identified with @xmath9 in eq.[eq : exp_prediction ] , can be found .
moreover , an exponential pair dispersion should extrapolate to the origin on this plot , and this is clearly not the case .
to our great surprise we have found that , for a significant fraction of the observation time , the mean relative pair dispersion evolves quadratically in time to leading order @xmath18 ; this observation is evident in the insets of fig.[fig : short_time_behaviour_forward ] , and stands in sharp contradiction to the interpretations of recent experimental results @xcite . to better understand the source for this scaling let us write the taylor expansion around @xmath25 @xmath26
substituting this in the expression for the relative pair dispersion and considering the ensemble average over pairs of the same initial separation @xmath27 we find that the leading order term at short times is indeed quadratic in @xmath28 the so - called _ ballistic _ regime .
this is a universal property which does not require any assumptions on the character of the flow .
a sign of this scaling beneath the dissipative scale has been observed in simulations of isotropic turbulence ( * ? ? ? * fig.5 ) . yet
neither quantitative analysis of the coefficients in eq.[eq : relative_dispersion_taylor ] nor any contrast with the exponential separation prediction were made .
other experimental @xcite and numerical @xcite results are limited to the inertial subrange of turbulence . to test this
further we rescale the relative pair dispersion by the pre - factor , the mean initial squared relative velocity @xmath29 . unlike the case of inertial turbulence , for elastic turbulence there are no exact results nor scaling arguments to derive the coefficients appearing in eq.[eq : relative_dispersion_taylor ] .
therefore we extract them from the experimental data ; the @xmath29 profile as function of @xmath22 is presented in fig.[fig : uo2_ro_profile]_a_. indeed we find that our data admits a scaling collapse with no fitting parameters , providing a convincing experimental evidence that these observations are well - described by the short time expansion of the relative pair dispersion .
the data shown in fig.[fig : short_time_behaviour_forward ] seem to deviate significantly from the quadratic scaling only after 23 seconds .
( image ) at ( 0,0 ) ; ( 0.01,0.9 ) node a ; ( image ) at ( 0,0 ) taylor_backwards_pair_dispersion_loglog ; ( 0.01,0.96 ) node b ; in order to expose the sub - leading contributions to the initial relative pair dispersion , we subtract the backwards - in - time dynamics from the forward one . this way the time - symmetric terms , even powers of @xmath28 ,
are eliminated .
the result , the time asymmetric contributions presented in fig.[fig : short_time_behaviour_backwards - forward ] , shows that indeed initially the next - to - leading order correction follows @xmath30 and that the curves do collapse onto one when rescaled by @xmath31 , the appropriate coefficient in eq.[eq : relative_dispersion_taylor ] .
the values of @xmath32 were , once again , extracted from the experimental data ; see fig.[fig : a.u_ro_profile ] for the @xmath22 profile . however , the deviations from this scaling are noticeable earlier than half a second , much earlier than the deviations from the ballistic behaviour discussed above .
this hints that the deviation observed in fig.[fig : short_time_behaviour_forward ] is in fact due to higher order terms , potentially an indication of a transition to another regime .
the fact that the transition takes place at an earlier time for the larger initial separations indicates the potential effects of the vessel size and its geometry , as well as the failure of the linear flow approximation ; see fig.[fig : uo2_ro_profile ] .
in the light of eq.[eq : relative_dispersion_taylor ] , the ballistic evolution at short times should not be too surprising . and
yet , to our knowledge it has not been discussed experimentally , nor confronted with the exponential separation prediction . moreover ,
the ` memory ' of the initial conditions in our data is longer than one may have expected . on the one hand ,
our observations are consistent with the time - scale @xmath33 derived from the ratio of the coefficients in eq.[eq : relative_dispersion_taylor ] ( see fig.[fig : dt_star_ro_profile ] ) , as deviations from the @xmath34 scaling are expected to be noticeable for @xmath35 .
this is also manifested in the rescaling of the data by @xmath33 , presented in fig.[fig : normalised_behaviour_forward ] , and to be compared with numerical simulations of inertial turbulence ( * ? ? ?
* fig.1 ) . on the other hand , during the time the statistics are dominated by the initial conditions , single realisations are not monotonic nor predictable ( see fig.[fig : intro ] ) , that is to say , our observations are not a trivial result of the mean flow alone , which in itself is inhomogeneous . moreover , during one second tracers typically travel distances larger than the width of the tube .
given the stochastic nature of the dynamics , the spatial structure of the flow and the limited range of the linear flow approximation ( see insets in fig.[fig : uo2_ro_profile ] ) , the taylor expansion in time eq.[eq : relative_dispersion_taylor ] was expected to have a rather restrictive radius of convergence .
further analysis is required to clarify whether higher order terms are of importance and to estimate the lagrangian correlation time , which accounts for both the temporal and the spatial structure of the flow .
hence , in - depth discussion of the transition away from the ballistic regime in our system is left for future work . nonetheless , the time for which the ballistic approximation holds is longer than a second , a memory of the initial relative velocities which allows a wide prediction horizon .
we are unaware of any previous discussion of pair dispersion short time statistics neither in the context of elastic turbulence nor for the broader class of wall - bounded viscous mixing flows .
the implications are far - reaching knowing the average velocity difference at a given scale , which is an eulerian property , one can determine the degree of dispersion , a lagrangian property , as was recently discussed in @xcite .
we have demonstrated its predictive power over a significant time interval , which provides a new paradigm for chaotic flows dominated by dissipation .
the exponential growth in eq.[eq : exp_prediction ] relies on two underlying assumptions @xcite : the velocity field admits a linear approximation in space throughout the observation time ; and the observation time is much longer than the correlation time of the velocity gradients .
these requirements are quite stringent and are not fulfilled by our experiment . to the best of our knowledge
these assumptions have not been met experimentally for tracer particles so far , and yet the exponential growth prediction seems to be the leading paradigm in interpreting experimental results @xcite .
our estimations for the experimental system presented here indicate these may be possibly relevant for @xmath36 and @xmath37 .
examining the experimental parameters reported in recent works @xcite we find that the trade - off between short correlation times and a large enough dissipation scale renders them difficult to reach in intense inertial turbulence , as batchelor himself anticipated @xcite ; also see @xcite .
the lesson in this case highlights the limitations of asymptotic analyses for the descriptions of real - world flows for practical purposes the asymptotic dynamics are not necessarily the generic description which should be considered but rather the short - time behaviour , here dictated by the eulerian averages at the initial time .
according to our findings the lyapunov exponents picture is not the appropriate one to describe the dynamics of wall - bounded elastic turbulence at scales larger than few percent of the vessel size .
this raises questions regarding the mechanism for polymer stretching once their end - to - end distance reaches few microns length .
following our work one may reach the conclusion that the batchelor prediction @xcite is irrelevant . by all means ,
this is not the case !
batchelor considered the evolution of the total length of a material line while this work , like the above mentioned reports @xcite , focused on the shortest distance between pairs of tracers starting with a finite initial separation .
therefore the question of what natural or experimental systems admit the assumptions underlying the exponential pair dispersion eq.[eq : exp_prediction ] for tracer particles persists .
[ [ methods - summary ] ] methods summary + + + + + + + + + + + + + + + the work presented here relies on constructing a database of trajectories in an elastic turbulence flow .
elastic turbulence is essentially a low reynolds number and a high weissenberg number phenomenon .
the former means the inertial non - linearity of the flow is over - damped by the viscous dissipation .
the latter estimates how dominant is the non - linear coupling of the elastic stresses to the spatial gradients of the velocity field compared with the dissipation of these stresses via relaxation .
this is the leading consideration in the design of the flow cell .
the lagrangian trajectories are inferred from passive tracers seeded in the fluid . in order to study the dynamics of pairs ,
the 3d positions of the tracers are needed to be resolved , even when tracers get nearby to each other .
the requirement of large sample statistics dictates the long duration of the experiment , which lasts over days .
the fluctuations due to the chaotic nature of the flow set the temporal resolution at milliseconds .
this leads to a data generation rate of about .
hence both the acquisition and the analysis processes are required to be steady and fully automated . the 3d positions of the fluorescent particles are determined using 2d single camera imaging , by measuring the diffraction rings generated by the out - of - focus particle . this way the particle localisation problem turns into a ring detection problem @xcite .
[ [ microfluidic - apparatus ] ] microfluidic apparatus + + + + + + + + + + + + + + + + + + + + + + the experiments were conducted in a microfluidic device , implemented in polydimethylsiloxane elastomer by soft lithography@xcite , consisting of a curvilinear tube having a rectangular cross - section .
the depth is measured to be , the width is approximately ( see fig.[fig : tube_geometry ] ) .
the geometry consists of a concatenation of 33 co - centric pairs of half circles .
the working fluid consists of polyacrylamide ( mw= at mass fraction of 80 parts per million ) in aqueous sugar syrup ( 1:2 sucrose to d - sorbitol ratio ; mass fraction of 78% ) , seeded with fluorescent particles ( 1 micron fluoresbrite yg carboxylate particles , polysciences inc . ) at number density of about 50 tracers in the observation volume .
the viscosity of the newtonian solvent , without the polymers , is estimated to be 1100 times larger than water viscosity at .
this leads to a polymer longest relaxation time of @xmath38 @xcite , which is the longest time scale characterising the relaxation of elastic stresses in the solution .
the flow was driven by gravity .
the maximal time - averaged velocity was measured to be roughly @xmath39{250}{\um\per\second}$ ] .
this results in a reynolds number @xmath40 and a global weissenberg number @xmath41 . @xmath42
using the estimation of the lyapunov exponent ; use ( * ? ? ?
20 ) [ [ imaging - system ] ] imaging system + + + + + + + + + + + + + + the imaging system consists of an inverted fluorescence microscope ( imt-2 , olympus ) , mounted with a plan - apochromat 20@xmath43/0.8na objective ( carl zeiss ) and a fluorescence filter cube ; a royal - blue led ( luxeonstar ) served for the fluorophore excitation .
a ccd ( gx1920 , allied vision technologies ) was mounted via zoom and 0.1@xmath43 c - mount adapters ( vario - orthomate 543513 and 543431 , leitz ) , sampling at , @xmath43 , covering @xmath43 laterally and the full depth of the tube .
the camera control was based on a modification of the motmot python camera interface package @xcite , expanded with a home - made plug - in , to allow real - time image analysis in the ram @xcite , recording only the time - lapse positions of the tracers to the hard drive .
[ [ lagrangian - particle - tracking ] ] lagrangian particle tracking + + + + + + + + + + + + + + + + + + + + + + + + + + + + to construct trajectories , the particle localisation procedure , introduced in @xcite , has to be complemented by a linking algorithm .
here we implemented a kinematic model , in which future positions are inferred from the already linked past positions .
we used the code accompanying @xcite as a starting point .
the algorithm was rewritten in python @xcite , generalised to n - dimensions , the kinematic model modified to account for accelerations as well , a memory feature was added to account for the occasional loss of tracers , and it was optimised for better performance .
the procedure accounts for the physical process of particles advected by a smooth chaotic flow and for the uncertainties .
these arise from the chaotic in time nature of the flow ( `` physical noise '' ) as well as from localisation and past linking errors ( `` experimental noise '' ) .
finally , natural smoothing cubic splines are applied to smooth - out the experimental noise and estimate the velocities and accelerations @xcite .
the smoothing parameter is chosen automatically , where vapnik s measure plays the role of the usual generalised cross - validation@xcite , adapted from the octave splines package @xcite@xcite . [
[ pairs - analysis ] ] pairs analysis + + + + + + + + + + + + + + within the trajectories database we have identified pairs of tracers which were found at some instant at a separation distance close to a prescribed initial separation @xmath44 , to within @xmath45 . the initial time @xmath46 for a trajectory was set by the instant at which the separation distance was closest to @xmath22 .
this way , each pair separation trajectory @xmath47 can contribute to an @xmath22 pairs ensemble at most once .
see fig.[fig : intro ] .
the number of pairs considered in each @xmath22 ensemble is plotted in fig.[fig : sample_size ] as function of @xmath28 .
examining the ensemble averages of the relative separation velocity at the initial time @xmath48 , presented in fig.[fig : ul_ro_profile ] , we do not find an indication that our sampling method introduces a bias for converging or diverging trajectories , at least for @xmath49 .
such a bias could manifest itself in non - vanishing relative separation velocity at these scales .
we have yet to resolve the reason for the negative values observed at the larger scales .
our data supports the linear flow approximation assumption at small enough scales , as indicated by the ensemble averages of the initial relative separation velocity ; see the inset of fig.[fig : uo2_ro_profile]_b _ where @xmath50 levels - off at @xmath51 .
the same regime is not reached for the relative velocity , yet the data in the inset of fig.[fig : uo2_ro_profile]_a _ does not rule out this possibility for smaller scales .
one note of warning should made as to the potential limitations when inferring the eulerian mean squared relative velocity from direct lagrangian particle trajectories , e.g. using @xmath52 in fig.[fig : uo2_ro_profile ] as an estimator for the eulerian second order structure function .
the tracer number density in our experiment sets the mean separation distance to a value which is larger than the initial separations considered in this work .
the analysis of the possible consequences are beyond the scope of this report .
yet we have no evidence for a discrepancy in the quantities we have considered here .
[ [ section-1 ] ] [ [ acknowledgements ] ] acknowledgements + + + + + + + + + + + + + + + + we thank a. frishman for the helpful and extensive discussions of the theory , o. hirschberg for useful discussions of the mathematical and statistical analysis , m. feldman for his insights regarding the implications to biology , and y. silberberg for his helpful advices .
ea had fruitful discussions with j. bec , s. musacchio , d. vincenzi , ew saw and r. chetrite , kindly organised by the latter ; both authors gained from discussions with v. lebedev and g. falkovich .
this work is supported by the lower saxony ministry of science and culture cooperation ( germany ) .
( image ) at ( 0,0 ) ; | the leading paradigm for chaotic flows dominated by dissipation predicts an exponential growth of the mean distance between pairs of fluid elements , in the long run .
this is reflected in the analysis of experimental results on tracer particles and the discussions which follow , as reported in recent experimental and numerical publications . to quantitatively validate this prediction ,
we have conducted a microfluidic experiment generating _ elastic turbulence _ , a flow characterised in the literature as smooth in space and random in time . to our great surprise , we discovered that the pair separation follows a much slower power - law also known as _ ballistic _ a notion overlooked so far for flows of this type .
we provide conclusive experimental evidence that this scaling is well - desribed by the same coefficients derived from the short - time dynamics .
our finding reinforces the role of the ballistic regime over a significant range in time and space , providing a quantitative estimation for the spreading of particles in mixing microfluidic flows based on the initial velocity snapshot .
finally we note that the conditions for the asymptotic exponential pair separation are quite stringent when it comes to tracer particles and are unlikely to be realised in wall - bounded flows
. therefore this work raises questions regarding the relevance and applicability of the currently leading paradigm .
* keywords . *
chaotic flows pair dispersion microfluidic fluid dynamics elastic turbulence [ [ section ] ] when we stir sugar in a cup of coffee we typically drive the liquid in circles using the tea - spoon , yet the flow quickly evolves into a three - dimensional chaotic one , tremendously accelerating the homogeneous distribution of the sweetener throughout the beverage .
turbulent flows , typical in our everyday life , are renowned for their efficient mixing and intensification of particle dispersion , having impact on natural processes as well as wide applications in the industry @xcite . at their smaller scales
the so - called dissipative or kolomogorov scales such flows are characterised by a velocity field which is smooth in space yet fluctuates strongly in time , also known as the _ batchelor regime _ @xcite .
understanding transport phenomena at small scales is of importance for the study of combustion , contamination and the deformation of polymers by the flow @xcite , to name but a few physical examples , as well as for the study of biological processes such as cytoplasmic streaming @xcite , communication via chemo - attractants and food searching @xcite .
moreover , it has immediate technological implications , mainly when it comes to miniature devices .
microfluidic systems have become a common tool in research and industry . as a prominent part of lab - on - a - chip apparatuses
, they are implemented for micro - chemistry @xcite , soft condensed matter @xcite , bioanalytics such as pcr @xcite , biomedicine and other research and engineering applications @xcite . nevertheless , there is limited knowledge of transport phenomena in these devices beyond the scenario of laminar steady flows , which is not the only case of interest @xcite .
the study of particle dispersion in flows is at the basis of the understanding of transport processes @xcite . in the early 1950s , g. k. batchelor predicted that the mean length of material lines in turbulent flows would grow exponentially in the course of time , in the long time limit ; this stems from the notion that the line elements it consists of can be considered short enough such that the distance between the ends of an element remain within the dissipative scale throughout the motion @xcite . in some recent works the terms _ material line
_ and _ material line element _ are used interchangeably to indicate the separation vector between two passive particles in the fluid @xcite . and
indeed batchelor s prediction has been later reformulated for tracer particles in the form of exponential pair separation ; see @xcite for example . to illustrate how this comes about ,
let us consider a pair of passive tracers separated by the vector @xmath0 , and whose relative velocity is @xmath1 .
the evolution of the squared separation distance @xmath2 follows @xmath3 where @xmath4 is the separation velocity , defined by the above relation .
the analysis then proceeds by assuming a linear flow approximation the pair separation is considered to be small enough such that the velocity of one tracer is linearly related to that of the other .
using this approximation eq.[eq : r2_eom ] reduces to @xmath5 where @xmath6 no longer depends on @xmath7 . for the case of chaotic flows
, @xmath6 can be modelled as a random variable , and analyses often focus on the expectation value of such equations @xcite . additionally assuming the correlation time of @xmath6 to be very short compared to the observation time @xcite , a generalisation of the central limit theorem the multiplicative ergodic theorem of oseledec @xcite is applied , resulting in the exponential pair dispersion @xmath8 this is a relation for the time evolution of the second moment of pair separation distances . in this form
one can identify @xmath9 with the generalised lyapunov exponent of the second order , which is generically not trivially related to the ordinary ( maximal ) lyapunov exponent ; see @xcite , @xcite , @xcite and others .
much of the theoretical and numerical literature discussing pair dispersion in the dissipative sub - range is devoted to the evaluation of @xmath10 in terms of the typical time - scale of the flow @xmath11 , that is @xmath12 ; see ( * ? ? ?
* eq.2.9 ) .
the value of @xmath13 is still under debate as can be learnt from @xcite as well as @xcite and references therein . at the time of writing
, the exponential pair dispersion is regarded as the leading paradigm for chaotic flows which are spatially smooth , as manifested by the analysis of recent experimental results and the discussions which follow .
jullien @xcite studied an instance of the batchelor regime flow in 2d turbulence , where the velocity field was inferred experimentally followed by numerical integration of tracers simulated on a computer ; the initial pair separation values were set to distances smaller than the measurements grid .
an exponential separation , referred to in that context as lin s law @xcite , was reported during an intermediate time interval of between one to twice the value of the estimated @xmath11 , after which a power - law scaling has been observed .
salazar and collins @xcite estimated @xmath13 from measurements of @xmath14 in 3d turbulence reported by guala et al .
@xcite ; luthi et al .
@xcite employed 3d particle tracking velocimetry for direct measurements of @xmath14 in 3d turbulence .
the authors did not infer a value for @xmath13 , nevertheless , in the lack of other sources we could try to get an estimation . the data in fig.10(a ) of @xcite and fig.1 of @xcite imply that @xmath14 saturates after about @xmath15 at a value of approximately , where the kolmogorov time - scale @xmath11 is estimated at 0.26 ; following eq .
2.1 and 2.6 in @xcite this would correspond to @xmath16 . based on @xcite . ]
this estimate should be taken with grain of salt not only because it is unclear whether these measurements are indeed restricted to the dissipative scales but also as , although related , the quantities @xmath14 and @xmath10 are not the same @xcitefootnote 32 p.582 . even more recently , ni and xia @xcite reported measurements in 3d turbulent thermal convection and inferred @xmath13 reported @xmath17 as concluded from exponential fits to the mean squared pair separation distance ; as presented in ( * ? ? ?
* fig .
1 ) , the fits are taken at time intervals of up to one kolmogorov time - scale , a time too short with respect to the underlying assumptions , and thereafter the data grows faster than the evaluated exponentials .
we therefore find it safe to conclude that the literature on the subject is lacking conclusive experimental evidence .
in this paper we report experimental results of particle pair dispersion where , in contrast to the above mentioned publications @xcite , the memory of the initial relative velocity prevails the dynamics , showing no signature of the asymptotic exponential growth . studying pair separation in the dissipative sub - range over long times in intense turbulence
poses a technological challenge .
the high velocities , typical of high reynolds flows , restrict the length of the obtained trajectories as exemplified by the above mentioned reports @xcite and other recent works @xcite .
this is one of the reasons for which the experimental literature on pair dispersion in smooth chaotic flows is lagging behind the theoretical one . and has yet to reach a conclusive evidence .
however , chaotic flows do not necessarily require fast velocities or large vessels
. one might expect viscous flows to be regular in microfluidic tubes and the question of mixing in these apparatuses may seem irrelevant due to their low reynolds nature .
nevertheless , when a minute amount of long flexible polymers , such as dna and protein filaments , are introduced , the flow may develop a series of elastic instabilities which render it irregular and twisted , a phenomenon termed _
elastic turbulence _
@xcite .
the study of pair separation dynamics in elastic turbulence , taking place inside a tiny tube , presents technical challenges : the positions of tracers are needed to be resolved over long times and distances , in particular when the tracers get nearby to each other , whereas the flow is chaotic and three - dimensional ; the scales at which the dynamics takes place require the use of a microscope , where 3d imaging is non - trivial ; the flow fluctuations in time dictate a high temporal resolution ; the statistical nature of the problem demands a large sample of trajectories , which in turn requires long acquisition times and reliable automation . to satisfy these requirements
a novel method was implemented .
the 3d positions of the fluorescent particles are determined from a single camera 2d imaging , by measuring the diffraction rings generated by the out - of - focus particle ; this way the particle localisation problem turns into a ring detection problem , which is addressed accurately and efficiently in ref .
@xcite .
below we report the results of an experiment designed to measure the value of @xmath13 using direct lagrangian particle tracking in an elastic turbulence flow in a microfluidic tube . |
let @xmath2 denote the simply connected space form of constant curvature @xmath3 where @xmath4 or @xmath5 and dimension @xmath6 .
let @xmath0 be a closed hypersurface of @xmath7 . when @xmath8 and @xmath0 a closed hypersurface of @xmath9 , bleecker
weiner @xcite proved that the first eigenvalue @xmath10 of the laplace operator of @xmath0 satisfies the inequality : @xmath11 , where @xmath12 denotes the square of the length of the second fundamental form of the hypersurface @xmath0 . in @xcite ,
reilly improved this inequality to show that @xmath13 , where @xmath14 is the mean curvature of the hypersurface @xmath0 .
these inequalities of bleecker weiner and reilly are also sharp for geodesic spheres in @xmath9 .
since then , reilly s inequality has been extended to hypersurfaces in other simply connected space forms ( see @xcite and @xcite for details and related results ) . while trying to understand these results , we noticed that one can obtain a similar sharp upper bound for the first eigenvalue @xmath10 , of closed hypersurfaces @xmath0 in rank-1 symmetric spaces .
namely @xmath15 where @xmath16 is the first eigenvalue of the geodesic sphere @xmath17 with center at a point @xmath18 called the center - of - mass of the hypersurface @xmath0 and radius @xmath19 .
we refer to @xcite and @xcite for the basic riemannian geometry used in this paper .
to state our results we need the notion of center - of - mass and result on the existence of center - of - mass for measurable subsets of @xmath1 .
let @xmath20 be a complete riemannian manifold .
for a point @xmath21 , we denote by @xmath22 , the convexity radius of @xmath20 at @xmath23 . for a subset @xmath24 for some @xmath25 , we let @xmath26 denote the convex hull of @xmath27 .
let @xmath28 be the exponential map and @xmath29 be the normal coordinates centered at @xmath30 .
we identify @xmath26 with @xmath31 and we also denote @xmath32 as @xmath33 for @xmath34 .
we now state and prove the center - of - mass theorem ( see also @xcite and @xcite ) .
[ thm1 ] let @xmath27 be a measurable subset of @xmath20 contained in @xmath35 for some point @xmath36 .
let @xmath37\to\r$ ] be a continuous function such that @xmath38 is positive on @xmath39 .
then there exists a point @xmath40 such that @xmath41 where @xmath42 denotes the geodesic normal coordinate system centred at @xmath23 . for @xmath43 , we define @xmath44 where @xmath45 is a geodesic normal coordinate system centred at @xmath30 .
we shall now show that the continuous vector field @xmath46 points inward along the boundary @xmath47 of @xmath26 .
then the theorem follows from the brouwer s fixed point theorem . since @xmath26 is convex ,
it is contained in the half - space for every @xmath48 , where @xmath49 denotes the outward normal to @xmath47 at @xmath30 .
this implies that @xmath50 for all @xmath51 .
thus @xmath46 points inward along the boundaryof @xmath26 .
we can find a @xmath52 such that @xmath53 for every @xmath43 .
then the continuous map @xmath54 defined by @xmath55 has a fixed point @xmath40 by the brouwer s fixed point theorem .
hence @xmath56 .
this completes the proof of the theorem.@xmath57 @xmath58 the point @xmath23 in the theorem is called a center - of - mass of the measurable subset @xmath27 with respect to the mass distribution function @xmath38 .
before we state our results , we fix some notations that we will be using throughout this paper .
we let @xmath59 denote any one of the following rank-1 symmetric spaces : the round sphere @xmath60 with constant sectional curvature @xmath61 , complex projective space @xmath62 , quaternionic projective space @xmath63 and the cayley projective plane @xmath64 with sectional curvature @xmath65 or their non - compact duals with sectional curvature @xmath66 .
we also write @xmath67 , where @xmath68 ; @xmath69 , @xmath70 , @xmath71 or @xmath72 .
we also let @xmath73 denote the ricci curvature of @xmath1 and also remark that @xmath73 is constant .
the round sphere @xmath60 with constant sectional curvature @xmath74 is denoted by @xmath75 given a point @xmath76 , we let @xmath17 denote the geodesic sphere of radius @xmath77 with center @xmath23 .
let @xmath78 denote the laplacian of the geodesic sphere @xmath17 with respect to the induced metric and @xmath79 denote the first eigenvalue of @xmath78 .
[ thm2 ] let @xmath0 be a closed hypersurface in a simply connected rank-@xmath74 symmetric space @xmath80 of compact type .
assume that @xmath0 is contained in a ball of radius @xmath81 .
let @xmath82 be the center - of - mass of @xmath0 with respect to the mass distribution function @xmath83
. then @xmath84 & = \frac{1}{{\rm vol}(m)}\int_m(\mid\!a(r)\!\mid^2+{\rm ric}_{\olm}),\end{aligned}\ ] ] where @xmath85 is the distance from the point @xmath82 to the point @xmath86 .
furthermore@xmath87 equality holds in the above inequality iff @xmath0 is a geodesic sphere with center @xmath82 .
[ thm3 ] let @xmath0 be a closed hypersurface in a simply connected rank-@xmath74 symmetric space @xmath80 of non - compact type .
let @xmath82 be the center - of - mass of @xmath0 with respect to the mass distribution function @xmath83
. then @xmath88 & = \frac{1}{{\rm vol}(m)}\int_m(\mid\!a(r)\!\mid^2+{\rm ric}_{\olm}),\end{aligned}\ ] ] where @xmath85 is the distance from the point @xmath82 to the point @xmath86 .
furthermore@xmath87 equality holds in the above inequality iff @xmath0 is a geodesic sphere with center @xmath82 .
let @xmath89 . let @xmath17 be a geodesic sphere of radius @xmath77 in @xmath59 centred at a point @xmath90 .
we identify @xmath17 with the inverse image @xmath91 with the metric @xmath92 .
let @xmath93 denote the laplacian of @xmath17 and @xmath94 the first eigenvalue of @xmath93 .
in this section we study the first eigenvalue @xmath94 of @xmath93 .
this is also done in @xcite and @xcite . if @xmath96 , the geodesic sphere @xmath17 is isometric to @xmath97 .
therefore the first eigenvalue of @xmath17 is @xmath98 .
we know that there is a canonical riemannian submersion @xmath99 with connected totally geodesic fibres , where @xmath100 is the simply connected compact symmetric space of dimension @xmath101 .
if @xmath102 , we can decompose the laplacian @xmath93 of @xmath17 as @xmath103 we also know that the euclidean co - ordinate functions @xmath104 s , for @xmath105 , are the first eigenfunctions of @xmath106 corresponding to the first eigenvalue @xmath107 ( see @xcite for details ) .
since the fibres are all totally geodesic , when we restrict these eigenfunctions to each fibre , they become eigenfunctions of @xmath108 with eigenvalue @xmath109 hence we get @xmath110 for @xmath105 .
let @xmath111 denote the horizontal laplacian of the riemannian submersion .
since @xmath112 all the eigenfunctions of @xmath113 are also eigenfunctions of @xmath93 with the same eigenvalues . in particular , the first non - zero eigenvalue @xmath114 occurs as an eigenvalue of @xmath93 also .
now @xmath115 iff @xmath116 since @xmath117 and @xmath6 , we see that @xmath118 .
therefore @xmath119 for @xmath120 and consequently the functions @xmath104 , for @xmath105 , are all first eigenfunctions of @xmath93 with the first eigenvalue @xmath121 we denote by @xmath122 the compact dual of @xmath1 .
if @xmath96 , the geodesic sphere @xmath17 is isometric to @xmath123 and hence the first eigenvalue @xmath94 of @xmath17 is @xmath124 .
we have the canonical riemannian submersion @xmath125 with connected totally geodesic fibres .
if @xmath102 , we can decompose the laplacian @xmath93 as @xmath126 we know that the euclidean coordinate functions @xmath104 s , for @xmath105 , are eigenfunctions of @xmath93 with eigenvalue @xmath127 since @xmath128 all the eigenfunctions of @xmath129 are also eigenfunctions of @xmath93 with the same eigenvalues . in particular , the first non - zero eigenvalue @xmath130 occurs as an eigenvalue of @xmath93 also .
now @xmath131 will be the first non - zero eigenvalue of @xmath93 so long as @xmath132 since the inequality above is valid for all @xmath133 , we get that @xmath134 for all @xmath133
. we will now relate @xmath94 of @xmath17 with the square of the length of the second fundamental form of @xmath17 and the ricci curvature of @xmath1 .
let @xmath135 be a geodesic starting at the point @xmath23 .
let @xmath136 be the symmetric endomorphism defined by @xmath137 , where @xmath138 is the curvature tensor of @xmath1 . for @xmath139
, we let @xmath140 denote the weingarten map @xmath141 of the smooth hypersurface @xmath142 at the point @xmath143 .
it is known that these family of symmetric endomorphisms @xmath140 satisfy the riccati equation @xmath144 along the geodesic @xmath135 .
therefore , by taking the trace of these endomorphisms we get @xmath145 & = |a(r)|^2+{\rm ric}_{\olm}(\gamma ' , \gamma')\end{aligned}\ ] ] where @xmath146 is the square of the length of the second fundamental form of the geodesic sphere @xmath17 and @xmath73 is the ricci curvature of @xmath1 .
let @xmath147 , @xmath148 , @xmath149 , @xmath150 be an orthonormal basis of @xmath151 such that the vectors @xmath152 , for @xmath153 , are tangent to the fibre of the canonical riemannian submersion @xmath154 ( \olm_{n-1}^ * , \hbox{d}s^2)&\text{if $ ( \olm , \hbox{d}s^2)$ is of non - compact type}. \end{cases}\ ] ] then an easy jacobi field computation shows that @xmath155 \coth r\!e_i & \text{if $ ( \olm , \hbox{d}s^2)$ is of non - compact type } \end{cases}\ ] ] for @xmath153 and @xmath156 \frac{1}{2}\coth(r/2 ) e_i & \text{if $ ( \olm , \hbox{d}s^2)$ is of non - compact type } \end{cases}\ ] ] for @xmath157 .
therefore @xmath158 \frac{k(n-1)}{2}\coth(r/2)+(k-1)\coth r & \text{if $ ( \olm , \hbox{d}s^2)$ is of non - compact type } \end{cases}\end{aligned}\ ] ] and @xmath159 .
we will now prove theorems [ thm2 ] and [ thm3 ] .
let @xmath165 for @xmath105 .
since @xmath166 , we can use these functions @xmath167 s as test functions in the rayleigh quotient .
therefore @xmath168 for @xmath105 , where @xmath169 denotes the gradient in @xmath0 . since @xmath170 , we see that @xmath171 .
hence @xmath172 if we now show that @xmath173 then we are through .
let us now recall that @xmath179 , where @xmath180 is the laplacian of @xmath59 , @xmath27 is the weingarten map of the hypersurface @xmath0 and @xmath181 is the unit outward normal to @xmath0 . using this identity
, we write that @xmath182 .
we decompose the laplacian @xmath180 of @xmath1 as @xmath186 along the radial geodesics starting at @xmath82 .
now @xmath187 for @xmath188 .
therefore @xmath189 . in [ sec2 ] , we have shown that @xmath190 for @xmath191 .
now @xmath192 & = \frac{1}{2}\frac{\partial^2}{\partial\nu^2}(f_i^2)- \left(\frac{\partial\,f_i}{\partial\nu}\right)^2.\end{aligned}\ ] ] therefore @xmath193 and @xmath194 similarly , @xmath195 & = 0.\end{aligned}\ ] ] we substitute these quantities in @xmath196 to get @xmath197 & = \int_m\lambda_1(s(r))\sum_{i=1}^{kn}f_i^2- \sum_{i=1}^{kn}\int_m\left(\frac{\partial\,f_i}{\partial\nu}\right)^2\nonumber\ ] ] @xmath198 & \leq\int_m\lambda_1(s(r))\label{ineq5}\\[.4pc ] & = \int_m(|a(r)|^2+{\rm ric}_{\olm})\nonumber.\end{aligned}\ ] ] this completes the proof of the inequality . for the equality part of the proof
, we notice that equality holds in the above inequality iff @xmath199 , for @xmath105 , at all points in @xmath0 .
this is true iff the unit outward normal field is same as the radial vector field @xmath200 .
hence the equality holds iff @xmath0 is a geodesic sphere of radius @xmath201 with center @xmath82 .
this completes the proof of theorems [ thm2 ] and [ thm3 ] .
[ thm4 ] let @xmath0 be a closed hypersurface in @xmath9 .
let @xmath82 be the center - of - mass of @xmath0 with respect to the mass distribution function @xmath83
. then @xmath202 & = \frac{1}{{\rm vol}(m)}\int_m|a(r)|^2\nonumber\\[.4pc ] & = \frac{n-1}{{\rm vol}(m)}\int_m\frac{1}{r^2(x)},\end{aligned}\ ] ] where @xmath85 is the distance from the point @xmath82 to the point @xmath86 .
furthermore@xmath87 equality holds in the above inequality iff @xmath0 is a geodesic sphere with center @xmath82 .
let us first observe that the inequality [ ineq5 ] in the proof of theorems [ thm2 ] and [ thm3 ] is valid in all riemannian manifolds in which the functions @xmath203 are first eigenfunctions of the geodesic sphere @xmath17 . this is true in @xmath9 .
further , the first eigenvalue @xmath94 of the geodesic sphere @xmath17 in @xmath9 is @xmath204 .
therefore @xmath205 & = ( n-1 ) \int_m \frac{1}{r^2(x)}.\end{aligned}\ ] ] hence @xmath206 | let @xmath0 be a closed hypersurface in a simply connected rank-1 symmetric space @xmath1 . in this paper , we give an upper bound for the first eigenvalue of the laplacian of @xmath0 in terms of the ricci curvature of @xmath1 and the square of the length of the second fundamental form of the geodesic spheres with center at the center - of - mass of @xmath0 .
p h # 1#2 # 1.2ex # 1 .2ex [ theore]*theorem * [ theore]definition [ theore]remark |
in recent years effective field theory ( eft ) techniques have found increasing acceptance in nuclear physics .
eft facilitates the systematic separation of the effects of high - momentum physics from those of low - momentum physics . in nuclear physics
there are many different momentum scales present .
but for the @xmath5 system the key ones would seem to be : @xmath6 with @xmath7 the binding energy of deuterium , @xmath8 and @xmath9 the pion and nucleon masses , and @xmath10 the scale of chiral - symmetry breaking , which is of the order of the mass of the @xmath11 meson . an eft based on
the first hierarchy@xmath12has had considerable success describing very - low - energy reactions in @xmath5 and @xmath13 ( for reviews see @xcite ) , and has recently been extended to @xmath14 @xcite . here
i will focus on the eft built on the second hierarchy@xmath15 .
this eft is chiral perturbation theory ( @xmath4pt ) ( for a review see @xcite ) . here
the low - momentum physics is described by a quantum field theory of nucleons and pions whose lagrangian is the most general one that respects the approximate chiral symmetry of qcd and the pattern of its breaking .
higher - energy effects of qcd appear in @xmath4pt as non - renormalizable contact operators .
any process involving nucleons and an arbitrary number of ( soft ) pions and photons should then be computable as an expansion in the ratio of nucleon or probe momenta ( denoted here by @xmath16 and @xmath17 ) and @xmath8 to @xmath10 . since typical momenta in light nuclei are @xmath18 we would expect to be able to calculate the response of such nuclei to low - momentum electroweak probes using @xmath4pt . as we will see below
, @xmath4pt provides a systematic way to compute what in nuclear - structure physics would be called `` bare operators '' .
but the operators computed in this fashion include pieces which account for the fact that they are not actually `` bare '' if one thinks about the physics at scale @xmath19 which has been `` integrated out '' of the low - momentum theory of nucleons and pions : high - momentum nucleon modes , the delta isobar , mesons such as the rho , pomerons , m - branes , etc . here
i will argue that these operators produce results for electromagnetic few - nucleon reactions which are model independent , can be systematically improved by going to higher order , and yield a good description of the data over the region of validity of the eft .
section 2 outlines the @xmath4pt expansion , and briefly discusses how to obtain deuteron wave functions using it .
section 3 then looks at electron - deuteron scattering in @xmath4pt as a probe of deuteron structure .
section 4 examines the use of compton scattering on the deuteron in particular as a way to extract isoscalar polarizabilities of the nucleon .
consider an elastic scattering process on the deuteron . if @xmath20 is the transition operator for this reaction
then the amplitude in question is simply @xmath21 , with @xmath22 the deuteron wave function . in this section ,
we follow weinberg @xcite , and divide the formulation of a systematic expansion for this amplitude into two parts : the expansion for @xmath20 , and the construction of @xmath22 .
chiral perturbation theory gives a systematic expansion for @xmath20 of the form @xmath23 where we have labeled the contributions to @xmath20 by their order @xmath24 in the small parameter @xmath25 , @xmath26 , @xmath27 .
to construct @xmath28 one first writes down the vertices appearing in the chiral lagrangian up to order @xmath24 .
one then draws all of the two - body , two - nucleon - irreducible , feynman graphs for the process of interest which are of chiral order @xmath29 .
the rules for calculating the chiral order of a particular graph are : * each nucleon propagator scales like @xmath30 ( provided that the energy flowing through the nucleon line is @xmath31 ) ; * each loop contributes @xmath32 ; * graphs in which both particles participate in the reaction acquire a factor of @xmath33 ; * each pion propagator scales like @xmath34 ; * each vertex from the @xmath24th - order piece of the chiral lagrangian contributes @xmath29 . in consequence
more complicated contributions involving two - body mechanisms , and/or higher - order vertices , and/or more loops are suppressed by powers of @xmath35 .
this leaves us with the problem of constructing a deuteron wave function that is consistent with the operator @xmath20 .
weinberg s proposal was to use the @xmath4pt expansion ( [ eq : expansion ] ) for the @xmath36 potential @xmath37 , and then solve the schrdinger equation to find the deuteron ( or other nuclear ) wave function @xcite .
matrix elements @xmath38 should then incorporate the consequences of chiral - symmetry breaking in a model - independent way .
calculations have shown that the @xmath36 phase shifts can be understood , and deuteron bound - state static properties reliably computed , with @xmath36 potentials derived from @xmath4pt @xcite .
this is now a sophisticated enterprise , with n@xmath39lo @xmath40o@xmath41 $ ] potentials recently having been obtained @xcite .
these potentials can reproduce @xmath36 data with an accuracy which rivals that of `` high - quality '' @xmath36 potentials , at least for @xmath42 mev
. in the goldstone - boson and single - nucleon sector @xmath4pt works because loop effects are generically suppressed by powers of the small parameter @xmath35 . in zero and
one - nucleon reactions the power counting in @xmath35 applies to the amplitude , and not to the two - particle potential .
however , the existence of nuclei tells us immediately that a power counting in which all loop effects are suppressed can not be correct for the multi - nucleon case , since if it were there would be no nuclei , and neither we , or this workshop , would occur .
weinberg s proposal to instead power - count the potential is one response to this dilemma .
however , its consistency has been vigorously debated in the literature ( see @xcite for reviews ) .
et al . _
@xcite have shown that weinberg s proposal is consistent to leading order in the @xmath43 channel , but there remain questions about the consistency in other channels and at higher orders @xcite . here
we will use wave functions calculated according to the weinberg proposal .
one of our objectives will be to assess whether such a procedure really does lead to convergent eft expansions for the processes under consideration , or whether problems along the lines of those raised in refs .
@xcite also occur in the chiral eft for electroweak reactions on deuterium .
one quantitative test of @xmath4pt s deuteron wave functions is provided by elastic electron - deuteron scattering .
we thus turn our attention to the deuteron electromagnetic form factors @xmath44 , @xmath45 , and @xmath46 .
these are matrix elements of the deuteron current @xmath47 , with : @xmath48 where we have labeled these ( non - relativistic ) deuteron states by the projection of the deuteron spin along the direction of the momentum transfer @xmath49 and @xmath50 .
@xmath44 , @xmath45 , and @xmath46 are related to the experimentally - measured @xmath51 , @xmath7 , and @xmath52 in the usual way , with @xmath52 being primarily sensitive to @xmath53 and @xmath7 depending only to @xmath46 . here
we will compare calculations of the charge and quadrupole form factor with the recent extractions of @xmath44 and @xmath45 from data @xcite .
both of these form factors involve the zeroth - component of the deuteron four - current @xmath54 .
here we split @xmath54 into two pieces : a one - body part , and a two - body part .
the one - body part of @xmath54 begins at order @xmath55 ( the proton charge ) with the impulse approximation diagram calculated with the non - relativistic single - nucleon charge operator for structureless nucleons .
corrections to the single - nucleon charge operator from relativistic effects and nucleon sub - structure are suppressed by two powers of @xmath35 , and thus arise at @xmath56 , which is the next - to - leading order ( nlo ) for @xmath44 and @xmath45 . at this order one might also expect meson - exchange current ( mec ) contributions .
however , all mecs constructed with vertices from @xmath57 are isovector .
these play a role in , e.g. @xmath58 @xcite . the first mec effect in @xmath59 does not occur until n@xmath60lo , or @xmath61 , where an @xmath62 vertex from @xmath63 gets included in an isoscalar two - body contribution to the deuteron charge operator ( see fig .
[ deuterongraphs](b ) ) .
the most important correction that arises at nlo is the inclusion of _ nucleon _ sub - structure . at @xmath56
the isoscalar nucleon form factors are dominated by short - distance physics , and so the only correction to the point - like leading - order result comes from the inclusion of the nucleon s electric radius . for the isoscalar combination of nucleon electric form factors @xmath4pt to @xmath56 gives : @xmath64 this description of nucleon structure breaks down at momentum transfers @xmath17 of order 300 mev .
there is a concomitant failure in the description of @xmath1 scattering data @xcite . in order to focus on _ deuteron _ structure , in the results presented below
i have chosen to circumvent this issue by using a `` factorized '' inclusion of nucleon structure @xcite : @xmath4pt is used to compute the ratio @xmath65 .
this allows us to use experimentally - measured single - nucleon form factors scattering data is one input to the extraction of the neutron electric form factor . ] in the calculation , thereby allowing us to test how far the theory is able to describe the @xmath36 dynamics .
the results for @xmath44 and @xmath45 are shown in fig .
[ fig - fcfq ] .
the figure demonstrates that convergence is quite good below @xmath66 mev .
the results shown are for the nlo chiral wave function .
it is clear that provided information from single - nucleon form factors is taken into account@xmath4pt is perfectly capable of describing the charge and quadrupole form factors of deuterium at least as far as the minimum in @xmath44 .
@xmath46 can be obtained in a similar way . the lo contribution to @xmath46
is @xmath67 .
the first two - body mechanisms enter at @xmath68 , when an undetermined two - body counterterm , as well as a long - range two - body current with an undetermined coefficient , appear @xcite .
results for @xmath46 at @xmath61 turn out to be of similar quality to those for @xmath44 @xcite , but are somewhat more sensitive to short - distance physics , as expected given the presence of a counterterm at nnlo in this observable .
using the nnlo chiral wave function , high - quality potential - model wave functions , or indeed simple wave functions which include only one - pion exchange regulated by a square well at short distances , does not modify the plots for @xmath44 and @xmath45 greatly below @xmath69 mev @xcite . as long as wave functions with the same long - distance parts
are employed these observables are apparently not very sensitive to the choice of @xmath22 . in the language of the no - core shell model they do not `` renormalize strongly '' .
( an important exception is the deuteron quadrupole moment , @xmath70 . )
this is good news for weinberg s power counting , because the first operators with which @xmath71 can be renormalized occur at @xmath72 ( fig .
[ deuterongraphs](c ) ) .
therefore it was to be expected that operator renormalization would have this small impact on @xmath44 and @xmath45 .
furthermore , the dominant piece of the operator renormalization can be taken into account by adjusting the coefficients of the @xmath72 operators to reproduce , e.g. @xmath73 and @xmath70 .
this would then push the omitted effects of operator renormalization in these observables to @xmath74 .
compton scattering on the nucleon at low energies is a fundamental probe of the long - distance structure of these hadrons .
this process has been studied in @xmath75 in @xcite .
the following results for the proton polarizabilities were obtained at @xmath76 : @xmath77 going to next order in the chiral expansion single - nucleon counterterms which shift the polarizabilities enter the calculation . using the @xmath78 amplitude computed by mcgovern to @xmath79 @xcite we fitted @xmath78 data over the kinematic range @xmath80 mev @xcite .
= 10.0 cm the best fit for the proton electric and magnetic polarizabilities is @xmath81 statistical ( 1-@xmath82 ) errors are inside the brackets .
an estimate of the theoretical error due to truncation of the expansion at @xmath83 is given outside the brackets .
the result ( [ eq : protpol1 ] ) is compatible with other extractions , although the central value of @xmath84 is higher @xcite .
one would like to perform a similar analysis for @xmath3 scattering data .
coherent compton scattering on a deuteron target has been measured at @xmath85 49 and 69 mev by the illinois group @xcite , at @xmath86 mev at saskatoon @xcite , and at @xmath87 and @xmath88 mev at maxlab at lund @xcite .
such data should permit an extraction of the isoscalar nucleon polarizabilities @xmath89 and @xmath90 .
these are interesting , not only as fundamental nucleon - structure parameters , but also because @xmath4pt predicts @xmath91 , @xmath92 at @xmath93 .
the amplitude for @xmath3 scattering involves mechanisms other than compton scattering on the individual constituent nucleons .
so , our desire to extract nucleon polarizabilities argues for a theoretical calculation of this reaction that is under control in the sense that it accounts for _ all _ mechanisms to a given order in @xmath75 . the compton amplitude we wish to evaluate is ( in the @xmath3 center - of - mass frame ) : @xmath94 where @xmath9 ( @xmath95 ) is the initial ( final ) deuteron spin state , and @xmath96 ( @xmath97 ) is the initial ( final ) photon polarization state , and @xmath98 ( @xmath99 ) the initial ( final ) photon three - momentum , which are constrained to @xmath100 .
the first integral is represented by the graphs of fig .
[ procomptonebodmod ] where the photon interacts with only one nucleon .
the second integral corresponds to the graphs of fig .
[ combined ] where the current is of a two - body nature . -2.2
cm the lo contribution to compton scattering on the deuteron is shown in fig . [ procomptonebodmod](a ) .
this graph involves a vertex from @xmath63 and so is @xmath101 .
this contribution is simply the thomson term for scattering on the proton .
there is thus no sensitivity to either two - body contributions _ or _ nucleon polarizabilities at this order . at @xmath102
there are several more graphs with a spectator nucleon ( figs .
[ procomptonebodmod](b),(c),(d ) ) , as well as graphs involving an exchanged pion with leading order vertices ( fig .
[ combined ] ) and one - loop graphs with a spectator nucleon ( fig .
[ procomptonebodmod](e ) ) @xcite .
graphs such as fig .
[ procomptonebodmod](e ) contain the physics of the proton and neutron polarizabilities at @xmath76 in @xmath4pt .
we employed a variety of wave functions @xmath22 , and found moderate wave - function sensitivity .
results shown here are generated with the nlo chiral wave function of ref .
[ cvgplot ] shows the results of a fit to the world s modern @xmath3 data .
the dotted line is the prediction at @xmath103 in the kernel , where the second contribution in eq .
( [ eq : gammad ] ) is zero , and the single - scattering contribution is given solely by fig .
[ procomptonebodmod](a ) .
it is quite remarkable how well the @xmath103 calculation reproduces the 49 mev data .
however , the agreement is somewhat serendipitous : there are significant @xmath76 corrections . at these lower photon energies
weinberg power counting begins to break down , since it is designed for @xmath104 , and does _ not _ recover the deuteron thomson amplitude as @xmath105 . correcting the power counting
to remedy this difficulty appears to improve the description of the 49 mev data , without significantly modifying the higher - energy results @xcite . at @xmath76
all contributions to the kernel are fixed in terms of known pion and nucleon parameters , so to this order @xmath75 makes _ predictions _ for deuteron compton scattering .
note that the @xmath76 corrections get larger as @xmath106 is increased as expected .
the agreement of the @xmath76 calculation with the intermediate - energy data sets is very good , although only limited conclusions can be drawn .
these results are not very different from other , potential - model , calculations @xcite . at @xmath79 terms
enter the single - nucleon amplitude which allow us to fit @xmath89 and @xmath90 to @xmath3 data .
a number of new two - body currents also appear . however , there are still no short - distance @xmath107 operators contributing to @xmath3 scattering at this order .
therefore an @xmath79 calculation allows us to test the single - nucleon physics which is used to predict the results of coherent scattering on deuterium , since there are no undetermined parameters in the two - body mechanisms that enter .
unfortunately the data is fairly sparse , and the sensitivity to the @xmath36 wave function quite large , so our fit does not tie down @xmath89 and @xmath90 very accurately . the result is : @xmath108 the errors inside the brackets are statistical and those outside are an estimate of the effect from higher - order terms .
( for details see ref .
the fit only includes data satisfying @xmath109 mev .
recent work suggests that to describe the backward - angle points at 95 mev requires the inclusion of an explicit delta degree of freedom @xcite .
the second error quoted on @xmath89 and @xmath90 is large because , as alluded to above , cross sections depend on the choice of @xmath110 at the 1020% level .
this change in the @xmath3 cross section has a sizable impact on the extraction of @xmath89 and @xmath90 .
while not all of the wave functions we employed are consistent with @xmath4pt , they do all have the correct long - distance behavior .
the differences we see are thus a consequence of different short - range behavior in , say , the nijm93 and nlo @xmath4pt @xmath36 potentials . as such they
should be renormalized by @xmath107 contact operators .
however , weinberg power counting predicts that the operators which do this do not appear until ( at least ) @xmath111 .
thus the degree of variability seen in our results with different wave functions could be viewed as inconsistent with this power counting .
indeed , it may be necessary to modify the power counting so that , e.g. @xmath107 contact operators appear at a lower order than is indicated by the naive @xmath4pt power counting we have used for @xmath20 .
the issue here is one of relevance to nuclear - structure physics : it speaks of the need to properly renormalize the effective @xmath107 operators found when high - energy degrees of freedom are integrated out of our eft of the @xmath36 system .
further understanding of this issue is also crucial if @xmath4pt is to be an accurate calculational tool for low - energy reactions on deuterium .
there is nothing in what i have presented so far that is particular to the photon as a probe .
analogous expansions can be made for axial currents .
the result is a systematic approach to reactions such as @xmath112 and @xmath113 .
i do not have room to properly describe the beautiful work on these reactions by park , ando , and collaborators @xcite .
in my view a particular strength of that work is the application of the same @xmath4pt electroweak operators to @xmath5 , 3 , and 4 . ref .
@xcite uses the well - known tritium beta decay rate to fix the strength of an ( otherwise poorly constrained ) two - body contribution to the gamow - teller transition operator .
this renormalizes the effective @xmath36 gamow - teller operator to such high accuracy that the @xmath114-fusion @xmath115 factor has an estimated error from higher - order terms of less than @xmath116% . the same one- and two - nucleon operators are then used to predict the reaction @xmath117 @xcite and @xmath118 scattering @xcite .
this is a great example of @xmath4pt s ability to give a systematic derivation of electroweak operators that can be used in a variety of light nuclear systems to yield accurate predictions .
i thank the organizers of the ` workshop on microscopic nuclear structure theory ' for their invitation to participate in a stimulating and timely meeting .
i am also grateful to silas beane , manuel malheiro , judith mcgovern , and bira van kolck for a profitable and enjoyable collaboration .
31 van kolck , u. , _ prog . part .
phys . _ , * 43 * , 409 ( 1999 ) . | i summarize two recent applications of chiral perturbation theory to electromagnetic reactions on deuterium : elastic electron - deuteron scattering , and compton scattering on deuterium . both calculations
have now been carried out to three orders in the chiral expansion .
the expansion shows good convergence and is able to reproduce data for @xmath0 mev in @xmath1 and for @xmath2 95 mev in @xmath3 .
these results demonstrate that @xmath4pt can be used to reliably compute operators and wave functions for low - momentum - transfer reactions in light nuclear systems . |
the newly born inter - disciplinar field of neutrino geoscience takes the advantage of the technologies developed by large - volume neutrino experiments and of the achievements of the elementary particle physics in order to study the earth interior with a new probe - geo - neutrinos .
geo - neutrinos are electron antineutrinos released in the decays of radioactive elements with lifetimes comparable with the age of the earth and distributed through the earth s interior .
the radiogenic heat released during the decays of these heat producing elements ( hpe ) is in a well fixed ratio with the total mass of hpe inside the earth .
geo - neutrinos bring to the earth s surface an instant information about the distribution of hpe .
thus , it is , in principle , possible to extract from measured geo - neutrino fluxes several geological information completely unreachable by other means .
these information concern the total abundance and distribution of the hpe inside the earth and thus the determination of the fraction of radiogenic heat contributing to the total surface heat flux .
such a knowledge is of critical importance for understanding complex processes such as the mantle convection , the plate tectonics , the geo - dynamo ( the process of generation of the earth s magnetic field ) , as well as the earth formation itself .
currently , only two large - volume , liquid - scintillator neutrino experiments , kamland in japan and borexino in italy , have been able to measure the geo - neutrino signal .
antineutrinos can interact only through the weak interactions .
thus , the cross - section of the inverse - beta decay detection interaction @xmath0 is very low .
even a typical flux of the order of @xmath1 geo - neutrinos @xmath2 s@xmath3 leads to only a hand - full number of interactions , few or few tens per year with the current - size detectors .
this means , the geo - neutrino experiments must be installed in underground laboratories in order to shield the detector from cosmic radiations .
the aim of the present paper is to review the current status of the neutrino geoscience .
first , in sec . [ sec : geonu ] we describe the radioactive decays of hpe and the geo - neutrino production , the geo - neutrino energy spectra and the impact of the neutrino oscillation phenomenon on the geo - neutrino spectrum and flux . the sec .
[ sec : earth ] is intended to give an overview of the current knowledge of the earth interior . the opened problems to which understanding the geo - neutrino studies can contribute are highlighted .
section [ sec : signal ] sheds light on how the expected geo - neutrino signal can be calculated considering different geological models .
section [ sec : experiments ] describes the kamland and the borexino detectors .
section [ sec : analysis ] describes details of the geo - neutrino analysis : from the detection principles , through the background sources to the most recent experimental results and their geological implications . finally , in sec .
[ sec : future ] we describe the future perspectives of the field of neutrino geoscience and the projects having geo - neutrino measurement among their scientific goals .
today , the earth s radiogenic heat is in almost 99% produced along with the radioactive decays in the chains of @xmath4th ( @xmath5 = 14.0 @xmath6 10@xmath7 year ) , @xmath8u ( @xmath5 = 4.47 @xmath6 10@xmath7 year ) , @xmath9u ( @xmath5 = 0.70 @xmath6 10@xmath7 year ) , and those of the @xmath10k isotope ( @xmath5 = 1.28 @xmath6 10@xmath7 year ) .
the overall decay schemes and the heat released in each of these decays are summarized in the following equations : @xmath11 @xmath12 @xmath13 @xmath14 @xmath15 since the isotopic abundance of @xmath9u is small , the overall contribution of @xmath8u , @xmath4th , and @xmath10k is largely predominant .
in addition , a small fraction ( less than 1% ) of the radiogenic heat is coming from the decays of @xmath16rb ( @xmath5 = 48.1 @xmath17 10@xmath7 year ) , @xmath18la ( @xmath19 10@xmath7 year ) , and @xmath20lu ( @xmath5 = 37.6 @xmath17 10@xmath7 year ) .
neutron - rich nuclides like @xmath8u , @xmath4th , and @xmath9u , made up @xcite by neutron capture reactions during the last stages of massive - stars lives , decay into the lighter and proton - richer nuclides by yielding @xmath21 and @xmath22 particles , see eqs .
[ eq : geo1 ] - [ eq : geo3 ] . during @xmath23 decays , electron antineutrinos ( @xmath24 ) are emitted that carry away in the case of @xmath8u and @xmath4th chains , 8% and 6% , respectively , of the total available energy @xcite . in the computation of the overall @xmath24 energy spectrum of each decay chain , the shapes and rates of all the individual decays has to be included : detailed calculations required to take into account up to @xmath2580 different branches for each chain @xcite .
the most important contributions to the geo - neutrino signal are however those of @xmath26bi and @xmath27pa@xmath28 in the uranium chain and @xmath29bi and @xmath30ac in the thorium chain @xcite .
geo - neutrino spectrum extends up to 3.26mev and the contributions originating from different elements can be distinguished according to their different end - points , i.e. , geo - neutrinos with e @xmath31 2.25mev are produced only in the uranium chain , as shown in fig .
[ fig : geonuspe ] .
we note , that according to geo - chemical studies , @xmath4th is more abundant than @xmath8u and their mass ratio in the bulk earth is expected to be @xmath32(@xmath4th)/@xmath32(@xmath8u ) = 3.9 ( see also sec . [
sec : earth ] ) . because the cross - section of the detection interaction from eq .
[ eq : invbeta ] increases with energy , the ratio of the signals expected in the detector is @xmath33(@xmath4th)/@xmath33(@xmath8u ) = 0.27 .
the @xmath10k nuclides presently contained in the earth were formed during an earlier and more quiet phase of the massive - stars evolution , the so called silicon burning phase @xcite . in this phase , at temperatures higher than 3.5 @xmath17 10@xmath7 k , @xmath22 particles , protons , and neutrons were ejected by photo - disintegration from the nuclei abundant in these stars and were made available for building - up the light nuclei up to and slightly beyond the iron peak ( a = 65 ) .
being a lighter nucleus , the @xmath10k , beyond the @xmath23 decay shown in eq .
[ eq : geo4 ] , has also a sizeable decay branch ( 10.7% ) by electron capture , see eq .
[ eq : geo5 ] . in this case ,
electron neutrinos are emitted but they are not easily observable because they are overwhelmed by the many orders of magnitude more intense solar - neutrino fluxes . luckily , the earth is mostly shining in antineutrinos ; the sun , conversely , is producing energy by light - nuclide fusion reactions and only neutrinos are yielded during such processes .
both the @xmath10k and @xmath9u geo - neutrinos are below the 1.8mev threshold of eq .
[ eq : invbeta ] , as shown in fig .
[ fig : geonuspe ] , and thus , they can not be detected by this process .
however , the elemental abundances ratios are much better known than the absolute abundances .
therefore , by measuring the absolute content of @xmath8u and @xmath4th , also the overall amount of @xmath10k and @xmath9u can be inferred with an improved precision .
geo - neutrinos are emitted and interact as flavor states but they do travel as superposition of mass states and are therefore subject to flavor oscillations . in the approximation @xmath34
@xmath25 @xmath35 @xmath36 @xmath37 , the square - mass differences of mass eigenstates 1 , 2 , and 3 , the survival probability @xmath38 for a @xmath24 in vacuum is : @xmath39 in the earth , the geo - neutrino sources are spread over a vast region compared to the oscillation length : @xmath40 for example , for a @xmath253mev antineutrino , the oscillation length is of @xmath25100 km , small with respect to the earth s radius of @xmath256371 km : the effect of the neutrino oscillation to the total neutrino flux is well averaged , giving an overall survival probability of : @xmath41 according to the neutrino oscillation mixing angles and square - mass differences reported in @xcite , @xmath42 .
while geo - neutrinos propagate through the earth , they feel the potential of electrons and nucleons building - up the surrounding matter .
the charged weak current interactions affect only the electron flavor ( anti)neutrinos . as a consequence
, the hamiltonian for @xmath24 s has an extra term of @xmath43 , where @xmath44 is the electron density .
since the electron density in the earth is not constant and moreover it shows sharp changes in correspondence with boundaries of different earth s layers , the behavior of the survival probability is not trivial and the motion equations have to be solved by numerical tracing .
it has been calculated in @xcite that this so called _ matter effect _ contribution to the average survival probability is an increase of about 2% and the spectral distortion is below 1% .
to conclude , the net effect of flavor oscillations during the geo - neutrino ( @xmath24 ) propagation through the earth is the absolute decrease of the overall flux by @xmath250.55 with a very small spectral distortion , negligible for the precision of the current geo - neutrino experiments .
the earth was created in the process of accretion from undifferentiated material , to which chondritic meteorites are believed to be the closest in composition and structure .
the ca - al rich inclusions in carbonaceous chondrite meteorites up to about a cm in size are the oldest known solid condensates from the hot material of the protoplanetary disk . the age of these fine grained structures was determined based on u - corrected pb - pb dating to be @xmath45 million years @xcite .
thus , these inclusions together with the so called chondrules , another type of inclusions of similar age , provide an upper limit on the age of the earth .
the oldest terrestrial material are zircon inclusions from western australia being at least 4.404 billion years old @xcite .
the bodies with a sufficient mass undergo the process of differentiation , e. g. , a transformation from an homogeneous object to a body with a layered structure .
the metallic core of the earth ( and presumably also of other terrestrial planets ) was the first to differentiate during the first @xmath2530 million years of the life of the solar system , as inferred based on the @xmath46hf - @xmath46w isotope system @xcite .
today , the core has a radius of 2890 km , about 45% of the earth radius and represents less than 10% of the total earth volume . due to the high pressure of about 330gpa
, the inner core with 1220 km radius is solid , despite the high temperature of @xmath255700k , comparable to the temperature of the solar photosphere . from seismologic studies , and namely from the fact that the secondary , transverse / shear waves do not propagate through the so called outer core , we know that it is liquid .
turbulent convection occurs in this liquid metal of low viscosity .
these movements have a crucial role in the process of the generation of the earth magnetic field , so called geo - dynamo .
the magnetic field here is about 25gauss , about 50 times stronger than at the earth s surface .
the chemical composition of the core is inferred indirectly as fe - ni alloy with up to 10% admixture of light elements , most probable being oxygen and/or sulfur . some high - pressure
, high - temperature experiments confirm that potassium enters iron sulfide melts in a strongly temperature - dependent fashion and that @xmath10k could thus serve as a substantial heat source in the core @xcite .
however , other authors show that several geo - chemical arguments are not in favor of such hypothesis @xcite .
geo - neutrinos from @xmath10k have energies below the detection threshold of the current detection method ( see fig .
[ fig : geonuspe ] ) and thus the presence of potassium in the core can not be tested with geo - neutrino studies based on inverse beta on free protons .
other heat producing elements , such as uranium and thorium are lithophile elements and due to their chemical affinity they are quite widely believed not to be present in the core ( in spite of their high density ) . there
exist , however , ideas as that of herndon @xcite suggesting an u - driven georeactor with thermal power @xmath4730tw present in the earth s core and confined in its central part within the radius of about 4 km .
the antineutrinos that would be emitted from such a hypothetical georeactor have , as antineutrinos from the nuclear power plants , energies above the end - point of geo - neutrinos from `` standard '' natural radioactive decays .
antineutrino detection provide thus a sensitive tool to test the georeactor hypothesis .
after the separation of the metallic core , the rest of the earth s volume was composed by a presumably homogeneous primitive mantle built of silicate rocks which subsequently differentiated to the present mantle and crust . above the core mantle boundary ( cmb )
there is a @xmath25200 km thick zone called d
( pronounced d - double prime ) , a seismic discontinuity characterized by a decrease in the gradient of both p ( primary ) and s ( secondary , shear ) wave velocities .
the origin and character of this narrow zone is under discussion and there is no widely accepted model . the lower mantle is about 2000 km thick and extends from the d
zone up to the seismic discontinuity at the depth of 660 km .
this discontinuity does not represent a chemical boundary while a zone of a phase transition and mineral recrystallization . below this zone , in the lower mantle , the dominant mineral phases are the mg - perovskite ( mg@xmath48fe@xmath49)sio@xmath50 , ferropericlase ( mg , fe)o , and ca - perovskite casio@xmath50
. the temperature at the base of the mantle can reach 3700k while at the upper boundary the temperature is about 600k . in spite of such high temperatures , the high lithostatic pressure ( 136gpa at the base )
prevents the melting , since the solidus increases with pressure .
the lower mantle is thus solid , but viscose and undergoes plastic deformation on long time - scales . due to a high temperature gradient and the ability of the mantle to creep ,
there is an ongoing convection in the mantle .
this convection drives the movement of tectonic plates with characteristic velocities of few cm per year .
the convection may be influenced by the mineral recrystallizations occurring at 660 km and 410 km depths , through the density changes and latent heat .
the mantle between these two seismic discontinuities at 410 and 660 km depths is called the transition zone .
this zone consists primarily of peridotite rock with dominant minerals garnet ( mostly pyrop mg@xmath50al@xmath51(sio@xmath52)@xmath50 ) and high - pressure polymorphs of olivine ( mg , fe)@xmath51sio@xmath52 , ringwoodite and wadsleyite below and above cca .
525 km depth , respectively . in the upper mantle above the 410 km depth discontinuity the dominant minerals are olivine , garnet , and pyroxene .
the upper mantle boundary is defined with seismic discontinuity called mohorovii , often referred to as moho .
it s average depth is about 35 km , 5 - 10 km below the oceans and 20 - 90 km below the continents .
the moho lies within the lithosphere , the mechanically defined uppermost earth layer with brittle deformations composed of the crust and the brittle part of the upper mantle , continental lithospheric mantle ( clm ) .
the lithospheric tectonic plates are floating on the more plastic astenosphere entirely composed of the mantle material .
partial melting is a process when solidus and liquidus temperatures are different and are typical for heterogeneous systems as rocks .
the mantle partial melting through geological times lead to the formation of the earth s crust .
typical mantle rocks have a higher magnesium - to - iron ratio and a smaller proportion of silicon and aluminum than the crust .
the crust can be seen as the accumulation of solidified partial liquid , which thanks to its lower density tends to move upwards with respect to denser solid residual .
the lithophile and incompatible elements , such as u and th , tend to concentrate in the liquid phase and thus they do concentrate in the crust .
there are two types of the earth s crust .
the simplest and youngest is the oceanic crust , less than 10 km thick .
it is created by partial melting of the transition - zone mantle along the mid - oceanic ridges on top of the upwelling mantle plumes .
the total length of this submarine mountain range , the so called rift zone , is about 80,000 km .
the age of the oceanic crust is increasing with the perpendicular distance from the rift , symmetrically on both sides .
the oldest large - scale oceanic crust is in the west pacific and north - west atlantic - both are up to 180 - 200 million years old .
however , parts of the eastern mediterranean sea are remnants of the much older tethys ocean , at about 270 million years old .
the typical rock types of the oceanic crust created along the rifts are mid - ocean ridge basalts ( morb ) .
they are relatively enriched in lithophile elements with respect to the mantle from which they have been differentiated but they are much depleted in them with respect to the continental crust .
the typical density of the oceanic crust is about 2.9 g @xmath53 .
the continental crust is thicker , more heterogeneous and older , and has a more complex history with respect to the oceanic crust .
it forms continents and continental shelves covered with shallow seas .
the bulk composition is granitic , more felsic with respect to oceanic crust .
continental crust covers about 40% of the earth surface .
it is much thicker than the oceanic crust , from 20 to 70 km . the average density is 2.7 g @xmath53 , less dense than the oceanic crust and so to the contrary of the oceanic crust , the continental slabs rarely subduct .
therefore , while the subducting oceanic crust gets destroyed and remelted , the continental crust persists . on average
, it has about 2 billion years , while the oldest rock is the acasta gneiss from the continental root ( craton ) in canada is about 4 billion years old .
the continental crust is thickest in the areas of continental collision and compressional forces , where new mountain ranges are created in the process called orogeny , as in the himalayas or in the alps .
there are the three main rock groups building up the continental crust : igneous ( rocks which solidified from a molten magma ( below the surface ) or lava ( on the surface ) ) , sedimentary ( rocks that were created by the deposition of the material as disintegrated older rocks , organic elements etc . ) , and metamorphic ( rocks that recrystallized without melting under the increased temperature and/or pressure conditions ) .
there are several ways in which we can obtain information about the deep earth .
seismology studies the propagation of the p ( primary , longitudinal ) and the s ( secondary , shear , transversal ) waves through the earth and can construct the wave velocities and density profiles of the earth .
it can identify the discontinuities corresponding to mechanical and/or compositional boundaries .
the first order structure of the earth s interior is defined by the 1d seismological profile , called prem : preliminary reference earth model @xcite .
the recent seismic tomography can reveal structures as large low shear velocity provinces ( llsvp ) below africa and central pacific @xcite indicating that mantle could be even compositionally non - homogeneous and that it could be tested via future geo - neutrino projects @xcite .
the chemical composition of the earth is the subject of study of geochemistry .
the direct rock samples are however limited .
the deepest bore - hole ever made is 12 km in kola peninsula in russia .
some volcanic and tectonic processes can bring to the surface samples of deeper origin but often their composition can be altered during the transport .
the pure samples of the lower mantle are practically in - existent . with respect to the mantle
, the composition of the crust is relatively well known .
a comprehensive review of the bulk compositions of the upper , middle , and lower crust were published by rudnick and gao @xcite and huang et al .
@xcite .
the bulk composition of the silicate earth , the so called bulk silicate earth ( bse ) models describe the composition of the primitive mantle , the earth composition after the core separation and before the crust - mantle differentiation .
the estimates of the composition of the present - day mantle can be derived as a difference between the mass abundances predicted by the bse models in the primitive mantle and those observed in the present crust . in this way
, the predictions of the u and th mass abundances in the mantle are made , which are then critical in calculating the predicted geo - neutrino signal , see sec .
[ sec : signal ] .
the refractory elements are those that have high condensation temperatures ; thus , they did condensate from a hot nebula , today form the bulk mass of the terrestrial planets , and are observed in equal proportions in the chondrites .
their contrary are volatile elements with low condensation temperatures and which might have partially escaped from the planet . u and th are refractory elements , while k is moderately volatile .
all u , th , and k are also lithophile ( rock - loving ) elements , which in the goldschmidt geochemical classification means elements tending to stay in the silicate phase ( other categories are siderophile ( metal - loving ) , chalcophile ( ore , chalcogen - loving ) , and atmophile / volatile ) .
the most recent classification of bse models was presented by rmek et al .
@xcite : * _ geochemical bse models : _ these models rely on the fact that the composition of carbonaceous ( ci ) chondrites matches the solar photospheric abundances in refractory lithophile , siderophile , and volatile elements .
these models assume that the ratios of refractory lithophile elements ( rle ) in the bulk silicate earth are the same as in the ci chondrites and in the solar photosphere .
the typical chondritic value of the bulk mass th / u ratio is 3.9 and k / u @xmath25 13,000 .
the absolute rle abundances are inferred from the available crust and upper mantle rock samples .
the theoretical petrological models and melting trends are taken into account in inferring the composition of the original material of the primitive mantle , from which the current rocks were derived in the process of partial melting . among these models
are mcdonough and sun ( 1995 ) @xcite , allgre ( 1995 ) @xcite , hart and zindler ( 1986 ) @xcite , arevalo et al .
( 2009 ) @xcite , and palme and oneill ( 2003 ) @xcite .
the typical u concentration in the bulk silicate earth is about @xmath54ppb . * _ cosmochemical bse models : _ the model of javoy et al .
( 2010 ) @xcite builds the earth from the enstatite chondrites , which show the closest isotopic similarity with mantle rocks and have sufficiently high iron content to explain the metallic core ( similarity in oxidation state ) .
the `` collisional erosion '' model of oneill and palme ( 2008 ) @xcite is covered in this category as well . in this model
, the early enriched crust was lost in the collision of the earth with an external body . in both of these models
the typical bulk u concentration is about 10 - 12 ppb . * _ geodynamical bse models : _ these models are based on the energetics of the mantle convection . considering the current surface heat flux , which depends on the radiogenic heat and the secular cooling , the parametrized convection models require higher contribution of radiogenic heat ( and thus higher u and th abundances ) with respect to geo and cosmochemical models .
the typical bulk u concentration is @xmath55ppb .
the surface heat flux is estimated based on the measurements of temperature gradients along several thousands of drill holes along the globe .
the most recent evaluation of these data leads to the prediction of @xmath56tw predicted by davies and davies ( 2010 ) @xcite , consistent with the estimation of jaupart et al .
( 2007 ) @xcite .
the relative contribution of the radiogenic heat from radioactive decays to this flux ( so called urey ratio ) is not known and this is the key information which can be pinned down by the geo - neutrino measurements .
the geochemical , cosmochemical , and geodynamical models predict the radiogenic heat of 20 @xmath57 4 , 11 @xmath57 2 , 33 @xmath57 3tw and the corresponding urey ratios of about 0.3 , 0.1 , and 0.6 , respectively .
the heat producing elements ( hpe ) predicted by these models are distributed in the crust and in the mantle .
the crustal radiogenic power was recently evaluated by huang et al . @xcite as @xmath58tw . by subtracting this contribution from the total radiogenic heat predicted by different bse models
, the mantle radiogenic power driving the convection and plate tectonics can be as little as 3tw and as much as 23tw . to determine this mantle contribution
is one of the main goals and potentials of neutrino geoscience .
the geo - neutrino signal can be expressed in several ways .
we recall that geo - neutrinos are detected by the inverse beta decay reaction ( see eq .
[ eq : invbeta ] ) in which antineutrino interacts with a target proton .
the most straightforward unit is the normalized event rate , expressed by the so called terrestrial neutrino unit ( tnu ) , defined as the number of interactions detected during one year on a target of @xmath59 protons ( @xmath251kton of liquid scintillator ) and with 100% detection efficiency .
conversion between the signal @xmath33 expressed in tnu and the oscillated , electron flavor flux @xmath60 ( expressed in @xmath61 ) is straightforward @xcite and requires a knowledge of the geo - neutrino energy spectrum and the interaction cross section , which scales with the @xmath24 energy : @xmath62 = 4.07 \cdot \phi ( ^{232}\rm{th } ) \label{eq : tnufluxth}\ ] ] @xmath63 = 12.8 \cdot \phi ( ^{238}\rm{u } ) \label{eq : tnufluxu}\ ] ] in order to calculate the geo - neutrino signal at a certain location on the earth s surface , it is important to know the absolute amount and the distribution of hpe inside the earth . as it was described in sec .
[ sec : earth ] , we know relatively well such information for the earth s crust , but we lack it for the mantle . instead , the bse models , also described in sec .
[ sec : earth ] , predict the total amount of hpe in the silicate earth ( so , excluding the metallic core , in which no hpe are expected ) .
thus , in the geo - neutrino signal predictions , the procedure is as follows .
first , the signal from the crust is calculated .
then , the total mass of the hpe concentrated in the crust is subtracted from the hpe mass predicted by a specific bse model ; the remaining amount of hpe is attributed to be concentrated in the mantle . due to the chemical affinity of hpe , the continental crust is their richest reservoir .
thus , for the experimental sites built on the continental crust , the total geo - neutrino signal is dominated by the crustal component .
it is important to estimate it with the highest possible precision since the mantle contribution can be extracted from the measured signal only after the subtraction of the expected crustal component . the first estimation of the crustal geo - neutrino signal @xcite modeled the crust as a homogeneous , 30 km thick layer .
since then , several much more refined models have been developed . in these models ,
different geochemical and geophysical data are used as input parameters .
the crust is divided in finite volume voxels with surface area of either @xmath64 @xcite , @xmath65 @xcite or , most recently , @xmath66 @xcite .
the oceanic and continental crust are treated separately .
the continental crust is further divided in different layers , as upper , middle , and lower continental crust . on the sites placed on the continental crust ,
a significant part of the crustal signal comes from the area with a radius of few hundreds of km around the detector @xcite .
thus , in a precise estimation of the crustal geo - neutrino signal , it is important to distinguish the contribution from the local crust ( loc ) and the rest of the crust ( roc ) @xcite . in estimating the loc contribution , it is crucial to consider the composition of real rocks surrounding the experimental site , while for the roc contribution it is sufficient to take into account the mean crustal compositions .
borexino and kamland , the only two experiments which have provided geo - neutrino measurements , are placed in very different geological environments .
borexino is placed on a continental crust in central italy .
kamland is situated in japan , in an area with very complicated geological structure around the subduction zone . in table
[ tab : signalexp ] we show the expected geo - neutrino signal for both experiments .
the loc contributions are taken from @xcite .
the calculations are based on six 2@xmath67 tiles around the detector , as shown in fig .
[ fig : kbtil ] .
the loc contribution in borexino , based on a detailed geological study of the lngs area from @xcite , is low , since the area is dominated by dolomitic rock poor in hpe .
the loc contribution in kamland is almost double , since the crustal rocks around the site are rich in hpe @xcite .
the roc contributions shown in table [ tab : signalexp ] are taken from @xcite .
this recent crustal model uses as input several geophysical measurements ( seismology , gravitometry ) and geochemical data as the average compositions of the continental crust @xcite and of the oceanic crust @xcite , as well as several geochemical compilations of deep crustal rocks .
the calculated errors are asymmetric due to the log - normal distributions of hpe elements in rock samples .
the authors of @xcite estimate for the first time the geo - neutrino signal from the continental lithospheric mantle ( clm ) , a relatively thin , rigid portion of the mantle which is a part of the lithosphere ( see also sec . [
sec : earth ] ) .
the mantle contribution to the geo - neutrino signal is associated with a large uncertainty .
the estimation of the mass of hpe in the mantle is model dependent .
the relatively well known mass of hpe elements in the crust has to be subtracted from the total hpe mass predicted by a specific bse model .
since there are several categories of bse models ( see sec .
[ sec : earth ] ) , the estimates of the mass of hpe in the mantle ( and thus of the radiogenic heat from the mantle ) varies by a factor of about 8 @xcite .
in addition , the geo - neutrino signal prediction depends on the distribution of hpe in the mantle , which is unknown . as it was described in sec .
[ sec : earth ] , there are indications of compositional inhomogeneities in the mantle but this is not proved and several authors prefer a mantle with homogeneous composition .
extremes of the expected mantle geo - neutrino signal with a fixed hpe mass can be defined @xcite : * _ homogeneous mantle : _ the case when the hpe are distributed homogeneously in the mantle corresponds to the _ maximal , high geo - neutrino signal_. * _ sunken layer : _ the case when the hpe are concentrated in a limited volume close to the core - mantle boundary corresponds to the _ minimal , low geo - neutrino signal_. * _ depleted mantle + enriched layer ( dm + el ) : _ this is a model of a layered mantle , with the two reservoirs ( dm and el ) in which the hpe are distributed homogeneously .
the total mass of hpe in the dm + el corresponds to a chosen bse model .
there are estimates of the composition of the upper mantle ( dm ) , from which the oceanic crust ( composed of mid - ocean ridge basalts , morb ) has been differentiated @xcite .
since in the process of differentiation the hpe are rather concentrated in the liquid part , the residual mantle remains depleted in hpe .
the measured morb compositions indicate that their source must be in fact depleted in hpe with respect to the rest of the mantle .
the mass fraction of the el is not well defined and in the calculations of rmek et al .
@xcite a 427 km thick el placed above the core - mantle boundary has been used .
an example of the estimation of the mantle signal for borexino and kamland , given in table [ tab : signalexp ] , is taken from @xcite . .expected geo - neutrino signal in borexino and kamland .
details in text . [ cols= " < , < , < " , ]
the kamland result is based on a total live - time of 2991 days , collected between march 2002 and november 2012 . in this 10-year time window
the backgrounds and detector conditions have changed . after the april 2011 earthquake the japanese nuclear energy production was strongly reduced and in particular in the april to june 2012 months all the japanese nuclear reactors were switched off with the only exception of the tomary plant which is in any case quite far ( @xmath25600 km ) from the kamland site .
this reactor - off statistics was extremely helpful to check all the other backgrounds and it is included in the present data sample even if with a reduced fiducial volume ( fv ) .
in fact , because of the contemporary presence of the inner balloon containing the xe loaded scintillator at the detector center , the central portion of the detector was not included in the analysis . the @xmath24 event rate in the kamland detector and in the energy window 0.9 - 2.6mev as a function of time is shown in fig .
[ fig : klrate]-left .
the measured excess of events with respect to the background expectations is constant in time , as highlighted in fig .
[ fig : klrate]-right , and is attributed to the geo - neutrino signal . to extract the neutrino oscillation parameters and the geo - neutrino fluxes , the @xmath24 candidates are analyzed with an unbinned maximum likelihood method incorporating the measured event rates , the energy spectra of prompt candidates and their time variations .
the total energy spectrum of prompt events and the fit results are shown in fig .
[ fig : kbspe]-right . by assuming a chondritic th / u mass ratio of 3.9 , the fit results in @xmath68 geo - neutrino events , corresponding to a total oscillated flux of @xmath69@xmath2 s@xmath3 .
it is easy to demonstrate that given the geo - neutrino energy spectrum , the chondritic mass ratio , and the inverse beta decay cross section , a simple conversion factor exists between the fluxes and the tnu units : 1 tnu = 0.113 @xmath17 10@xmath70 @xmath24 @xmath2s@xmath3 . by taking this factor
we could translate the kamland result to @xmath71tnu . while the precision of the kamland result is mostly affected by the systematic uncertainties arising from the sizeable backgrounds , the extremely low background together with the smaller fiducial mass
( see tables [ tab : bxkl2 ] and [ tab : bxkl3 ] ) makes the statistical error largely predominant in the borexino measurement .
the borexino result , shown in fig .
[ fig : kbspe]-left , refers to the statistics collected from december 2007 to august 2012 .
the levels of background affecting the geo-@xmath72 analysis were almost constant during the whole data taking , the only difference being an increased radon contamination during the test phases of the purification campaigns .
these data periods are not included in the solar neutrino analysis but can be used in the geo - neutrino analysis .
a devoted data selection cuts were adopted to make the increased background level not significant , in particular , an event pulse - shape analysis and an increased energy threshold have been applied for delayed candidates .
the borexino collaboration selected 46 antineutrino candidates ( fig .
[ fig : kbspe]-left ) , among which 33.3 @xmath57 2.4 events were expected from nuclear reactors and 0.70 @xmath57 0.18 from the non-@xmath24 backgrounds . an unbinned maximal likelihood fit of the light - yield spectrum of prompt candidates was performed , with the th / u mass ratio fixed to the chondritic value of 3.9 , and with the number of events from reactor antineutrinos left as a free parameter . as a result ,
the number of observed geo - neutrino events is 14.3 @xmath57 4.4 in ( 3.69 @xmath57 0.16 ) @xmath17 10@xmath73 proton @xmath17 year exposure .
this signal corresponds to @xmath24 fluxes from u and th chains , respectively , of @xmath60(u ) = ( 2.4 @xmath57 0.7 ) @xmath17 10@xmath70@xmath2s@xmath3 and @xmath60(th ) = ( 2.0 @xmath57 0.6 ) @xmath17 10@xmath70@xmath2s@xmath3 and to a total measured normalized rate of ( 38.8 @xmath57 12)tnu .
the measured geo - neutrino signals reported in table [ tab : bxkl3 ] can be compared with the expectations reported in table [ tab : signalexp ] .
the two experiments placed very far form each other have presently measured the geo - neutrino signal with a high statistical significance ( at @xmath25@xmath74 c.l . ) and in a good agreement with the geological expectations .
this is an extremely important point since it is confirming both that the geological models are working properly and that the geo - neutrinos are a reliable tools to investigate the earth structure and composition . in the standard geo - neutrino analysis
, the th / u bulk mass ratio has been assumed to be 3.9 , a value of this ratio observed in ci chondritic meteorites and in the solar photosphere , and , a value assumed by the geo - chemical bse models .
however , this value has not yet been experimentally proven for the bulk earth .
the knowledge of this ratio would be of a great importance in a view of testing the geo - chemical models of the earth formation and evolution .
it is , in principle , possible to measure this ratio with geo - neutrinos , exploiting the different end - points of the energy spectra from u and th chains ( see fig . [
fig : geonuspe ] ) . a mass ratio of @xmath32(th)/@xmath32(u ) = 3.9 corresponds to the signal ratio @xmath33(u)/@xmath33(th ) @xmath253.66 .
both kamland and borexino collaborations attempted an analysis in which they tried to extract the individual u and th contributions by removing the chondritic constrain from the spectral fit . in fig .
[ fig : uth ] , the confidence - level contours from such analyses are shown for borexino ( left ) and for kamland ( right ) .
borexino has observed the central value a @xmath33(u)/@xmath33(th ) of @xmath252.5 while kamland of @xmath2514.5 but they are not in contradiction since the uncertainties are still very large and the results not at all conclusive .
both the best fit values are compatible at less than 1@xmath75 level with the chondritic values . as discussed in sec .
[ sec : earth ] , the principal goal of geo - neutrino measurements is to determine the hpe abundances in the mantle and from that to extract the strictly connected radiogenic power of the earth .
the geo - neutrino fluxes from different reservoirs sum up at a given site , so the mantle contribution can be inferred from the measured signal by subtracting the estimated crustal ( loc + roc ) components ( sec .
[ sec : signal ] ) . considering the expected crustal signals from table [ tab : signalexp ] and the measured geo - neutrino signals from table [ tab : bxkl3 ] , such a simple subtraction results in mantle signals measured by kamland @xmath76 and borexino @xmath77 of : @xmath78 @xmath79 a graphical representation of the different contributions in the measured signals
is shown in fig .
[ fig : kbstrati ] . the kamland result seems to highlight a smaller mantle signal than the borexino one .
such a result pointing towards mantle inhomogeneities is very interesting from a geological point of view , but the error bars are still too large to get statistically significant conclusions .
indeed , recent models predicting geo - neutrino fluxes from the mantle not spherically symmetric have been presented @xcite .
they are based on the hypothesis , indicated by the geophysical data , that the large low shear velocity provinces ( llsvp ) placed at the base of the mantle beneath africa and the pacific represent also compositionally distinct areas . in a particular ,
the tomo model @xcite predicts a mantle signal in borexino site higher by 2% than the average mantle signal while a decrease of 8.5% with respect to the average is expected for kamland .
we have performed a combined analysis of the borexino and kamland data in the hypothesis of a spherically symmetric mantle or a not homogeneous one as predicted by the tomo model .
the @xmath80 profiles for both models are shown in fig .
[ fig : kbdchi2 ] . for the homogeneous mantle
we have obtained the signal @xmath81 of @xmath82 instead , when the borexino and kamland mantle signals have been constrained to the ratio predicted by the tomo model , the mean mantle signal @xmath83 results to be @xmath84 there is an indication for a positive mantle signal but only with a small statistical significance of about 1.5@xmath75 c.l .
the central values are quite in agreement with the expectation shown in table [ tab : signalexp ] .
a slightly higher central value is observed for the tomo model .
we stress again the importance of a detailed knowledge of the local crust composition and thickness in order to deduce the signal coming from the mantle from the measured geo - neutrino fluxes . in fig .
[ fig : kbsra ] , we compare the measured mantle signal @xmath81 from eq .
[ eq : homo ] with the predictions of the three categories of the bse models according to @xcite which we have discussed in sec . [ sec : signal ] , i.e. the geochemical , cosmochemical , and geodynamical ones . for each bse model category ,
four different hpe distributions through the mantle have been considered : a homogeneous model and the three dm + el models with the three different depleted mantle compositions as in @xcite .
all the earth models are still compatible at 2@xmath75 level with the measurement , as shown in fig .
[ fig : kbsra ] , even if the present combined analysis slightly disfavors the geodynamical models .
we remind that these models are based on the assumption that the radiogenic heat has provided the power to sustain the mantle convection over the whole earth story .
it has been recently understood @xcite the importance of the water or water vapor embedded in the crust and mantle to decrease the rock viscosity and so the energy supply required to promote the convection . if this is the case the geodynamical models are going to be reconciled with the geochemical ones .
it is , in principle , possible to extract from the measured geo - neutrino signal the earth s radiogenic heat power .
this procedure is however not straightforward : the geoneutrino flux depends not only on the total mass of hpe in the earth , but also on their distributions , which is model dependent .
the hpe abundances and so the radiogenic heat in the crust are rather well known , as discussed in secs .
[ sec : earth ] and [ sec : signal ] . as the main unknown remains the radiogenic power of the earth s mantle
. figure [ fig : kbtw ] summarizes the analysis we have performed in order to extract the mantle radiogenic heat from the measured geo - neutrino signals .
the geo - neutrino luminosity @xmath85 ( @xmath24 emitted per unit time from a volume unit , so called _ voxel _ ) is related @xcite to the u and th masses @xmath86 contained in the respective volume : @xmath87 where the masses are expressed in units of 10@xmath88 kg , and the luminosity in units of 10@xmath89s@xmath3 .
the measured geo - neutrino signal at a given site can be deduced by summing up the u and th contributions from individual voxels over the whole earth , and by weighting them by the inverse squared - distance ( geometrical flux reduction ) and by the oscillation survival probability .
we have performed such an integration for the mantle contribution to the geo - neutrino signal .
we have varied the u and th abundances ( with a fixed chondritic mass ratio th / u = 3.9 ) in each voxel .
the homogeneous and sunken layer models of the hpe distributions in the mantle ( sec .
[ sec : signal ] ) were taken into account separately .
for each iteration of different u and th abundances and distributions , the total mantle geo - neutrino signal ( taking into account eq .
[ eq : lm ] ) and the u + th radiogenic heat power from the mantle ( considering eq . 4 from @xcite ) can be calculated .
the result is shown in figure [ fig : kbtw ] showing the u + th mantle radiogenic heat power as a function of the measured mantle geo - neutrino signal .
the solid lines represent the sunken - layer model , while the dotted lines the homogeneous mantle . the individual u and th contributions , as well as their sums are shown .
the measured mantle signal @xmath90 from the combined borexino and kamland analysis quoted in eq .
[ eq : homo ] is demonstrated on this plot by the vertical solid ( orange ) line indicating the central value of 7.7tnu while the filled ( light brown ) triangular area corresponds to @xmath91tnu band of 1@xmath75 error .
the central value of @xmath81 = 7.7tnu corresponds to the mantle radiogenic heat from u and th of 7.5 - 10.5tw ( orange double arrow on @xmath92-axis ) , for sunken - layer and homogeneous hpe extreme distributions , respectively .
if the error of the measured mantle geo - neutrino signal is considered ( @xmath91tnu ) , the corresponding interval of possible mantle radiogenic heat is from 2 to 19.5tw , indicated by the black arrow on @xmath92-axis .
the two independent geo - neutrino measurements from the borexino and kamland experiments have opened the door to a new inter - disciplinary field , the neutrino geoscience .
they have shown that we have a new tool for improving our knowledge on the hpe abundances and distributions .
the first attempts of combined analysis has appeared @xcite , showing the importance of multi - site measurements .
the first indication of a geo - neutrino signal from the mantle has emerged .
the present data seem to disfavor the geo - dynamical bse models , in agreement with the recent understanding of the important role of water in the heat transportation engine . these results together with the first attempts to directly measure the th / u ratio are the first examples of geologically relevant outcomes .
but in order to find definitive answers to the questions correlated to the radiogenic heat and hpe abundances , more data are needed . both borexino and
kamland experiments are going on to take data and a new generation of experiments using liquid scintillators is foreseen .
one experimental project , sno+ in canada , is in an advanced construction phase , and a new ambitious project , daya - bay 2 in china , mostly aimed to study the neutrino mass hierarchy , has been approved .
other interesting proposals have been presented , lena at pyhsalmi ( finland ) or frjus ( france ) and hanohano in hawaii . the sno+ experiment in the sudbury mine in canada @xcite , at a depth of 6080 m.w.e .
, is expected to start the data - taking in 2014 - 2015 .
the core of the detector is made of @xmath25780 ton of lab ( linear alkylbenzene ) with the addition of ppo as fluor .
a rate of @xmath2520 geo - neutrinos / year is expected and the ratio of geo - neutrino to reactor @xmath24 events should be around @xmath251.2 .
the site is located on an old continental crust and it contains significant quantities of felsic rocks , which are enriched in u and th . moreover , the crust is particularly thick ( ranging between 44.2 km and 41.4 km ) , approximately 40% thicker than the crust surrounding the gran sasso and kamioka sites . for these reasons ,
a strong loc signal is expected , around 19 tnu .
a very detailed study of the local geology is mandatory to allow the measurement of the mantle signal .
the main goal of the daya bay 2 experiment in china @xcite is to determine the neutrino mass hierarchy .
thanks to a very large mass of 20kton it would detect up to 400 geo - neutrinos per year .
a few percent precision of the total geo - neutrino flux measurement could be theoretically reached within the first couple of years and the individual u and th contributions could be determined as well .
unfortunately , the detector site is placed by purpose very close to the nuclear power plant .
thus , under the normal operating conditions , the reactor @xmath24 flux is huge ( @xmath2540 detected events / day ) .
data interesting for the geo - neutrino studies could be probably taken only in correspondence with reactor maintenance or shutdowns .
lena is a proposal for a huge , 50kton liquid scintillator detector aiming to the geo - neutrino measurement as one of the main scientific goals @xcite .
two experimental sites have been proposed : frjus in france or pyhsalmi in finland . from the point of view of the geo - neutrino study
, the site in finland would be strongly preferable , since frjus is very close to the french nuclear power plants .
lena would detect about 1000 geo - neutrino events per year : a few percent precision on the geo - neutrino flux could be reached within the first few years , an order of magnitude improvement with respect to the current experimental results .
thanks to the large mass , lena would be able to measure the th / u ratio , after 3 years with 10 - 11% precision in pyhsalmi and 20% precision in frjus .
another very interesting project is hanohano @xcite in hawaii , placed on a thin , hpe depleted oceanic crust . the mantle contribution to the total geo - neutrino flux should be dominant , @xmath2575% . a tank of 26 m in diameter and 45 m tall , housing a 10kton liquid scintillator detector , would be placed vertically on a 112 m long barge and deployed in the deep ocean at 3 to 5 km depth .
the possibility to build a recoverable and portable detector is part of the project . a very high geo - neutrino event rate up to about @xmath25100 per year
would be observed with a geo - neutrino to reactor-@xmath24 event rate ratio larger than 10 . in conclusion ,
the new inter - disciplinary field has formed .
the awareness of the potential to study our planet with geo - neutrinos is increasing within both geological and physical scientific communities this is may be the key step in order to promote the new discoveries about the earth and the new projects measuring geo - neutrinos .
the authors declare that there is no conflict of interests regarding the publication of this article .
c. rolfs and w. rodney , `` cauldron in the cosmos , '' _ nuclear astrophysics _ , university of chicago press , 1988 .
g. fiorentini , m. lissia , and f. mantovani , `` geo - neutrinos and earth s interior , '' _ physics reports _ , vol .
117 - 172 , 2007 .
s. enomoto , `` using neutrinos to study the earth : geo - neutrinos , '' in _ talk given at the neutel 2009 conference _ ,
venice , italy , march 2009 .
s. enomoto , `` neutrino geophysics and observation of geo - neutrinos at kamland , '' ph .
d. thesis , tohoku university , japan , 2005 .
fogli et al .
, `` global analysis of neutrino masses , mixings and phases : entering the era of leptonic cp violation searches , '' _ physical review d _ , vol .
1 , article i d 013012 , 10 pages , 2012 .
j. n. connelly et al .
, `` the absolute chronology and thermal processing of solids in the solar protoplanetary disk , '' _ science _ , vol .
6107 , pp .
651 - 655 , 2012 .
s. a. wilde et al . , `` evidence from detrital zircons for the existence of continental crust and oceans on the earth 4.4 gyr ago , '' _ nature _ , vol .
409 , no . 6817 , pp .
175 - 178 , 2001 .
t. kleine et al . , `` rapid accretion and early core formation on asteroids and the terrestrial planets from @xmath46hf - @xmath46w chronometry , '' _ nature _ , vol .
952 - 955 , 2002 .
v. r. murthy , w. van westrenen , and y. fei , `` experimental evidence that potassium is a substantial radioactive heat source in planetary cores , '' _ nature _ , vol .
6936 , pp .
163 - 165 , 2003 .
w. f. mcdonough , `` compositional model for the earth s core , '' in : r.w .
carlson ( ed . ) the mantle and core , _ treatise on geochemistry _ , vol .
547 - 568 , elsevier , oxford , 2003 .
j. m. herndon , `` substructure of the inner core of the earth , '' _ proceedings of the national academy of sciences of the united states of america _ , vol .
646 - 648 , 1996 .
a. m. dziewonski and d. l. anderson , `` preliminary reference earth model , '' _ physics of the earth and planetary interiors _ , vol .
297 - 356 , 1981 .
y. wang and l. wen , `` mapping the geometry and geographic distribution of a very low velocity province at the base of the earth s mantle , '' _ journal of geophysical research _
109 ( b10 ) , article i d b10305 , 18 pages , 2004 .
o. rmek et al .
, `` geophysical and geochemical constraints on geo - neutrino fluxes from earth s mantle , '' _ earth and planetary science letters _ , vol . 361 , pp . 356 - 366 , 2013 .
r. l. rudnick and s. gao , `` composition of the continental crust , '' in the crust , vol.3 , _ treatise on geochemistry _ , edited by r. l. rudnick , pp .
1 - 64 , elsevier , oxford , 2003 . y. huang et al . , `` a reference earth model for the heat producing elements and associated geo - neutrino flux , '' doi : 10.1002/ggge.20129 , 2013 . w. f. mcdonough and s .- s .
sun , `` the composition of the earth , '' _ chemical geology _ , vol .
223 - 253 , 1995 . c. j. allgre , j. p. poirier , e. humler , and a. w. hofmann , `` the chemical composition of the earth , '' _ earth and planetary science letters _
515 - 526 , 1995 .
s. r. hart and a. zindler , `` in search of a bulk - earth composition , '' _ chemical geology _
247 - 267 , 1986 .
r. arevalo , w. f. mcdonough , and m. luong , `` the k / u ratio of the silicate earth : insights into mantle composition , structure and thermal evolution , '' _ earth and planetary science letters _ ,
36 - 369 , 2009 .
h. palme and h. s. c. oneill , `` cosmochemical estimates of mantle composition , '' in : r.w .
carlson ( ed . ) the mantle and core , vol
. 2 , _ treatise of geochemistry _ , elsevier , oxford , pp . 1 - 38 , 2003 . m. javoy et al . , `` the chemical composition of the earth : enstatite chondrite models , '' _ earth and planetary science letters _ , vol .
259 - 268 , 2010 .
h. s. c. oneill and h. palme , `` collisional erosion and the non - chondritic composition of the terrestrial planets , '' _ philosophical transactions of the royal society a : mathematical , physical and engineering sciences _ , vol .
366 , pp . 4205 - 4238 , 2008 . j. h. davies and d. r. davies , `` earth s surface heat flux , '' _ solid earth _
1 , pp . 5 - 24 , 2010 . c. jaupart , s. labrosse , and j. c. mareschal , `` temperatures , heat and energy in the mantle of the earth , '' in : d.j .
stevenson ( ed . ) _ treatise of geophysics _ , elsevier , amsterdam , pp . 1 - 53 , 2007 .
g. l. fogli , e. lisi , a. palazzo , and a. m. rotunno , `` combined analysis of kamland and borexino neutrino signals from th and u decays in the earth s interior , '' _ physical review d _ ,
82 , article i d 093006 , 9 pages , 2010 .
l. m. krauss , s. l. glashow , and d. n. schramm , `` antineutrino astronomy and geophysics , '' _ nature _ , vol .
191 - 198 , 1984 . c. g. rothschild , m. c. chen , and f. p. calaprice , `` antineutrino geophysics with liquid scintillator detectors , '' _ geophysical research letters _ , vol
. 25 , pp .
1083 - 1086 , 1998 .
s. enomoto , e. ohtani , k. inoue , and a. suzuki , `` neutrino geophysics with kamland and future prospects , '' _ earth and planetary science letters _ , vol . 258 , pp . 147 - 159 , 2007 . g. l. fogli , e. lisi , a. palazzo , and a. m. rotunno , `` geo - neutrinos : a systematic approach to uncertainties and correlations , '' _ earth , moon , and planets _ , vol .
111 - 130 , 2006 .
f. mantovani , l. carmignani , g. fiorentini , and m. lissia , `` antineutrinos from earth : a reference model and its uncertainties , '' _ physical review d _ , vol .
69 , article i d 0130011 , 12 pages , 2004 .
g. fiorentini et al .
, `` mantle geo - neutrinos in kamland and borexino , '' _ physical review d _ , vol .
86 , article i d 033004 , 11 pages , 2012 . f. mantovani , `` geo - neutrinos : phenomenology and experimental prospects , '' in _ talk given at the aap11 conference _ , wien , austria , 2011
. f. mantovani , `` geo - neutrinos : combined kamland and borexino analysis , and future , '' in _
talk give at the neutrino geoscience 2013 conference _ , takayama , japan , 2013 .
m. coltorti et al . , ``
u and th content in the central apennines continental crust : a contribution to the determination of the geo - neutrinos flux at lngs , '' _ geochimica et cosmochimica acta _ ,
2271 - 2294 , 2011 .
g. fiorentini , m. lissia , f. mantovani , and r. vannucci , `` how much uranium is in the earth ?
predictions for geo - neutrinos at kamland , '' _ physical review d _ , vol .
72 , article i d 033017 , 11 pages , 2005 .
w. m. white and e. m. klein , `` the oceanic crust , '' in the crust , vol
. 3 , _ treatise on geochemistry _ , edited by r. l. rudnick , elsevier , oxford , 2003 .
arevalo and w. f. mcdonough , `` chemical variations and regional diversity observed in morb , '' _ chemical geology _ , vol .
271 , no . 1 - 2 , pp .
70 - 85 , 2010 . v. j. m. salters and a. stracke , `` composition of the depleted mantle , '' _ geochemistry , geophysics , geosystems _ , vol . 5 , no .
5 , article i d q05b07 , 2004 .
r. k. workman and s. r. hart , `` major and trace element composition of the depleted morb mantle ( dmm ) , '' _ earth and planetary science letters _ , vol .
231 , no . 1 - 2 , pp . 53 - 72 , 2005
. kamland collaboration,``kamland : a liquid scintillator anti - neutrino detector at the kamioka site , '' proposal for us involvement , stanford - hep-98 - 03 , rcns-98 - 15 , jul 1998 .
b. e. berger et al . , ( kamland collaboration ) , `` the kamland full - volume calibration system , '' _ journal of instrumentation _ ,
4 , article i d p04017 , 30 pages , 2009 .
g. alimonti at al . , ( borexino collaboration ) , `` the borexino detector at the laboratori nazionali del gran sasso , '' _ nuclear instruments and methods in physics research section a _ , vol . 600 , pp .
568 - 593 , 2009 .
g. alimonti et al . , ( borexino collaboration ) , `` the liquid handling systems for the borexino solar neutrino detector , '' _ nuclear instruments and methods in physics research section a _ , vol .
58 - 78 , 2009 .
h. back et al . , ( borexino collaboration ) , `` borexino calibrations : hardware , methods , and results , '' _ journal of instrumentation _ , vol . 7 , no . 10 ,
article i d 10018 , 36 pages , 2012 .
k. eguchi et al . , ( kamland collaboration ) , `` first results from kamland : evidence for reactor antineutrino disappearance , '' _ physical review letters _ , vol . 90 , article i d 021802 , 6 pages , 2003 . t. araki et al . , ( kamland collaboration ) , `` measurement of neutrino oscillation with kamland : evidence of spectral distortion , '' _ physical review letters _ ,
94 , article i d 081801 , 5 pages , 2005 .
s. abe et al , ( kamland collaboration ) , `` precision measurement of neutrino oscillation parameters with kamland , '' _ physical review letters _ ,
100 , article i d 221803 , 5 pages , 2008 .
t. araki et al . , ( kamland collaboration ) , `` experimental investigation of geologically produced antineutrinos with kamland , '' nature , vol .
499 - 503 , 2005 .
a. gando et al . , ( kamland collaboration ) , `` partial radiogenic heat model for earth revealed by geo - neutrino measurements , '' _ nature geoscience _ ,
647 - 651 , 2011 .
a. gando et al . ,
( kamland collaboration ) , `` reactor on - off antineutrino measurement with kamland , '' _ physical review d _ ,
vol.88 , article i d 033001 , 10 pages , 2013 .
a. gando et al . , ( kamland collaboration ) , `` measurement of the double-@xmath93 decay half - life of xe with the kamland - zen experiment , '' arxiv : 1201.4664v2 .
g. alimonti et al . , ( borexino collaboration ) , `` ultra - low background measurements in a large volume underground detector , '' _ astroparticle physics _ , vol . 8 , pp.141 - 157 , 1998 . g. alimonti et al .
, `` a large - scale low - background liquid scintillation detector : the counting test facility at gran sasso , '' _ nuclear instruments and methods in physics research section a _ , vol .
411 - 426 , 1998 . c. arpesella et al .
, ( borexino collaboration ) , `` first real time detection of @xmath94be solar neutrinos by borexino , '' _ physics letters b _ , vol .
101 - 108 , 2008 . c. arpesella et al . , ( borexino collaboration ) , `` direct measurement of the @xmath94be solar neutrino flux with 192 days of borexino data , '' _ physical review letters _ , vol
. 101 , article i d 091302 , 6 pages , 2008 .
g. bellini et al . , ( borexino collaboration ) , `` precision measurement of the @xmath94be solar neutrino interaction rate in borexino , '' _ physical review letters _ , vol .
14 , article i d 141302 , 5 pages , 2011 .
g. bellini et al . , ( borexino collaboration ) , `` absence of a day - night asymmetry in the @xmath94be solar neutrino rate in borexino , '' _ physics letters b _ , vol .
22 - 26 , 2012 .
g. bellini et al . , ( borexino collaboration ) , `` first evidence of pep solar neutrinos by direct detection in borexino , '' _ physical review letters _ ,
108 , article i d 051302 , 6 pages , 2012 .
g. bellini et al . , ( borexino collaboration ) , `` measurement of the solar @xmath95b neutrino rate with a liquid scintillator target and 3 mev energy threshold in the borexino detector , '' _ physical review d _ ,
82 , article i d 033006 , 10 pages , 2010 .
g. bellini et al .
, ( borexino collaboration ) , `` observation of geo - neutrinos , '' _ physics letters b _ , vol .
299 - 304 , 2010 .
g. bellini et al . , ( borexino collaboration ) , `` measurement of geo - neutrinos from 1353 days of borexino , '' _ physics letters b _ , vol .
295 - 300 , 2013 .
g. bellini et al . , ( borexino collaboration ) , `` sox : short distance neutrino oscillations with borexino , '' arxiv:1304.7721 accepted for publication by journal of high energy physics .
mueller et al . , `` improved predictions of reactor antineutrino spectra , '' _ physical review c _ , vol .
83 , article i d 054615 , 17 pages , 2011 .
p. huber , `` determination of antineutrino spectra from nuclear reactors , '' _ physical review c _ , vol .
84 , article i d 024617 , 16 pages , 2011 .
g. mention et al .
, `` the reactor antineutrino anomaly , '' _ physical review d _ , vol .
83 , no . 7 , article i d 073006 , 20 pages , 2011 .
b. ricci et al . , `` reactor antineutrinos signal all over the world , '' in _ poster presented at the neutel 2013 conference _ ,
venice , italy , 2013 .
j. w. crowley , `` mantle convection and heat loss , '' in _ talk given at the neutrino geoscience 2013 conference _ , takayama , japan , 2013 .
m. chen , `` geo - neutrinos in sno@xmath96 , '' _ earth , moon and planets _ , vol .
221 - 228 , 2006 .
m. chen , `` sno@xmath96 , '' in _ talk given at the neutrino geoscience 2013 conference _ , takayama , japan , 2013 .
z. wang , `` update of dayabay ii jiangmen anti - neutrino observation spectrometer , '' in _ talk given at the neutrino geoscience 2013 conference _ , takayama , japan , 2013 .
m. wurm et al . , `` the next - generation liquid - scintillator neutrino observatory lena , '' _ astroparticle physics _ , vol .
685 - 732 , 2012 .
j. g. learned , s.t .
dye , and s. pakvasa , `` hanohano : a deep ocean anti - neutrino detector for unique neutrino physics and geophysics studies , '' arxiv : 0810.4975 , 2008 . | geo - neutrinos , electron antineutrinos from natural radioactive decays inside the earth , bring to the surface unique information about our planet .
the new techniques in neutrino detection opened a door into a completely new inter - disciplinary field of neutrino geoscience .
we give here a broad geological introduction highlighting the points where the geo - neutrino measurements can give substantial new insights .
the status - of - art of this field is overviewed , including a description of the latest experimental results from kamland and borexino experiments and their first geological implications .
we performed a new combined borexino and kamland analysis in terms of the extraction of the mantle geo - neutrino signal and the limits on the earth s radiogenic heat power .
the perspectives and the future projects having geo - neutrinos among their scientific goals are also discussed . |
the properties of galaxies in the local universe appear to depend strongly on their environment .
this issue was highlighted by @xcite with the so - called morphology density relation .
namely , massive ellipticals and s0 galaxies are preferentially found in crowded regions , such as cluster cores , while spiral and disk galaxies prefer less dense environments .
it is also well established that a rather tight correlation exists between morphological type and level of star formation ( sf ) activity . in general ,
disk galaxies tend to have a higher sf rate ( sfr ) than spheroidal systems .
recently , the nature of this relation has been carefully studied up to @xmath3 by @xcite , through the use of the deep _ herschel _ surveys and well - calibrated complementary sfr indicators on the major blank fields , such as the goods ( great observatories origin deep survey ; @xcite ) and cosmos ( cosmological evolution survey ; @xcite ) fields .
this work highlights that the so - called main sequence ( ms ) of star - forming systems , observed at any redshift ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , corresponds to a well defined sequence of disk galaxies , while spheroidal systems tend to live below the ms . in light of this finding , the sfr
density relation can be seen as an alternative way to study the morphology density relation .
a galaxy s sfr is on average anti - correlated with the galaxy density in the local universe @xcite .
in fact , highly - star - forming galaxies are mostly found in low - density environments , while the cores of massive clusters are full of massive , early - type galaxies dominated by old stellar populations .
however , the way this relation evolves with redshift is still a matter of debate .
it has been argued that as we approach the epoch at which early - type galaxies form the bulk of their stars at @xmath4 ( e.g. * ? ? ?
* ) , the sfr
density should progressively reverse , such that high - density regions host highly star - forming galaxies at earlier cosmic time .
@xcite and @xcite observe this reversal already at @xmath5 in the goods field and the deep2 galaxy redshift survey , respectively . using _
pacs ( photodetecting array camera and spectrometer , @xcite ) data , @xcite detect the reversal only for high - mass galaxies .
according to the authors , this is due to high - mass galaxies being more likely to host active galactic nuclei ( agn ) . since agn exhibit a slightly higher sfr with respect to galaxies of the same stellar mass @xcite ,
agn hosts tend to be star - forming ( see also @xcite ) . on the other hand , @xcite , @xcite and @xcite
find no reversal in the cosmos field , arguing that the reversal , if any , must occur at @xmath6 .
the aforementioned studies , use different sfr indicators . @xcite and
@xcite convert the [ oii ] emission line flux into a sfr , while @xcite , @xcite , and @xcite use _ spitzer _ mips 24 data to measure the sf activity of their galaxy sample .
in addition , @xcite complement the estimates of sfr derived from the 24 flux with those from ultra - violet ( uv ) emission .
all of these estimators can be heavily affected by dust extinction uncertainties , by agn contamination , and/or by metallicity ( e.g. @xcite ) .
these problems can be overcome by measuring the sfr from the far - infrared ( ir ) luminosity , as done in @xcite .
pacs data cover the wavelength range at which the bulk of the uv light is re - emitted by dust , at least up to @xmath7 @xcite .
this enables an accurate estimate of the sfr and avoids possible contamination by agn emission , more common in the mid - ir spectral range @xcite .
also the definition of the environment estimated via the local galaxy density is somewhat arbitrary . indeed ,
several works measure the distance to the nth nearest neighbour ( e.g. * ? ? ?
this method is strongly dependent on n : small values probe high - density regions better though they smooth the low - density ones , while high values of n could wash out the information on over - densities when the number of galaxies in a given halo is less than n @xcite .
other authors measure the density of neighbours within a fixed co - moving volume centred on each galaxy ( e.g. * ? ? ?
* ; * ? ? ?
all of these methods rely on the assumption that the local number density of galaxies is a good representation of the environment . however , if the environment is defined as the halo mass of the parent halo to which the galaxy belongs , this is not necessarily the case . indeed , a filament ( interconnecting `` nodes '' of the same large scale structure )
, the outskirts of a massive galaxy cluster , and the core of a galaxy group could exhibit the same galaxy density , even being sites of quite different physical processes ( on multiple scales these environments can be separated , see e.g. @xcite ) .
further complication is added by the interplay of mass and density . according to @xcite , mass and galaxy density
are coupled , with the high - mass galaxies segregated in the densest environments .
this relation was already in place at @xmath8 @xcite .
therefore , the evidence for a clear sfr density trend could be due to a different contribution of massive and less massive galaxies favouring different density regimes . in order to shed light on the relation between sfr , density , and halo mass , we take advantage of the combination of the deepest available _ spitzer _ and _ herschel _ surveys of the extended chandra deep field south ( ecdfs ) and the goods - south and -north fields ( goods - s and goods - n , respectively ) , observed in the pacs evolutionary probe ( pep , @xcite ) and goods-_herschel _ @xcite surveys .
the combined goods data from these two surveys are described in @xcite . in this work
we use a spectroscopic selected sample as already done in ( * ? ? ?
* z13 hereafter ) .
we first study the sfr density relation up to @xmath9 in its standard definition , by estimating the local galaxy density parameter . in the second part of the paper
, we propose an alternative definition of the sfr
environment relation : we distinguish between galaxy group members , `` filament - like '' environments and galaxies that are isolated or more likely associated with lower mass haloes . for this analysis
, we use the galaxy group sample studied in z13 .
in addition , we try to break the mass density ( environment ) degeneracy , by studying the location of group galaxies in the sfr@xmath10 ( @xmath10 ) plane as a function of environment up to @xmath0 .
this paper is organized as follows : in section [ dataset ] we briefly describe our dataset and analysis . in sec .
[ sfr_density ] we present our results and we discuss them in sec . [ discussion ] . eventually , we draw our conclusions in sec .
[ summary_conclusions ] . throughout our analysis
we adopt a @xcite initial mass function ( imf ) and the following cosmological parameters : @xmath11 , @xmath12 and @xmath13 .
in z13 we create a clean irac ( infrared array camera , @xcite ) 3.6 @xmath14 m selected galaxy sample in the ecdfs and goods fields .
this sample includes only galaxies with a spectroscopic redshift and is drawn from the galaxy catalogues of @xcite , @xcite , and @xcite , in the ecdfs , the goods - s and the goods - n field , respectively .
the group sample studied in z13 also includes the x - ray groups identified in the cosmos field by @xcite , and @xcite , and employs the group membership defined by @xcite .
however , given the rather low spectroscopic completeness in the cosmos field ( 40% in the @xmath10 range of interest , see z13 for a complete discussion ) , this region is not included in our current analysis .
indeed , it is not possible to reliably estimate the local galaxy density parameter on the basis of the pure spectroscopic data .
the use of both spectroscopic and photometric redshifts , as done in @xcite , is preferable in the cosmos field , where the sampling rate is spatially very inhomogeneous .
thus , since the ecdfs and the goods fields show an extremely high spectroscopic completeness ( 60 - 80% in @xmath10 ) , we prefer to restrict our analysis to these regions .
we measure the sfr by using the deepest available _ spitzer _ multiband imaging photometer ( mips ) 24 @xmath14 m data combined with the deepest _ herschel _ pacs 100 and 160 @xmath14 m data . in order to overcome any blending issue , the _ spitzer _ and _ herschel _ flux densities are derived with a point - spread - function - fitting analysis guided by the position of sources detected in deep irac images ( see * ? ? ?
* ; * ? ? ?
this method solves a large part of the blending issues encountered ( see results of dedicated monte - carlo simulations in @xcite ) and provides a straightforward association between irac , mips and pacs sources .
furthermore , even if in high density regions the prior psf - fitting method does not solve all blending issues , it should still provide reliable estimates of the total infrared fluxes of these clustered regions and , thus , of their total sfr activity .
the sfr is estimated with the use of the ir templates of @xcite . for sources undetected in pacs and with only mips detections
, we use the `` main sequence '' template , which turns out to provide the most accurate estimate of the sfr from mid - ir data . in order to complement the sfr derived from ir data ( available for the bulk of the star - forming population ) with the sfr of the low - star - forming or rather inactive galaxies ( i.e. undetected in the mid- and far - ir surveys )
, we measure the sfr via multi - wavelength sed fitting by using _ le phare _
( photometric analysis for redshift estimations ; @xcite ) and the @xcite library .
for this purpose we use the aforementioned multi - wavelength photometric catalogues @xcite .
z13 provide a careful calibration of the sfr derived via sed fitting with respect to the more reliable sfr derived from ir data .
we find consistent estimates of sfr though the scatter is quite large , ranging from 0.5 to 0.7 dex depending on the redshift range .
the sed fitting technique is also useful for estimating stellar masses .
the comparison of our estimates with those derived from the same catalogues via different methods and/or templates shows that we can accurately estimate @xmath10 within a factor of two ( see z13 for more details ) .
the spectroscopic data used for the construction of the density field and the dynamical analysis of the galaxy group sample are taken from a collection of publicly available high - quality spectroscopic redshifts in the ecdfs ( @xcite , see z13 for further details about the combination of the different catalogues ) .
the spectroscopic catalogue of the goods - n field is taken from @xcite . as a function of redshift for all groups considered in our sample .
filled circles represent the x - ray mass estimates , while empty circles show the dynamical mass estimates .
we highlight in blue the ecdfs sample and in red the goods groups . ]
all the blank fields considered in our analysis are observed extensively in the x - ray with @xmath15 and xmm-@xmath16 . the x - ray data reduction and the creation of the x - ray group catalogues
are explained in detail in @xcite and in finoguenov et al .
( in prep . ) .
as explained in z13 , we select a sub - sample of x - ray selected groups with clear optical ( spectroscopic ) identification ( we do not include groups with more than one redshift peak of similar strength along the line of sight ) , without close companions that might affect the membership determination , and with at least 10 members , to reliably estimate the velocity dispersion and the membership .
this selection leads to a sample of 22 x - ray detected groups in the ecdfs and 2 groups in the goods
- n field .
we also consider a large scale structure spectroscopically confirmed at @xmath0 by @xcite .
[ fig : m200_z ] shows the group mass ( where @xmath17 ) is defined as @xmath18 , where @xmath19 is the radius at which the density of a cluster is equal to @xmath20 times the critical density of the universe ( @xmath21 ) ] estimates as a function of redshift for the ecdfs ( in blue ) and goods ( in red ) fields , respectively .
we also show the dynamical mass estimates for the groups in the goods fields from @xcite .
the dynamical analysis of each structure is based on spectroscopic data . for details on the dynamical analysis and group membership , see @xcite and z13 . in order to follow the evolution of the relation between sfr and environment , we divide our galaxy sample into four redshift bins , @xmath22 , @xmath23 , @xmath24 , and @xmath25 , according to the redshift distribution of our group sample .
we note that the last redshift bin is populated only by the structure at @xmath0 @xcite .
this is a likely super - group or a cluster in formation as suggested by the x - ray emission from different extended sources in the structure ( finoguenov et al . in prep . ) . when we analyse the sfr
environment relation by distinguishing group members from systems in other environments , we consider , in each redshift bin , all group galaxies together as members of a composite group .
this is done to increase the statistics of group galaxies which otherwise would be too low when considering individual systems . to limit the selection effects and at the same time to control the different levels of spectroscopic completeness in different redshift bins ( see e.g. fig . 5 in z13
) , we apply a common stellar mass cut at @xmath26 .
this mass cut corresponds to an irac 3.6 @xmath14 m apparent magnitude brighter than the 5@xmath27 detection limit in each considered field up to @xmath28 , enabling a high spectroscopic completeness .
moreover , the considered mass range is still dominated by sources with mips and/or pacs detections , in other words with robust sfr estimates .
the uncertainties due to the spectroscopic incompleteness of our galaxy sample is evaluated with dedicated monte carlo simulations based on the mock catalogues of @xcite drawn from the millennium simulation @xcite .
the key ingredient for building a reliable density field is very high and spatially uniform spectroscopic coverage .
this is reached in the ecdfs ( see @xcite ) and in the goods fields ( see @xcite ) , for which we reconstruct the density around each galaxy up to @xmath28 .
we use a method similar to @xcite to compute the projected local galaxy density , @xmath29 , around each spectroscopically - confirmed galaxy with @xmath30 .
we count all galaxies located inside a cylinder of radius @xmath31 , within a fixed velocity interval around each galaxy of @xmath32 , about ten times the typical velocity dispersion of galaxy groups ( @xmath33 @xmath34 ) , and above a redshift dependent mass limit ( @xmath35 ) .
given the spectroscopic completeness as a function of @xmath10 in the four redshift bins considered in our analysis ( see z13 ) , we choose as a cut the @xmath10 value where the 40 - 50% completeness limit is reached in each redshift bin : @xmath36 at @xmath37 , @xmath38 at @xmath39 , @xmath40 at @xmath41 and @xmath42 at @xmath43 .
this does not lead to a different density field definition as a function of redshift bin , but only to a more robust density estimate in the bins where the spectroscopic completeness is still very high at low masses .
indeed , only the absolute value of the density parameter changes , but the relative difference between high and low density regions is kept the same with respect to the choice of a fixed @xmath44 at any redshift .
the density field obtained with the chosen @xmath35 , rather than that at a fixed mass cut of @xmath44 , allows us to distinguish between galaxies residing in dark matter haloes of different masses .
however , the density fields obtained with a lower mass cut show , as expected , higher values and a slightly higher accuracy in distinguishing between galaxies located in parent haloes of different masses .
we estimate that , on average , the projected density obtained with a mass cut of @xmath36 at @xmath37 is a factor of 7 higher than that at @xmath42 . at @xmath39 ,
a mass cut of @xmath38 leads to a density a factor of 5 higher than the cut at lower @xmath10 , while at @xmath45 the density with a mass cut of @xmath40 is a factor of 2.5 higher .
these calibrations are discussed in depth in a dedicated forthcoming paper ( popesso et al .
, in prep . ) .
a more physical definition of the density field would require a mass cut which takes into account the evolution of the characteristic magnitude of the stellar mass function .
in our case this would translate to selecting only galaxies at masses larger than @xmath46 at any redshift @xcite , given the restriction imposed by the completeness level of the galaxy sample in the higher redshift bins .
this would imply that at lower redshifts we would select only galaxies with @xmath47 , limiting in a significant way the statistics for defining the density field .
thus , distinguishing between galaxies residing in parent haloes of different masses would be inefficient .
a simple exercise on the mock catalogues of @xcite reveals that this density definition would be able to distinguish only isolated galaxies from galaxies in the core of massive clusters and it would provide the same density for galaxies in haloes with masses ranging from @xmath48 to @xmath49 . in order to consider the effect of spectroscopic incompleteness , we correct the density @xmath29 by accounting for the possibly missing galaxies .
we consider , for each galaxy , the cylinder along the line of sight , with radius of 0.75 mpc at the redshift of the considered source , and with redshift limits @xmath50 and @xmath51 equal to the limits of the redshift bin to which the source belongs .
the spectroscopic completeness is given by the number of sources with spectroscopic redshifts divided by the total number of galaxies , considering only sources with @xmath52 .
since the redshift bins are more than 10 times larger than the error on the photometric redshift , this uncertainty is only marginally affecting our completeness estimate .
we correct for incompleteness by dividing @xmath29 by this ratio . in order to test the reliability of our density estimate , we measure , with the same method , the density field in 100 randomly - extracted catalogues from the mock catalogues of @xcite .
we compare the density obtained in this way with that measured in the parent light - cone mock catalogues , free of selection biases . in order to simulate the photometric redshift uncertainty , before estimating the incompleteness correction
, we assign a random error in the range @xmath53 to the redshift of the parent mock catalogue galaxies , where @xmath54 is the photometric redshift error provided in the photometric catalogues .
we find a very good agreement between the original density estimated in the @xcite mock catalogues and the local density retrieved with our method ( fig . [ comp ] ) .
we also use this approach for estimating the error per density bin as the dispersion of @xmath55 .
nicely separates the group - dominant regime from the field - dominant regime .
indeed , 75% of field galaxies are found at densities below this threshold and 92% of group galaxies above that . ]
our density definition takes advantage of the high level of mass segregation observed in the local universe @xcite and at least up to @xmath1 .
since we estimate the density of rather massive galaxies around each system , our density estimator should be able to better distinguish between high - density regions , generally dominated by massive galaxies , from low density regions , more populated by low - mass systems . indeed , fig .
[ density ] shows that our method is able to nicely isolate galaxies identified as group spectroscopic members ( red histogram ) from isolated galaxies ( the peak below @xmath56 ) .
a similar figure is shown in ( * ? ? ?
* their fig .
11 ) based on the third - nearest neighbour density estimator .
the comparison of the two figures shows that our density estimator is more efficient in distinguishing isolated systems from galaxy group members .
in fact , although groups occupy the highest density bins in ( * ? ? ?
* their fig .
11 ) , they are not clearly isolated from field galaxies as in our case .
we first build the sfr density relation by studying the statistical correlation between the sfr and density parameters , as usually done in the literature .
this lets us compare our results with previous works . as a second approach we use a dynamical definition of environment by differentiating among massive bound structures , less - massive bound or unbound structures and relatively isolated galaxies .
we follow the evolution of the relation in both cases up to @xmath57 and we test and compare our results with the predictions of simulations . fig .
[ sfr_density_relation ] shows the sfr
density relation for all galaxies with @xmath58 in four redshift bins .
the errors in fig .
[ sfr_density_relation ] are derived from our error analysis using the mock catalogues of @xcite , as explained in sec .
[ mock_errors ] .
we find a significant anti - correlation up to @xmath59 , confirmed by the spearman test at @xmath60 confidence level . at @xmath61
we find an anti - correlation but with lower significance ( @xmath62 ) . in the highest redshift bin , comprising the @xcite large scale structure
, we do not find any significant anti - correlation ( @xmath63 significance level ) .
thus , we can exclude with high confidence level ( from the spearman test ) any positive correlation in the last two redshift bins as claimed in previous works ( e.g. * ? ? ?
* ; * ? ? ?
we only observe a progressive flattening towards higher redshifts , but no reversal of the relation . in different redshift bins ( solid lines ) .
the dashed line represents the sfr density relation at @xmath37 for all galaxies with @xmath64 .
errors are derived using the mock catalogues of @xcite , as explained in the text . ]
the shapes of the relations shown in fig .
[ sfr_density_relation ] are noisy and not even linear in log - log space .
thus , it can not be easily fit by a simple fitting function . in order to quantify the steepness of the relation
, we simply estimate the ratio between the mean sfr at densities below and above the median local galaxy density @xmath29 .
below @xmath65 , where we see an anti - correlation , although with different significances depending on the redshift bin , the mean sfr in low density regions spans a range of 1.4 - 2.1 times the mean sfr in high density regions . in the highest redshift bin
we do not observe a significant difference between the sfr in low and high density regions .
what is the role of group galaxies in shaping the relations ? in order to check this , we remove from the sample all galaxies dynamically associated with either extended x - ray emitting sources or the structure at @xmath0 .
we also remove all galaxies associated with extended x - ray emitting sources not included in the final group sample .
in the two lowest redshift bins , the significance of the anti - correlation decreases much below the @xmath60 level . in the highest two redshift bins we exclude both a reversal and any sign of anti - correlation .
this clearly shows the dominant role of group environments in shaping the sfr density anti - correlation observed in the local universe and at intermediate redshift . in different redshift bins ( solid lines ) .
errors are derived using the mock catalogues of @xcite , as explained in the text .
the dashed line represents the @xmath10density relation at @xmath37 for all galaxies with @xmath64 .
the normalisation of the dashed line is artificially increased to higher value to make it close to the blue solid line and make the comparison easier . ] in principle , a prominent mass segregation together with a high fraction of low - star - forming galaxies ( typical of the group and cluster environment ) could easily lead to the sfr
density anti - correlation observed at low and intermediate redshift .
[ mass_density ] shows the @xmath10density relation for the same sample of galaxies in the four redshift bins . we do not measure a strong mass segregation in the galaxy sample used for the sfr
density relation analysis .
the first redshift bin exhibits a slightly different behaviour with respect to the relation in the other redshift bins .
however , given the large errors ( for their computation see sec .
[ mock_errors ] ) , we can not draw a definitive conclusion on the @xmath10density trend .
the spearman test confirms only a mild level of mass segregation at @xmath39 .
thus , the sfr density anti - correlation observed in the first and second redshift bins is not caused by mass segregation . in the local universe , mass segregation
is observed with high significance by @xcite on the basis of a large sample of sloan digital sky survey ( sdss , @xcite ) galaxies .
are our results at odds with previous findings ?
the main difference with respect to @xcite is the mass cut applied to our sample . indeed , for spectroscopic completeness issues
, we are considering only massive galaxies , i.e with @xmath66 . the dashed blue line in fig .
[ mass_density ] shows the mass density relation obtained after applying a mass cut of @xmath67 in the lowest redshift bin .
this analysis is possible without strong biases only at low redshift where the spectroscopic completeness is rather high even at low stellar masses ( see z13 ) .
the spearman test reveals a mild positive correlation .
the absence of a stronger correlation , as found e.g. in @xcite , could be due to the lack of many massive spectroscopically - detected galaxies at low redshift ( see fig .
5 of z13 ) , since in ecdfs this type of galaxies was targeted for spectroscopy only at high redshift ( e.g. * ? ? ?
we point out that we observe an even more significant ( according to the spearman test ) sfr density anti - correlation ( blue dashed line in fig . [ sfr_density_relation ] ) in the lowest redshift bin after applying the lower mass cut .
this probably indicates that , in a broader mass regime , mass segregation enhances the significance of the sfr
density relation in the parent sample . in order to take into account all possible biases inherent in our spectroscopic selection , we study the sfr
density relation in a simulated universe .
we analyse the sfr
density relation obtained using the @xcite mock catalogues ( 5 different light cones ) by applying our definition of local galaxy density .
we observe an anti - correlation in all redshift bins ( @xmath68 significance ) .
thus , our results ( fig . [ sfr_density_relation ] ) are at least qualitatively in agreement with the prediction of @xcite , except in the highest redshift bin .
however , we point out that the sfr density relations measured using the mock catalogues are observed in an area of sky that is a magnitude larger than the ecdfs and the goods - n regions .
thus , the mock catalogues sample a much broader range of densities due to the presence of massive clusters , while our dataset comprises only groups . in order to check the effect of cosmic variance
when using a rather small area , we estimate the sfr
density relation in 1000 different regions of the @xcite light - cones with areas similar to the sum of the ecdfs and goods - n areas . after running a spearman test
, we detect an anti - correlation with at least @xmath60 significance in all regions in the two redshifts bins below @xmath69 . at @xmath70
we measure an anti - correlation in 98% of the cases and at higher redshift in 70% of the cases .
the non - correlation in the observed @xmath71 redshift bin could be due to the low probability of finding massive large scale structures in such a small area and at high redshift in the @xmath72cdm cosmology .
it could also be due to larger errors on environment washing out the signal ( see e.g. @xcite).thus , cosmic variance could considerably affect the significance of an anti - correlation . in order to simulate the spectroscopic completeness in the ecdfs and goods regions , we randomly extract a sub - sample of galaxies from the millennium mock catalogues mimicking the spectroscopic completeness of the observed data .
we extract randomly a percentage of galaxies consistent with the spectroscopic completeness in one of the available photometric bands and for each magnitude bin of our galaxy sample .
this procedure randomly extracts 100 different catalogues that nicely reproduce the selection function of our sample ( see z13 for further details ) .
we point out that while the galaxy mock catalogues of the millennium simulation provide a rather good representation of the local universe , at higher redshifts ( @xmath73 ) they fail to reproduce the correct distribution of star - forming galaxies in the sfr-@xmath10 plane .
indeed , @xcite find that at @xmath41 the galaxy sfr is under - estimated , on average , by a factor of two , at fixed @xmath10 , with respect to the observed values . by performing the same exercise with our dataset
, we find that this under - estimate ranges by factors of 2.5 to 3 at @xmath43 .
however , this does not represent a problem with our approach .
indeed , the aim of this analysis is to understand what is the bias introduced by a selection function similar `` in relative terms '' to the spectroscopic selection function of our dataset .
thus , for our needs , it is sufficient that the randomly extracted mock catalogues reproduce the same bias in selecting , on average , the same percentage of most star - forming and most massive galaxies of the parent sample , as shown in z13 .
the bias of our analysis is estimated by comparing the results obtained with and without our galaxy sampling .
since the under - estimate of the sfr or the @xmath10 of high redshift galaxies is common to both , biased and unbiased , samples , it does not affect the result of this comparative analysis .
we also stress that the aim of this analysis is not to provide correction factors for our observational results but a way to interpret our results taking into account possible biases introduced by the spectroscopic selection function . in order to account for both the spectroscopic completeness and the cosmic variance , we repeat the exercise performed with the complete @xcite mock catalogues by extracting 1000 different regions with areas similar to the sum of the ecdfs and goods - n areas in the incomplete mock catalogues .
we estimate the sfr density relation in each region as done on the real dataset . the probability of non - correlation increases to @xmath745% in lower redshift bins , 12% at @xmath70 and 45% at @xmath43 .
this suggests that small areas ( thus cosmic variance ) , in addition to spectroscopic incompleteness , could hide a possible anti - correlation in the highest redshift bin or reduce the significance of the anti - correlation in the lower redshift bins .
this last exercise allows us to quantify the possible bias in the estimate of the sfr
density relation due to our spectroscopic selection .
we measure the mean sfr by using the same binning in density in the incomplete and in the original complete catalogues . in this way
, we can compute the residual @xmath75 , where @xmath76 is the mean sfr estimated in the incomplete catalogue at the given density bin , and @xmath77 is the mean sfr estimated in the complete @xcite mock catalogues at the same density bin .
we estimate @xmath78 in 1000 sky regions , extracted from 5 light cones , with the area similar to the sum of the ecdfs and goods - n area , as explained above .
we estimate the mean and the dispersion of the @xmath78 distribution in each density bin .
the mean indicates whether there is any bias in the spectroscopic selection that leads to an over- or under - estimate of the mean sfr per density bin .
the dispersion provides the error on the mean sfr per bin . ) and the sfr from the complete ones ( @xmath79 ) versus density for all four redshift bins used in this work .
we do not find any bias in the slope of the sfr
density relation of fig .
[ sfr_density_relation ] as confirmed by the spearman test . ] as shown in fig .
[ sfr_density_nobias ] , the @xmath80 derived from the incomplete mock catalogues is on average a factor of 2 - 4 ( depending on the redshift bin ) larger than the `` true '' one obtained from the complete @xcite mock catalogues .
thus , the incompleteness leads to a large over - estimate of the mean sfr in each density bin .
this is easily understandable , since the simulated spectroscopic selection favours highly - star - forming galaxies ( see z13 ) .
however , the ratio of the observed and true mean sfr is constant as a function of local galaxy density and is of the same order at any redshift .
this implies that using our dataset we are likely over - estimating the mean observed sfr in the same way at any density without biasing the slope of the relation .
thus , our estimate of the sfr
density relation is rather robust despite the spectroscopic incompleteness .
we use the dispersion estimated with this procedure to define the errors on the observed sfr density relation at the corresponding density .
this is possible because we find a correspondence between galaxy density field and galaxy parent halo mass both in the observed and simulated datasets , at least in the group regime .
the final point of our environmental approach focuses on understanding the disagreement between our findings and previous works claiming a reversal of the sfr
density relation at @xmath1 .
the fairest comparison is with @xcite and @xcite , since our dataset includes the sky regions covered by their dataset . at @xmath5 .
galaxies in the ecdfs and goods - n fields are shown in orange and green , respectively .
the open symbols connected by a dotted line show the sfr
density relation of @xcite for goods - north . ]
[ sfr_density_z1 ] shows the sfr
density relation at @xmath45 for the ecdfs and goods - n regions separately . in the goods - n region
we observe an anti - correlation between sfr and density with high significance , as confirmed by the spearman test .
we do not observe any relation between sfr and density in the ecdfs region which contains only a very poor group at @xmath81 , differently from the goods - n region that comprises , in the same redshift bin , two very massive groups ( @xmath82 ) .
errors are estimated as in sec .
[ mock_errors ] .
[ sfr_density_z1 ] also shows the relation of @xcite for the goods - n region with open symbols connected by a dotted line .
we can compare our results with @xcite only qualitatively , since the definitions of the density parameter and of the galaxy sample differ considerably .
in fact , @xcite include all galaxies with hst - acs @xmath83 mag without any mass cut .
given the broad redshift range considered ( @xmath45 ) , this apparent magnitude cut corresponds to a difference of 0.75 mag from the lowest to the highest redshift limit , introducing a bias with respect to our physical stellar mass selection ( see also @xcite ) .
this could explain the offset between our sfr
density relation and that of @xcite .
@xcite find a positive correlation in goods - n up to the point in which the sfr reaches its maximum and then a rapid decline at higher density .
they refer to this as a `` reversal '' of the sfr density relation .
even if we do not detect any reversal , we point out that the trend we observe for the goods - n field has a shape similar to that of @xcite
. the analysis of the sfr density relation of @xcite is based on the estimate of the mean sfr per density bin , rather than on statistical tests such as the spearman test used in this work .
in addition , the errors estimated in @xcite seem to be under - estimated with respect to ours .
indeed , elbaz et al . ( 2007 ) use a bootstrap technique that can not take into consideration the effect of cosmic variance due to the relatively small fields considered in the analysis .
@xcite show that the use of pacs data provides a big advantage ( with respect to the mips data ) in measuring the unbiased sfr of agn hosts , whose sfr could be enhanced with respect to non - active galaxies of similar @xmath10 .
thus , given the high fraction of agn ( 17% ) measured at least in the highly star - forming population of the goods - s and goods - n fields , @xcite conclude that the reversal of the sfr observed by @xcite could be due to a bias introduced by the sfr of agn host galaxies measured with mips data . in building the sfr
density relation , we are including all galaxies above @xmath84 with sfr much below the lirg limit used by @xcite .
taking advantage of the agn sample of @xcite for the goods - n region and the agn sample of @xcite for the ecdfs , constructed with similar criteria and x - ray flux limits , we investigate whether agn can bias our sample .
we observe an agn fraction of 3 - 5% in the ecdfs and goods - n region above our mass cut .
this fraction is much lower with respect to the work of @xcite , who show that the fraction of agn is much higher in highly ir luminous galaxies .
since we include galaxies spanning a wide range in sfr , the agn fraction is diluted in our sample .
if we remove the agn from our sample , the significance of the sfr density relation does not change at all , in agreement with @xcite . we conclude that the previously observed reversal of the sfr density relation at @xmath1 is most likely due to a combination of different effects : the galaxy sample selection , a rather high fraction of agn in the selected sample and a possibly biased definition of the density parameter , which can hide a redshift dependence .
in addition , we point out that the significance of this reversal is probably due to an under - estimate of the error on the mean sfr , since the cosmic variance is neglected . as shown in fig .
[ density ] , on the right of the dashed green line , where @xmath85 of group galaxies are located , there is still a large number of galaxies at densities comparable to groups but not associated with any extended emitting source identified by the x - ray catalogue of finoguenov et al .
( in prep . ) .
those galaxies are likely located in unbound large scale structures , such as filaments , or in dark matter haloes of lower mass with respect to the detection limits of the cdfs ( chandra deep field south ) 4ms @xcite .
if the relative vicinity to other galaxies is the main driver in quenching the galaxy sf , we should not observe any difference in the level of sf activity between galaxies showing the same local galaxy density .
if , instead , processes related to the dark matter halo play a stronger role , we should observe a difference in the level of sf activity between group galaxies and systems at high density but not related to massive dark matter haloes . to check this issue
, we investigate the sfr
density relation with a novel `` dynamical '' approach .
we distinguish between galaxies in three different environments : a ) group members , as identified via dynamical analysis , b ) `` filament - like '' galaxies identified as systems at the same density as group galaxies but not associated with any of the extended x - ray sources or to the @xcite structure , c ) isolated galaxies with local galaxy density @xmath86 @xmath87 ( on the left - hand side of the green line of fig . [ density ] ) , i.e. where we find a low fraction of group galaxies ( 8% ) .
we build a new version of the sfr
density relation by comparing the mean sfr in the three environments for all galaxies with @xmath88 .
this method allows us to isolate the contribution of groups with halo mass @xmath89 in the sfr density relation .
the left panel of fig .
[ fig : plots_xbin1 ] , shows the sfr
density relation according to our new definition .
we see a strong evolution with redshift of the mean sfr in groups ( within @xmath90 ) with respect to the other two environments . indeed , as shown in the right panel of fig .
[ fig : plots_xbin1 ] , the ratio of @xmath91 is strongly evolving and it shows that the higher the redshift the lower the difference between the level of sf activity in groups and that in the field .
this is consistent with the significance of the sfr
density ( in its standard definition ) anti - correlation decreasing with redshift ( see sec . [ sfr_density_environmental ] ) .
the right panel of fig .
[ fig : plots_xbin1 ] also shows the ratio between @xmath91 of groups and `` filament - like '' galaxies .
although the errors are large , it is possible to appreciate how the ratio increases with redshift , with the sf activity of group galaxies being twice that of `` filament - like '' galaxies at @xmath0 .
if this trend were real , the structure at @xmath9 would provide some hints of the enhancement of the sf activity in groups with respect to filaments .
however , since the errors are quite large we can not draw any definitive conclusion .
the evolution of the sf activity in different environments allows us to better understand the traditional sfr density relation .
in fact , the mix of galaxies in different environments , but at the same densities , hides the strong evolution observed in the left panel of fig .
[ fig : plots_xbin1 ] .
our results also suggest that quenching processes related to a massive dark matter halo must play a decisive role in the strong evolution of the sf activity of group members with respect to galaxies in other environments . ) , `` filament - like '' and field galaxies .
data points are artificially shifted for clarity .
errors are derived using the mock catalogues of @xcite , as explained in the text .
the dashed line represents the mean stellar mass for the different environments at @xmath37 for all galaxies with @xmath64 .
the normalisation of the dashed line is artificially increased to higher value to make it close to the blue solid line only for comparison . ]
we also check if the strong evolution of the newly defined sfr density relation depends on a similar evolution of the @xmath10density relation , using the same approach .
[ fig : plots_xbin2 ] shows the @xmath10density relation in the usual four redshift bins according to our novel dynamical definition .
the large errors do not allow us to see any strong mass segregation .
as in fig .
[ mass_density ] , the lowest redshift bin exhibits a different behaviour with respect to the relation at higher redshifts .
a mass cut at @xmath64 ( dashed line and open symbols ) allows us to highlight , once again , our spectroscopic bias on the lack of massive galaxies at low redshift . with the same mass cut , the @xmath80 appears lower with respect to that derived with a higher mass cut at @xmath92 ( dashed line and open symbols in fig .
[ fig : plots_xbin1 ] ) , as expected by the ms evolution ( e.g. * ? ? ?
* ; * ? ? ?
thus , we conclude that , even with this approach , the strong difference between groups and low - density regime observed at @xmath93 is likely not ascribable to a strong mass segregation . for completeness , we also analysed the evolution of the ssfr
density relation .
this relation evolves in the same way as the sfr
density relation , since the mass density relation is only slightly evolving . the errors on the mean sfr for group ,
`` filament - like '' and field galaxies are estimated in a similar way as in sec .
[ mock_errors ] .
we randomly extract 100 catalogues ( 1000 regions with an area equal to the sum of the ecdfs and goods - n regions ) in which we identify all haloes with masses between @xmath94 and all their members .
this information is obtained by linking the mock catalogues of @xcite to the parent halo properties provided by the `` friends - of - friends '' algorithm @xcite and the @xcite semi - analytic model tables of the millennium database . in the same regions , we define the `` filament - like '' galaxies in the mock catalogues as the ones at the same density of the group galaxies but belonging to haloes with masses below @xmath95 .
finally , field galaxies are defined as sources with densities below the threshold in the real dataset .
we measure the mean sfr for group , `` filament - like '' and field galaxies ( @xmath96 , @xmath97 and @xmath98 , respectively ) by using the galaxy members of each respective environment as in the observational dataset .
we measure in the same way the mean galaxy sfr ( @xmath99 ) for each population in the original ( bias - free ) @xcite mock catalogues .
we estimate , then , the difference @xmath100 for each population , where @xmath101 is the mean sfr of the given population in the @xmath102 region .
the dispersion of the distribution of the residual @xmath103 provides the error on our mean sfr .
this error takes into account the bias due to incompleteness , the cosmic variance ( due to the fact that we are considering small areas of the sky ) and the uncertainty in the mean due to a limited number of galaxies per redshift bin .
the bias introduced by the spectroscopic selection leads to an over - estimate of the mean sfr by the same amount , as expected , as in the case of the `` environmental approach '' .
the same over - estimate is observed in each of the three populations .
this is due to our assumption of a spatially - uniform sampling rate as the one guaranteed by the spectroscopic coverage of the ecdfs and goods - n fields .
the errors on the mean @xmath10 are estimated with the same procedure used for retrieving the errors on the mean sfr for each population . in this section
we analyse the location of group , `` filament - like '' and low density ( field ) galaxies in the sfr@xmath10 plane .
this is done to identify the causes for the strong evolution of the sfr
density relation defined according to our `` dynamical '' definition . as already mentioned ,
@xcite , @xcite , @xcite and several other authors find a well defined sequence of star - forming galaxies in the sfr@xmath10 plane from @xmath107 to @xmath108 .
the relation shows a rather small scatter of 0.2 - 0.4 dex .
the region below the main sequence is populated by quiescent galaxies ( qg ) in a scattered cloud , while only a small fraction ( 2% ) of outliers is found to be located above ( by a factor of 4 ) the ms in the starburst region ( see * ? ? ?
@xcite suggest that the same set of physical processes governs the sf activity in galaxies on this smooth sequence .
if `` mass quenching '' ( e.g. * ? ? ?
* ) is the dominant mechanism for moving a galaxy off of the ms , the location of star - forming galaxies in high density regions should not be different from that of the bulk of the star - forming galaxies in other environments .
conversely , if the environment plays a role in the evolution of the galaxy s sf activity , the position of the group galaxies along or across the ms should be different with respect to the bulk of the star - forming galaxies .
to shed light on this topic , we analyse the position of group , `` filament - like '' and field galaxies with respect to the ms in the ecdfs and goods regions and in four redshift bins . in other words
, we follow the behaviour of different environments defined in our dynamical approach ( see section [ sfr_density_dynamical ] ) in the sfr@xmath10 plane . since the ms is well studied in the literature ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , and the goal of this work is not to fit this relation again , we use the best - fit relations available in the literature for the considered redshift bins . when no fit is available for a specific redshift bin , we interpolate the best - fit relations of the two closest redshift bins .
[ fig : sfr_m ] shows the sfr@xmath10 planes for the different redshift bins and above the mass threshold considered in this work . in each plot , the grey filled circles show the field galaxy population , while green and blue filled circles represent the `` filament - like '' and group galaxy , respectively .
we highlight with red empty circles all galaxies detected in the ir bands ( @xmath109 ) by pacs and mips . on the other hand ,
the sfrs derived via the sed fitting technique are useful for defining the cloud of quiescent or low - star - forming galaxies below the ms .
the issue with this sfr estimator is that it can create artificial discreet trails when plotting sfr and @xmath10 .
however , we point out that this does not affect our study . in the lowest redshift bin
only a few galaxies populate the sfr@xmath10 plane .
this reflects the choice of giving a higher priority to spectroscopic targets like massive galaxies at high redshift ( see z13 and @xcite ) . in order to define the ms in the four redshift bins , we use the equations already used in the literature which best represent our data . in the @xmath110 redshift bin ,
we use a ms fit of ( * ? ? ?
* their eq .
5 ) based on sdss star - forming galaxies at @xmath107 . at @xmath111 we do not find a fit in the literature , thus , we interpolate the ms relation of @xcite at @xmath112 based on sdss galaxies and that of ( * ? ? ?
* their eq .
4 ) at @xmath5 based on spitzer mips detected galaxies .
the latter equation is used for the @xmath113 bin , with an offset of @xmath114 . as for the second redshift bin ,
the ms relation is not available in the literature for @xmath115 .
thus , we interpolate between the ( * ? ? ?
* their eq .
4 ) ms relation at @xmath1 and the ms relation at @xmath6 of @xcite based on uv data . in all cases
we find a rather good agreement between our field galaxy distribution and the best - fit relations , with the mean of the distribution peaked at @xmath116 in the @xmath105 residual at all redshifts ( left panel of fig .
[ fig : ms_evolution ] ) .
we define @xmath117 as the residual of the @xmath104 relation , where @xmath118 is the observed galaxy sfr and @xmath119 is the sfr predicted by the ms best - fit .
we estimate @xmath120 for all the galaxies with mass above our mass cut ( @xmath121 ) and belonging to the three different environments in the usual redshift bins . at this point
we identify and quantify the difference between the location across the ms of group galaxies in each bin with respect to the low density and `` filament - like '' galaxies .
at all redshifts , the distribution of the @xmath105 residuals shows a bimodal distribution : the gaussian representing the ms location with a peak around 0 , and a tail of quiescent / low - star - forming galaxies at low negative values of @xmath105 .
this distribution is reminiscent of the bimodal behaviour of the @xmath122 galaxy colour distribution observed by @xcite in the sdss galaxy sample .
following the example of @xcite , we identify the minimum value of the valley between the ms gaussian and the peak of the broader quiescent / low - star - forming galaxies distribution .
at all redshifts , the value @xmath123 turns out to be the best separation between the two galaxy populations .
since the observed scatter of the ms at any redshift varies between 0.2 - 0.4 dex @xcite , the limit at @xmath123 should be consistent with a @xmath60 cut from the best - fit ms relation .
we use this value to separate ms galaxies from quiescent / low - star - forming galaxies in the three considered environments .
we measure the mean difference in sfr ( @xmath105 ) from the ms location of the galaxy population in each environment , selecting only normally star forming galaxies in the range @xmath124 . by definition
, @xmath105 should be consistent with 0 for the bulk of the ms galaxies .
thus , the mean of our gaussian distribution centred around 0 confirms that our choice of the ms relation represents well the mean of the normally star forming galaxies within our sample ( but note the slope might not be so well represented , see the end of section [ sec : ms_qg_evolution ] ) .
we stress once again that given the depth of the pacs and _ spitzer _ mips observations of the ecdfs and goods fields , the ms is fully sampled ( 80% ) by ir - derived sfrs with very small ( 10% ) uncertainties .
the sed fitting derived sfrs populate the region below the ms at @xmath125 , where we measure the quiescent galaxy fraction , @xmath106 .
thus , our estimate of the @xmath105 should not be affected by the large error ( 0.5 - 0.6 dex ) in the determination of the sfr via sed fitting ( see z13 for details ) .
we estimate the error in @xmath105 ( @xmath106 ) with the same approach used in sec .
[ mock_errors ] .
we use the usual 1000 regions with an area equal to the sum of the ecdfs and goods - n regions with a simulated spectroscopic incompleteness similar to our dataset .
we identify group , `` filament - like '' and field galaxies in each region as explained in the error analysis of previous section .
our aim is to apply the same technique used to analyse the real dataset .
thus , we need the residual @xmath105 with respect to the ms relation to measure the mean distance from the ms at @xmath126 . however , the evolution of the ms predicted by the @xcite catalogues is different from the one observed at the highest redshift considered in our work ( see also @xcite ) . indeed ,
simulated star - forming galaxies , in particular at high redshift , tend to be less star - forming than in observations . thus , the location of the ms using the mock catalogues at @xmath73 tend to be below the observed ms in the same redshift bin . in order to cope with this problem we change the normalisation of the observed ms relation keeping the observed slope . fitting the simulated
ms provides similar results . in each area
we measure the mean distance from the ms at @xmath124 as is done in the real data ( @xmath127 ) .
we follow the same procedure in the original and complete @xcite mock catalogues by measuring @xmath128 .
we measure , then , the difference @xmath129 for each population , where @xmath130 is the residual of the considered population in the @xmath102 region .
the dispersion of the distribution of the residual @xmath131 provides the error on the observed @xmath132 . as in the previous case ,
this error takes into account the bias due to incompleteness , the cosmic variance and the uncertainty in the measure of the mean due to a limited number of galaxies per redshift bin .
we apply the same technique to estimate the @xmath106 error in each of the three populations at different redshift .
the left panel of fig .
[ fig : ms_evolution ] shows the evolution of the @xmath133 for the ms galaxies in low density regions ( grey stars and line ) , `` filament - like '' environments ( green stars and line ) and groups ( blue stars and line ) up to @xmath0 . in the first two redshift bins , the @xmath134 of the star forming group galaxies is systematically below 0 . at @xmath135
the star - forming group galaxies are perfectly on sequence , consistently with the lower density environments .
moreover , the `` filament - like '' ms galaxies appear to be placed between the low density environment and the group galaxies .
this result shows , for the first time , that at least below @xmath59 the sf activity in star - forming group galaxies is lower than in the bulk of the star - forming galaxies . here
, we show that a certain amount of pre - processing ( galaxies being pre - processed in groups before entering clusters , @xcite ) happens even before star - forming galaxies enter the group environment .
in fact , some quenching is already in place when galaxies fall along filaments or lower mass groups that could eventually merge to form more massive structures .
thus , the speed of the evolution of the sf activity in star - forming galaxies depends , at least since @xmath8 , on the galaxy environment . for completeness of the analysis
, we also investigate the evolution of the galaxy - type mix for each environment .
the galaxy - type mix is expressed through the fraction of quiescent galaxies , @xmath106 .
as already mentioned , we define as qg all those systems with @xmath136 , i.e. all the sources in the cloud below the ms .
the right panel of fig .
[ fig : ms_evolution ] shows the evolution of @xmath106 in the three environments . low density ( grey stars and line ) and `` filament - like '' galaxies ( green stars and line )
exhibit the same galaxy - type mix at any redshift and no evolution is observed in these environments at least in the mass range considered in our analysis .
the galaxy - type mix in groups exhibits a higher @xmath106 with respect to the other two environments , at any redshift .
we note that the first redshift bin is affected by the spectroscopic selection function of our sample .
the evolution of @xmath106 is stronger in groups that in the other environments .
in particular , we note that at @xmath137 the fraction of quiescent galaxies is twice the mean fraction observed at high redshift .
the two panels of fig .
[ fig : ms_evolution ] show two different aspects of the role of environment in the evolution of galaxy sf activity .
some degree of partial quenching is observed as suggested by the `` environmental gradient '' as a function of distance from the ms . on the other hand
, the right panel shows that the density is not responsible for the different galaxy - type mix . indeed the group- and `` filament - like '' regimes cover , by definition , the same range of local galaxy density .
the main difference is that , in the group regime , galaxies likely belong to a massive ( @xmath138 , see z13 for the sample mass distribution ) bound dark matter halo , while in the `` filament - like '' regime galaxies likely belong to unbound structures , such as filaments , or lower mass haloes .
thus , the different evolution of the galaxy - type mix of the two environments , similar in projected density but not in dynamical properties , indicates that the galaxy - type mix is connected indicates that a high @xmath106 requires a massive parent dark matter halo rather than simply requiring an over density of galaxies . in order to verify that our result is robust
, we perform some sanity checks .
this choice is driven by the distribution of galaxies in the @xmath104 plane .
in fact , massive galaxies lie mainly below the ms ( as defined , see fig . [
fig : sfr_m ] ) , with the locus of galaxies apparently steepening towards lower sfr ( see e.g * ? ? ?
* ; * ? ? ?
thus , a non - zero @xmath139 could result from an environmental dependent distribution of mass coupled with a ms relation which does not represent the data at all masses .
the kolmogorov - smirnov ( ks ) test reveals that the mass distributions of star forming galaxies in different environments are consistent with one another in all redshift bins , except the second one .
this could be a problem since a more significant population of the most massive galaxies is expected in groups compared to the field . to check how our findings are affected by this issue
, we investigate the evolution of @xmath139 and @xmath106 in two different stellar mass bins : @xmath140 and @xmath141 ( fig .
[ fig : ms_evolution_mbin ] ) .
reassuringly , we find results consistent with fig . [
fig : ms_evolution ] .
we note that the lowest redshift bin is populated by only a few galaxies and so we focus on higher redshifts . in general ,
in the low - mass bin we observe a similar trend in @xmath139 and @xmath106 as in the total sample , while in the high - mass bin @xmath139 is systematically below 0 ( bottom left panel of fig .
[ fig : ms_evolution_mbin ] ) .
this illustrates that massive star - forming galaxies are not well represented by a linear ms relation ( cf .
however , since the evolution of @xmath139 for the other environments is similar to the total sample , we conclude that the relative offset observed in fig .
[ fig : ms_evolution ] for `` filament - like '' and group galaxies is real .
also the evolution of @xmath106 remains similar to the total sample in both stellar mass bins .
even though it is less populated than the low mass bin , the high - mass bin better highlights the different quenching in groups , `` filament - like '' environment and field .
we have investigated the sfr density relation using two different approaches .
one , more traditional , studies the relation between sfr and density for all galaxies , while the other ( the `` dynamical '' approach ) isolates the contribution of group galaxies with respect to other environments in the sfr density relation .
our results show that the sfr
density relation has progressively lower significance towards high redshift , but it does not reverse at @xmath5 .
in addition , a careful analysis of the biases due to the spectroscopic selection leads to the conclusion that we can also not exclude an anti - correlation at @xmath142 . the observed sfr
density anti - correlation at @xmath93 is not simply ascribable to mass segregation ( most massive galaxies are generally passive galaxies , low - mass galaxies are , on average , star - forming ) .
indeed , we observe only a mild mass segregation in any redshift bin .
our results seem to be at odds with @xcite who find strong mass segregation at least in the local universe . at higher
redshift the effect has never been thoroughly analysed , except for the results of @xcite and @xcite who showed that already at @xmath8 mass and galaxy density are coupled with the most massive galaxies segregated in the most dense environment .
we must note that our stellar mass cut ( @xmath143 ) is rather high .
indeed , a lower mass cut ( @xmath144 ) in the lowest redshift bin leads to a stronger mass segregation , although of small amplitude .
this would be in agreement with the recent finding of @xcite , who observe mass segregation within 10 for groups , by only considering low - mass galaxies .
given the flattening of the sfr
density relation observed after excluding group galaxies from the sample , we conclude that group members are mostly responsible for the observed anti - correlation at @xmath93 .
thus , galaxies living in relatively massive dark matter haloes must have a suppressed mean sfr with respect to the field , at least up to @xmath137 .
this is confirmed by the sfr
density relation analysed with our `` dynamical '' approach .
one of the most striking findings in our analysis is the lack of reversal of the sfr
density relation at @xmath1 .
this result is at odds with recent findings .
in particular , @xcite and @xcite observe the reversal of the sfr
density relation at @xmath5 in the goods and the deep2 fields , respectively , using a spectroscopically defined density parameter .
we have extensively compared our analysis with that of @xcite and @xcite , since our dataset includes the sky regions analysed in their dataset ( fig .
[ sfr_density_z1 ] ) .
in particular , we have considered the possibility that the fraction of agn could affect the sfr estimate .
in fact , since the fraction of agn is found to be higher in groups as the redshift increases @xcite and the sfr of agn host galaxies could be enhanced with respect to non active galaxies of similar stellar mass @xcite , we use pacs data to measure , without biases , their sfr @xcite .
this can not be done using mips data or the [ oii ] doublet . removing the agn from our sample
does not affect the significance of the sfr density relation .
this could be due to cosmic variance , since the fraction of agn present in the goods - s field is higher ( 17% in the highly star - forming galaxies , @xcite ) with respect to ecdfs and goods - n ( 3 - 5% ) .
finally , we consider the possibility that the density definition itself could be responsible for the differences we observe .
in fact , as explained in section [ sec : galaxy_density ] , our density estimate is based on a stellar mass cut .
@xcite show that the density definition adopted by @xcite , based on a galaxy apparent magnitude cut ( @xmath83 mag ) , could lead to a strong redshift bias .
thus , we conclude that the previously observed reversal of the sfr density relations is most likely due to the combination of different effects : the galaxy sample selection , high fraction of agn and a possibly biased definition of the density parameter , which can hide a redshift dependence .
our results are , instead , in agreement with @xcite , who find no dependence of the sfr and lirg fraction on environment , arguing that the reversal , if any , must occur at @xmath2 .
according to @xcite the reversal found by @xcite and @xcite might be due to the contribution of galaxies at lower stellar mass and sfr comprised in @xcite and @xcite galaxy sample .
however , since we consider a wide range of sfr and @xmath10 , we disagree with this conclusion .
the advantage of the @xcite study is the use of cosmos data , which , due to its wide field , is less affected by cosmic variance ( although @xcite is the least impacted , covering a larger volume spread over 4 distinct fields ) . on the other hand , it must be noted that the authors use sources with both spectroscopic and photometric redshifts to define the density field .
this could dilute any over - density present in the field .
our approach is , therefore , more rigorous in this sense . in our analysis
we estimated the sfr using both ir data and multi - wavelength sed fitting .
the latter is used for all galaxies undetected in the ir bands , i.e. below the ms .
this sfr estimator gives a larger uncertainty with respect to the ir data , as shown in z13 .
however , in case of an under - estimate of the @xmath145 , i.e. for all galaxies that would be on the ms , we should have an ir detection , thus a more robust sfr estimate . on the other hand , in case @xmath145 over - estimates the real value , the signal could be washed out @xcite .
for example , any anti - correlation at @xmath146 could be hidden by the errors .
in fact , as shown also by @xcite , the uncertainties on the sfr(sed ) increase as a function of redshift .
we discuss our possible biases by using the mock catalogues of @xcite . in all cases ,
we measure a @xmath80 higher than the one predicted by the simulations .
this comparison also assures that we are not suffering from any bias in the slope of the sfr density relation and , thus , that our results are robust .
moreover , we note that in some cases the cosmic variance ( or the big uncertainty in the @xmath145 ) could wash out the signal in the sfr
density anti - relation in the highest redshift bins .
the use of the standard approach for the study of the sfr density relation , can be ineffective if the local galaxy density is not directly connected to the sf activity , due either to mass segregation or to sf quenching processes linked to galaxy - galaxy interactions .
for this reason , we analyse the sfr
density relation with a novel `` dynamical '' approach .
this technique allows us to separate the contribution to the highest galaxy density bins of groups and `` filament - like '' galaxies .
this is not possible in a more classical `` environmental approach '' ( although see @xcite ) .
our results show that the bulk of the sf is quenched in groups .
this is what drives the trend of the sfr
density relation .
the `` filament - like '' environment has a slower evolution in sf compared to the groups , thus the density ( galaxy - galaxy interaction ) itself can not be responsible for the bulk of the quenching .
we note that we use a fairly large velocity window to compute densities , much larger than the typical velocity differences at which galaxy - galaxy interactions are effective .
this means that there can be a certain dynamic range in the efficiency of our density estimates . in the x - ray groups , galaxies have typically smaller velocity differences than chance projections , but larger differences than close pairs .
both the latter types contribute to `` filament - like '' environments .
this result is not necessarily inconsistent with z13 . in z13 , we show that the sf activity of galaxies is not affected by the local environment of groups , but here we find that it does depend on the global environment . in other words , the level of sf activity is generally low in groups ( with respect to the other environments ) even if it is independent from the group - centric distance ( and from the density , as discussed in z13 ) . rather than being inconsistent with z13 ,
this strengthens our results .
in fact , we show , once again , that the density is not responsible for the bulk of the quenching ( although we can not exclude that some quenching is happening for galaxy - galaxy interactions ) , but that processes related to a massive dark matter halo are more effective .
the high sfr in the `` filament - like '' galaxies at low redshift , more consistent with the field than with the group @xmath80 , is in agreement with the recent finding of @xcite and @xcite .
they show that the filament around the super - cluster a1763 hosts the highest fraction of ir - emitting galaxies .
similarly , the `` filament - like '' region contains the highest total sfr per unit galaxy .
our findings are also consistent with those of @xcite , who have used optical data to discover an enhanced star - forming activity among galaxies associated with filaments in the nearby pisces - cetus super - cluster . in order to investigate the cause of the strong evolution of sf activity in our sample
, we have studied the position of group , `` filament - like '' and field star - forming galaxies with respect to the ms galaxy population @xcite .
many works focus mainly on the study of the ms in field galaxies .
for example , according to @xcite , this sequence suggests that the same small set of physical processes governs the sf activity in galaxies . thus ,
if galaxy evolution is driven mainly by their @xmath147 , there should be no difference among ms star - forming galaxies , regardless of their environment .
on the other hand , if a galaxy depends on the environment in which it lies , a group member should have a different level of sf with respect to the bulk of star - forming galaxies on the ms .
this last point reflects our main result from the analysis of the @xmath104 relation .
in particular , we have studied , for the first time , the location of galaxies in different environments on the @xmath104 plane .
the evolution of @xmath134 shows that , at least below @xmath59 , the sf activity in group galaxies is quenched with respect to the bulk of star - forming galaxies . at earlier epochs , group , `` filament - like '' , and field galaxies have comparable sf .
interestingly , the density seems to play a role in the distance from the ms , since the filaments represent a somewhat intermediate environment in the evolution of @xmath134 .
therefore , we show , with high significance , that the speed of the evolution of sf activity in star - forming galaxies depends , at least since @xmath8 , on the galaxy environment , defined according to our `` dynamical '' approach .
in addition , we find that the fraction of qg , @xmath106 , evolves faster in groups with respect to both filaments and field .
the latter two environments show a similar evolution of @xmath106 .
this confirms that quenching processes related to a rather massive dark matter halo ( @xmath148 ) are more efficient than those associated with a generally dense region .
thus , strangulation @xcite , ram pressure stripping @xcite and harassment @xcite , are likely to be much more effective than the simple galaxy - galaxy interaction .
our result is in contrast with the analysis of @xcite .
these authors argue that central star - forming galaxies are equivalent to field galaxies .
they claim that there is no difference in the main sequence relation of central and satellite galaxies .
however , it is not clear how they discern between star - forming and passive galaxies .
in fact , over their whole study , @xcite use the red / blue galaxy dichotomy to distinguish between passive / star - forming galaxies respectively . as shown by @xcite ,
about 30% of the sdss red sequence galaxies , identified in the colour - magnitude diagram , lie on the ms , which is also populated by green valley galaxies @xcite .
@xcite confirm this point , finding two different ms for blue and star - forming galaxies .
this implies that selecting blue galaxies misses many red , dusty , star - forming sources .
moreover , @xcite use the catalogue of ( * ? ? ?
* based on a friends - of - friends algorithm ) to explore the properties of group galaxies .
as we already mentioned , optical selection of group is much more prone to projection effects than x - ray selection .
in fact , the optically selected group catalogues do not contain virialized , relatively high - mass , x - ray emitting groups and include many more low - mass and unvirialized groups , as well as some pure projections . finally ,
( part of ) optically selected groups could be classified as `` filament - like '' galaxies , implying an @xmath106 evolution similar to that of field galaxies .
recently , @xcite , computing the sfr from uv emission for nearby group galaxies , found a ms broadly consistent but flatter than the ms of field galaxies at the same redshift .
they argue that a flattening could be expected if the sfr of low - mass galaxies is suppressed in groups . at a median mass of @xmath149 ,
their ms predicts a mean ssfr which is @xmath150 lower than that expected for the field .
this could be consistent with our results , although we do not cover the same mass range .
our findings support the pre - processing scenario ( galaxies age in groups before entering clusters ; @xcite ) .
this is consistent with the result of @xcite , who find a strong suppression of sf activity in a sample of @xmath151 groups .
they estimate a stronger quenching for galaxies with @xmath152 , where the fraction of star - forming galaxies falls down to @xmath7412% .
our results are also in agreement with an analogous analysis done by @xcite on the sample of 2df groups .
the authors show that the group star - forming galaxies are located below the field ms , but above the location of the bulk of cluster star - forming galaxies .
this suggests that , although some pre - processing is present in groups , a stronger quenching must happen in more dense and massive systems , like clusters . in this work
we also show that a certain amount of pre - processing happens when galaxies are falling along the filaments , before they enter the group environment .
however , halo - related processes seem to be more effective in quenching the star formation . the pre - processing scenario is also supported by models .
@xcite , using semi - analytic models , show that the fraction of galaxies that can be pre - processed in a group - size halo of mass @xmath153 is significant ( @xmath154 which raise to @xmath155 for galaxies with @xmath156 ) . furthermore , comparing observations with their theoretical predictions , they argue that satellite galaxies become passive after they have spent 5 - 7 gyr in haloes more massive than @xmath157 . similarly , @xcite , using the stellar masses and merger trees produced by the semi - analytic galaxy catalogues , suggest that all clusters in their sample exhibit a significant fraction of their galaxies accreted through galaxy groups .
for instance , they propose that this fraction is 40@xmath158 for @xmath159 clusters at @xmath160 and only @xmath74 25 @xmath158 at higher redshifts ( @xmath7 ) .
our results show qualitative agreement with this prediction .
conversely , @xcite , using cosmological @xmath72cdm n - body simulations , suggest that on average , @xmath7470% of cluster galaxies fall into the cluster potential directly from the field . on the other hand ,
less than @xmath7412% of cluster galaxies are accreted as members of groups with five or more galaxies . the pre - processing scenario is also reflected in our analysis on the qg fraction in groups , `` filament - like '' environments and field .
our findings suggest that these environments have a different galaxy - type mix up to @xmath1 , with groups being the most efficient at quenching the sf . at higher redshift ,
the galaxy population of groups , filaments and field is similar .
our results find support in several works in the literature .
for example , @xcite show that galaxy star - formation and colour transformation rates are higher in the group regions than in lower density areas at @xmath8 .
in addition , @xcite suggest that galaxy colours are particularly affected by the group environment ( with respect to the field ) on short time - scales in a redshift range @xmath161 .
finally , @xcite and @xcite show that the group galaxy population becomes bluer as the redshift increases , but it maintains a systematic difference with respect to the global galaxy population , and an even larger difference with respect to the isolated galaxy population .
in this work we have investigated the sfr density relation in different environments up to @xmath9 .
we have used multi - wavelength data from the ecdfs and goods fields to study the evolution of sf activity in four redshift bins . moreover , the use of deep mips 24 and _ herschel _ pacs data has assured an accurate estimate of the sfr for all detected ir sources ( in particular ms galaxies ) .
this rich dataset has enabled the use of two different approaches to investigate the evolution of the sfr
density relation : an `` environmental '' approach , which is the traditional method used in the literature , and a novel `` dynamical '' approach , which splits the sample into group , `` filament - like '' and field galaxies . by studying the sfr
density relation in the standard way , we have found an anti - correlation up to @xmath59 but no correlation at higher redshift .
although the significance found by the spearman test decreases as the redshift increases , we did not observe any reversal of the sfr density relation .
after checking for the presence of biases using the mock catalogues of @xcite , we have verified that we have constantly over - estimated the values of sfr at all redshifts , due to the spectroscopic incompleteness of our catalogues , but that the slope of our sfr density relation is not affected by any bias .
we have also found that the role of agn is rather marginal in shaping the relation and that the anti - correlation at @xmath93 is dominated by spectroscopic group members .
since our galaxy sample shows only a mild mass segregation at any redshift bin , we conclude that the sfr
density relation is not driven by a strong mass segregation . by using the `` dynamical '' definition of environment ,
we have asserted that the bulk of quenching happens in groups .
indeed , group spectroscopic members show a much lower mean sfr than galaxies at similar density but not belonging to bound structures , at least up to @xmath8 .
on the other hand , galaxies in unbound structures exhibit a similar evolution of sfr as field isolated galaxies .
group galaxies only reach the same level of sf activity as field galaxies at @xmath73 .
however , even with this alternative approach , we have not detected any significant sfr density reversal .
thus , we conclude that group galaxies experience a much faster evolution with respect to galaxies in other environments . in addition , the strong difference in the evolution of the group galaxies with respect to non - group galaxies at similar density ( i.e. `` filament - like '' ) reveals that processes related to the presence of a massive dark matter halo ( ram pressure stripping , strangulation , harassment ) must be dominant in the suppression of the sf activity in group galaxies below @xmath5 . on the other hand , purely density - related processes ( close encounters , tidal tripping ) play a secondary role in the quenching . in order to understand the cause of the faster evolution in group galaxies
, we have also studied the location of group , `` filament - like '' and field galaxies in the @xmath104 plane .
this has been done to identify if the lower @xmath162 in groups at @xmath163 with respect to field galaxies is due to a general quenching of the sf in all galaxies or to a faster evolution of the galaxy - type mix .
we have found that the ms of group galaxies is offset with respect to that of field galaxies up to @xmath59 , i.e. it is shifted towards lower sfrs . at higher
redshift the star - forming group galaxies are on sequence .
`` filament - like '' galaxies occupy a halfway position between groups and field .
this suggests that both the density- and halo - related processes are playing a role in quenching the sf activity of actively star - forming galaxies , but that density seems to play a secondary role .
interestingly , the qg fraction evolves faster in groups than in the other two environments up to @xmath59 , beyond which the fractions are comparable .
we conclude that the strong evolution observed in the sfr
density relation , analysed in the dynamical approach , is likely to be the driver of the different galaxy type mix in groups across cosmic epochs .
we thank the anonymous referee for her / his constructive comments .
pacs has been developed by a consortium of institutes led by mpe ( germany ) and including uvie ( austria ) ; kul , csl , imec ( belgium ) ; cea , oamp ( france ) ; mpia ( germany ) ; ifsi , oap / aot , oaa / caismi , lens , sissa ( italy ) ; iac ( spain ) .
this development has been supported by the funding agencies bmvit ( austria ) , esa - prodex ( belgium ) , cea / cnes ( france ) , dlr ( germany ) , asi ( italy ) , and cicyt / mcyt ( spain ) .
this research has made use of nasa s astrophysics data system , of ned , which is operated by jpl / caltech , under contract with nasa , and of sdss , which has been funded by the sloan foundation , nsf , the us department of energy , nasa , the japanese monbukagakusho , the max planck society , and the higher education funding council of england .
the sdss is managed by the participating institutions ( www.sdss.org/collaboration/credits.html ) .
cooper m. c. , newman j. a. , madgwick d. s. , gerke b. f. , yan r. , davis m. , 2005 , apj , 634 , 833 cooper m. c. , et al .
, 2008 , mnras , 383 , 1058 cooper m. c. , et al . , 2010 , mnras , 409 , 337 cooper m. c. , et al . , 2012 , mnras , 425 , 2116 daddi e. , et al . , 2007 , apj , 670 , 156 | we investigate the evolution of the star formation rate ( sfr)density relation in the extended chandra deep field south ( ecdfs ) and the great observatories origin deep survey ( goods ) fields up to @xmath0 .
in addition to the `` traditional method '' , in which the environment is defined according to a statistical measurement of the local galaxy density , we use a `` dynamical '' approach , where galaxies are classified according to three different environment regimes : group , `` filament - like '' , and field . both methods show no evidence of a sfr density reversal .
moreover , group galaxies show a mean sfr lower than other environments up to @xmath1 , while at earlier epochs group and field galaxies exhibit consistent levels of star formation ( sf ) activity .
we find that processes related to a massive dark matter halo must be dominant in the suppression of the sf below @xmath1 , with respect to purely density - related processes .
we confirm this finding by studying the distribution of galaxies in different environments with respect to the so - called main sequence ( ms ) of star - forming galaxies .
galaxies in both group and `` filament - like '' environments preferentially lie below the ms up to @xmath1 , with group galaxies exhibiting lower levels of star - forming activity at a given mass . at @xmath2 ,
the star - forming galaxies in groups reside on the ms .
groups exhibit the highest fraction of quiescent galaxies up to @xmath1 , after which group , `` filament - like '' , and field environments have a similar mix of galaxy types .
we conclude that groups are the most efficient locus for star - formation quenching .
thus , a fundamental difference exists between bound and unbound objects , or between dark matter haloes of different masses .
[ firstpage ] galaxies : groups : general galaxies : evolution galaxies : star formation rate infrared : galaxies |
the work was supported by the engineering and physical sciences research council ( epsrc ep / m027961/1 and ep / m013812/1 ) , the leverhulme trust ( rpg-2014 - 238 ) and the royal society ( rg140457 ) .
33ifxundefined [ 1 ] ifx#1 ifnum [ 1 ] # 1firstoftwo secondoftwo ifx [ 1 ] # 1firstoftwo secondoftwo `` `` # 1''''@noop [ 0]secondoftwosanitize@url [ 0 ]
+ 12$12 & 12#1212_12%12@startlink[1]@endlink[0]@bib@innerbibempty link:\doibase 10.1002/adma.201300588 [ * * , ( ) ] link:\doibase 10.1038/nature13487 [ * * , ( ) ] link:\doibase 10.1038/srep20347 [ * * , ( ) ] link:\doibase 10.1063/1.4916232 [ * * ( ) , 10.1063/1.4916232 ] link:\doibase 10.1103/physrevb.51.11164 [ * * , ( ) ] link:\doibase 10.1038/srep07234 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.105.266801 [ * * , ( ) ] link:\doibase 10.1002/lpor.201000046 [ * * , ( ) ] link:\doibase 10.1021/nn404703z [ * * , ( ) ] link:\doibase 10.1063/1.4723564 [ * * ( ) , 10.1063/1.4723564 ] link:\doibase 10.1021/nl501738b [ * * , ( ) ] link:\doibase 10.1039/c4fd00187 g [ * * , ( ) ] link:\doibase 10.1021/acsphotonics.5b00226 [ * * , ( ) ] link:\doibase 10.1063/1.4939263 [ * * ( ) , 10.1063/1.4939263 ] link:\doibase 10.1126/science.1210713 [ * * , ( ) ] link:\doibase 10.1038/ncomms3807 [ * * , ( ) ] link:\doibase 10.1038/ncomms3665 [ * * , ( ) ] link:\doibase 10.1364/oe.21.00a268 [ * * , ( ) ] link:\doibase 10.1063/1.3206091 [ * * , ( ) ] link:\doibase 10.1038/nphoton.2010.7 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.117.053902 [ * * , ( ) ] link:\doibase 10.1073/pnas.0907744106 [ * * , ( ) ] link:\doibase 10.1103/physrevb.80.155112 [ * * ( ) , 10.1103/physrevb.80.155112 ] link:\doibase 10.1002/adom.201300415 [ * * , ( ) ] link:\doibase 10.1073/pnas.1307879110 [ * * , ( ) ] link:\doibase 10.1109/jstqe.2016.2528125 [ * * , ( ) ] link:\doibase 10.1038/srep19325 [ * * , ( ) ] link:\doibase 10.1039/c4cp06024e [ * * , ( ) ] link:\doibase 10.1364/optica.3.000763 [ * * , ( ) ] link:\doibase 10.1103/physrevb.87.165116 [ * * , ( ) ] link:\doibase 10.1103/physrevb.91.020201 [ * * , ( ) ] , link:\doibase 10.1038/srep13301 [ * * , ( ) ] link:\doibase 10.1016/s1748 - 0132(06)70120 - 7 [ _ _ ] ( , , ) | hyperuniform geometries feature correlated disordered topologies which follow from a tailored k - space design . here
we study gold plasmonic hyperuniform metasurfaces and we report evidence of the effectiveness of k - space engineering on both light scattering and light emission experiments .
the metasurfaces possess interesting directional emission properties which are revealed by momentum spectroscopy as diffraction and fluorescence emission rings at size - specific k - vectors .
the opening of these rotational - symmetric patterns scales with the hyperuniform correlation length parameter as predicted via the spectral function method .
coherent control of optical waves by scattering from 2d nanostructured surfaces is revolutionising the way we shape the wavefront of an incoming light beam , opening new avenues for miniaturised optical components for integrated optical circuits @xcite , flat display technology @xcite , and energy harvesting @xcite .
metallic surfaces are in particular attractive due to the strong light - matter interaction associated with surface plasmons , enabling diffraction control through plasmonic crystals @xcite and metal nano - particle arrays @xcite , broadband operation and increase of the plasmon mode density @xcite , enhanced omni - directional light extraction and coupling @xcite , broadband absorption @xcite , fluorescence enhancement @xcite and lasing @xcite , and more recently the realisation of ultra thin lenses @xcite and metasurface holograms @xcite .
whereas periodic geometries suffer from limited rotational symmetries , aperiodic and disordered topologies , with their richer symmetries and patterns , can lead to superior optical functionalities @xcite , as in omnidirectional absorption for solar applications @xcite , scattering - induced light localisation @xcite and light extraction from led / oled @xcite .
moreover , disordered metasurfaces are expected to be more resilient against fabrication imperfection and therefore more apt for technological implementation . given the vast possible designs of non - periodic topologies , ranging from random to correlated - disordered ,
their full potential is still to be fully explored .
there exists a general class of disordered systems , called hyperuniform disordered ( hud ) photonic structures , which are of particular interest because they exhibit wide and isotropic photonic band gaps @xcite , rotational symmetry and broadband k - space control , and can be systematically generated through a specific design rule via universal tessellation protocol @xcite .
pioneering experiments on photonic hud systems have explored ir light diffraction in 3d dielectric structures @xcite , microwave band - gaps formation @xcite , polarization filtering @xcite and random quantum cascade lasers @xcite .
theoretical proposals have been put forward for surface enhanced raman scattering @xcite , transparency design @xcite , high - q optical cavities and low - loss waveguides @xcite , and microwave photonic circuits @xcite .
hud structures fabrication is improving quickly , reaching already the ir range @xcite but not yet the visible . here , we report visible light scattering and light emission experiments from hud plasmonic gold metasurfaces .
we find that scattering from the metasurfaces is principally directed into an annular angular pattern indicating reciprocal space engineering .
moreover , we investigate directional emission from near - field coupled emitters which , as confirmed by theoretical modelling , is shaped into a ring from the effective band folding into the light cone by scattering processes .
presents a structure factor ( panel b ) with a typical zero around @xmath0 and a broad isotropic diffraction maximum ( inset ) .
the hud connected network ( panel c ) broadly preserves the @xmath1 space characteristics ( panel d ) of the point pattern .
sketches and sem images of the resulting pillar - type ( panel e and g ) and network - type ( panel f and h ) metasurfaces.,width=264 ] the metasurface design is derived from a stealth hyperuniform point pattern with @xmath2 comprising @xmath3 points @xcite and generated under a periodic boundary condition , for a given average inter - scatterer distance @xmath4 . a section of the point pattern , decorated by discs of radius @xmath5 ,
is shown in fig .
[ fig : kingsdesigns]a . the structure factor of the point pattern is shown in fig .
[ fig : kingsdesigns]b : the point distribution exhibits significant local structural correlations : the typical exclusion region around @xmath0 which characterizes stealthy hyperuniform patterns and a broad isotropic diffraction maximum peaked around @xmath6 .
next , a delaunay tessellation protocol @xcite is performed , obtaining a strictly trivalent continuous network topology ( fig .
[ fig : kingsdesigns]c ) with walls of thickness @xmath7 . the structure factor of this network is presented in fig .
[ fig : kingsdesigns]d .
the stealthiness of the architecture has been significantly reduced : the diffraction spectrum ( fig .
[ fig : kingsdesigns]d , inset ) exhibits low intensity diffuse scattering around @xmath0 , in contrast to the sharp exclusion zone of the simple point pattern ( fig . [ fig : kingsdesigns]b , inset ) .
nonetheless , the general form of the point pattern structure factor dominated by a single broad and isotropic resonance around @xmath8 is reproduced by the network .
gold metasurfaces were fabricated by electron beam lithography on a glass substrate for various size scaling parameters @xmath4 .
the samples are of two kind : _ pillar - type _
samples comprising isolated pentagonal , hexagonal and heptagonal gold pillars ( sketched in fig .
[ fig : kingsdesigns]e and sem in fig .
[ fig : kingsdesigns]g ) , and _ network - type _ designs ( identical but inverted ) consisting of a connected network of gold ( sketch in fig .
[ fig : kingsdesigns]f and sem in fig .
[ fig : kingsdesigns]h ) .
the samples are labelled as @xmath9 for pillars and @xmath10 for networks , with @xmath11 : for example @xmath12 for @xmath13 nm has a side dimension of @xmath14 m .
network designs larger or smaller than @xmath15 m have been cropped or periodically repeated , respectively , to cover a @xmath16m@xmath17 area .
fig .
[ fig : kingsdiffraction ] presents light scattering experiments on the pillar - type metasurfaces whose sem images are shown in fig .
[ fig : kingsdiffraction]a .
similar experiments performed on the network - type samples led to comparable results and are not shown here .
the measured farfield intensity distribution of each pillar metasurface is shown in fig . [
fig : kingsdiffraction]b .
the samples were illuminated through the glass substrate with a collimated laser ( @xmath18@xmath19 nm ) while the back - scattered light was recorded in the farfield by imaging the fourier plane of a microscope objective ( oil immersion , na=1.45 ) .
the maximum observable momenta is overlaid as a coloured dashed circle .
all @xmath20 samples exhibit broad and statistically isotropic scattering rings which resemble the primary resonance of the designed structure factor shown in the inset of fig .
[ fig : kingsdesigns]d .
the momentum associated with the scattering resonance peak increases with the downscaling of the sample ; in @xmath21 the scattering ring crosses over the observable momentum limit . the bright spot at the centre of each farfield results from the specular reflection along the axis of incidence . fig .
[ fig : kingsdiffraction]c shows the azimuthally averaged farfield intensity distributions .
as the sample correlation length is reduced in size , the momentum of the primary scattering peak increases with a small intensity decrease .
the structures @xmath22 are all characterised by a single scattering peak with some finer structure that varies from sample to sample .
interestingly , the @xmath23 sample exhibits a double peak , while @xmath24 only a secondary low shoulder peak . the linear scaling with the reciprocal of the correlation length , shown in the insets of figs .
[ fig : kingsdiffraction]c , can be expected by simple diffraction theory , while the finer structures are captured by numerical finite - difference time - domain ( fdtd ) simulations shown in fig .
[ fig : kingsdiffraction]d . in particular
, fdtd predicts the multi - peak signature of the @xmath23 sample , suggesting that this feature is a genuine property of the sample .
we attribute this finer structure to the interplay of the surface plasmon at the air / gold interface sustained by a single pillar with the linear diffraction dispersion .
in fact , the hud structure comprises elements of different sizes , much larger than the plasmon wavelength , which present high - order resonances in the visible range and an overall response close to that of a surface plasmon resonance of an infinite film which peaks around @xmath25 .
we performed also a broadband scattering characterization of the @xmath12 design .
the sample was illuminated from below with white light , and the reflected and scattered light was spectrally decomposed into its wavelength and momentum components ( fig . [ fig : kingswhitelightscattering]b ) by spectrally imaging the fourier plane of the sample . in this way
an energy - wavevector dispersion diagram can be constructed as shown in fig [ fig : kingswhitelightscattering]b ( experiments ) and fig [ fig : kingswhitelightscattering]a ( fdtd calculations ) .
both images display a bold diagonal slash , from low - scattering angle to high - scattering angle , and the evolution , for increasing wavelengths , towards larger momenta of the primary scattering peak . from this linear relationship
we conclude that the light diffraction follows the designed structure factor with a single main peak ( fig .
[ fig : kingsdesigns]d ) . metasurface .
both theory and experiment match
well and show that the scattering peak momentum increases linearly with wavelength .
, width=264 ] so far , we have investigated the ability of our gold metasurfaces to mediate between incident and scattered light .
we now seek to characterise the hud metasurface electromagnetic modes and their momentum distribution .
the metasurface modes are excited by a @xmath26 nm layer of poly(methyl methacrylate ) polymer highly doped with fluorescent 4-(dicyanomethylene)-2-methyl-6-(4-dimethylaminostyryl)-4h - pyran ( dcm ) dye molecules which was spin coated on the samples ( fig .
[ fig : kingsfluorescence ] , top panel ) .
figs .
[ fig : kingsfluorescence]a and b present the theoretical and experimental frequency - momentum distribution , or dispersion plot , of the fluorescence light emitted from the @xmath12 structure when excited with a green laser at a wavelength of 532 nm . of @xmath12 slab modes calculated summing incoherently the farfields of 48 randomly oriented dipole ( panel a ) and measured by fluorescence emission ( panel b).,width=264 ] the pair of intense emission bands just outside the light lines ( white dashed lines indicating @xmath27 ) results from the characteristic radiation profile of a dipole near the glass - air interface @xcite . inside the light cone , we observe annular features which describe the decomposition of the metasurface slab modes into their in - plane momentum components .
these may be viewed as the generalised dispersion relation @xmath28 of the slab modes .
similar dispersion diagrams were measured for the @xmath12 , @xmath29 , @xmath30 , @xmath31 and @xmath23 metasurfaces . in order to further analyse the dispersion plots of figs .
[ fig : kingsfluorescence ] , we convert the emission wavelength and in - slab momentum to the dimensionless quantities @xmath32 and @xmath33 respectively .
the experimental dispersion diagrams can then be stacked into a single image to illustrate the dispersion over a large normalized frequency range ( fig .
[ fig : kingsspectralfn ] , left panel ) .
it shows that the fluorescent light is emitted isotropically from the hud surface , into a cone with varying opening angle , which depends on the emission wavelength to correlation length ratio . as the frequency increases from @xmath34 , the dominant momentum of the slab modes first decreases , reaching a zero in the region of @xmath35 , at which point light
is emitted normal to the slab plane , before gradually opening back out .
the experiments are in good agreement with fdtd calculations ( fig .
[ fig : kingsspectralfn ] , right panel ) obtained with the spectral function method @xcite . specifically , decomposition of a slab s eigenmodes into a plane wave basis can directly predict the farfield angular profile of its directional emission .
these results show that the electromagnetic dispersion diagram of the hud metasurface follows the designed structure factor ( figure [ fig : kingsdesigns]d ) and exhibits band folding resulting from bragg - like processes . ,
@xmath29 , @xmath30 , @xmath31 and @xmath23 are stacked together as a function of @xmath32 .
right panel , fdtd simulations which confirm the crossing at @xmath32 = 0.9 where the light is emitted normal to the metasurface plane.,width=264 ] in conclusion , we designed , fabricated and characterized gold metasurfaces derived from hud connected networks . by reciprocal space engineering
we designed the structure factor to be dominated by a single , broad scattering resonance which was observed to dictate both the annular farfield light scattering and directional emission properties of the metasurfaces .
the observed hud metasurface dispersion corresponds to an effective medium band that is back - folded by bragg - like processes .
surface plasmon resonance where found to contribute to the diffraction by inducing additional peaks from their interaction with the structural diffraction peak .
the observed light emission and scattering engineering has important applications for light extraction from light - emitting devices , absorption in solar cells and annular redirection for displays , by exploiting the full azimuthal symmetry provided by the hud metasurfaces design . |
over the last few decades , stellar population synthesis has become one of the most popular and powerful techniques to study the formation and evolution histories of galaxies too distant to be resolved into individual stars . in particular , the measurement of absorption line - strengths combined with stellar population models has been used to investigate the luminosity - weighted age , metallicity and abundance ratios in integrated stellar populations .
extensive work has been carried out on stellar populations in early - type galaxies , mainly using long - slit spectra ( see for example @xcite , davies , sadler & peletier 1993 , fisher , franx & illingworth 1995 , fisher , franx & illingworth 1996 , @xcite , @xcite , @xcite , @xcite ) and recently also integral - field spectroscopy ( kuntschner et al . 2006 ,
hereafter paper vi , mcdermid et al .
2006 , hereafter paper viii ) .
the main advantage brought by the integral - field observations is that one can obtain two - dimensional maps and easily identify extended structures . in particular
, the two - dimensional coverage gives the possibility to straightforwardly connect the stellar populations with the kinematical structures and therefore represents a powerful tool towards a comprehensive understanding of the structure of the galaxy .
+ some literature is also available on absorption line - strengths in early - type spiral galaxies , mainly s0 and sa galaxies ; ( see for example jablonka , martin & arimoto 1996 , @xcite , @xcite , @xcite , and peletier et al .
2007 , hereafter paper xi ) .
in particular , @xcite presents two - dimensional absorption line - strength maps of 24 spiral galaxies , mostly of type sa , observed with the integral - field spectrograph sauron ( bacon et al .
2001 , hereafter paper i ) ; these reveal young populations in sa galaxies , possibly formed in mini - starbursts in structures such as nuclear discs , circular star forming rings and bars .
these mini - starbursts generate a large scatter in index - index diagrams , larger than observed for elliptical galaxies . as a consequence
, there is a large range in luminosity - weighted ages .
different star formation modes -starbursts or quiescent- are also reflected in the wide range observed in the abundance ratios .
all the galaxies of paper xi lie on or below the mg_b _ - @xmath1 relation for ellipticals determined by jrgensen , franx & kjrgaard ( 1996 ) .
the authors argue that , if one considers that relation as valid for old galaxies , then the early - type spirals present a scatter in age , with the biggest scatter for the smallest velocity dispersions .
+ spiral galaxies towards the end of the hubble sequence are , on the other hand , still poorly - studied objects , due to the difficulty of obtaining reliable measurements in these low - surface brightness objects , full of dust and star forming regions , whose spectra can be dramatically contaminated by emission - lines .
in particular , the study of their stellar populations is extremely complicated . an approach based on broad - band colours
is basically invalidated by the huge amount of dust , which is very difficult to take into account , especially when working with data from ground - based telescopes . a spectroscopic approach , relying for example on stellar absorption - strength indices , has instead to deal with the almost ubiquitous presence of gas with consequent emission , which makes it necessary to accurately remove the emission - lines from the spectra before performing any population analysis . only in the last decade
several hst - based imaging surveys have targeted these late - type galaxies . particularly relevant
are the papers by @xcite , carollo & stiavelli 1998 , carollo , stiavelli & mack 1998 , @xcite , @xcite , @xcite , @xcite , where the authors reveal the presence of a variety of structures in the inner regions , such as bulges , nuclear star clusters , stellar discs , small bars , double bars and star forming rings whose formation and evolutionary pattern are not properly understood yet .
@xcite also performed a population analysis on nine late - type spiral galaxies ( types between sa and sc ) on the basis of hst acs and nicmos optical and near - infrared colours .
the high spatial resolution of their space - based data allowed them to mask dust features and measure the properties of the stellar populations from the colours ; this method , together with the high spatial resolution provided by hst , produced results that do not suffer much from the effects of dust ; only a smooth dust distribution would not be detected . from their analysis , they found a large range in the colour properties of bulges in late - type spirals , reflecting a large range in their stellar population properties . from a comparison with population models
, they concluded that in about half of their bulges the bulk of the stellar mass formed at early epochs ( more than 50% of their mass was formed more than 9 gyr ago ) , and that in the other half a non - negligible fraction of the stellar mass ( up to 25% ) was formed recently , in the last @xmath2 3 gyr .
+ until now , only little spectroscopic counterpart for the quoted imaging was available , due to the mentioned observational difficulties .
the published work ( @xcite , zaritsky , kennicutt & huchra 1994 , @xcite , @xcite , @xcite ) mainly refers to the emission - line properties , to the characteristics of the hii regions or to the nature of the innermost component . in particular , no study has addressed yet in a systematic way the stellar populations of late - type spiral galaxies with a two - dimensional coverage .
existing studies on abundances concern topics such as elemental abundances in hii regions and the distribution of oxygen within discs @xcite , not properties inferred from the integrated stellar absorption spectrum . as for the absorption - lines properties ,
there is ongoing work by macarthur , gonzalez & courteau ( 2007 ) , who analyse long - slit spectra ( gemini - gmos ) of eight spiral galaxies of type between sa and scd on a wide spectral range , measuring many absorption indices out to 1 - 2 disc scale - lengths .
also , @xcite presented line - strengths in the bulges and inner discs of 38 galaxies of morphological type between s0 and sc for which they acquired long - slit spectra on a broad spectral range including balmer lines , mg and fe features .
they concluded that the central regions of bulges span a wide range both in ssp metallicity and age , confirming the results of @xcite .
they also found that luminosity - weighted metallicities and abundance ratios are sensitive to the value of the central velocity dispersion and of the maximum disc rotational velocity , and that red bulges ( defined as those with b@xmath6k @xmath7 4 ) of all types are similar to luminous ellipticals and obey the same scaling relations ; for the blue bulges ( b@xmath6k @xmath8 4 ) they observed instead some differences .
they also addressed the radial variations of the population parameters ; in most cases , they measured negative metallicity gradients with increasing radius , in the bulge - dominated region ; positive gradients in age were found only in barred galaxies .
+ as mentioned , there is still a lack of work on populations in late - type spirals on a full two - dimensional field
: we therefore started a project on a sample of 18 late - type spiral galaxies , with the purpose of investigating the nature of their inner regions , addressing the bulge and disc formation and evolution , the interconnection between stellar and gaseous components and the star formation history .
given the high complexity in the inner regions of these objects , integral - field spectroscopy has to be preferred with respect to long - slit , providing a full picture of the kinematics and populations on a two - dimensional field of view .
we observed our 18 late - type spiral galaxies using sauron , an integral - field spectrograph that was built for a representative census of ellipticals , lenticulars and early - type spiral galaxies , to which in the rest of this paper we will refer as ` the sauron survey ' .
our project can be regarded as an extension of the sauron survey towards later - type objects .
first results based on our data are presented by @xcite and focus on the stellar and gas kinematical maps .
the main findings of that paper are that in many cases the stellar kinematics suggests the presence of a cold inner region , as visible from a central drop in the velocity dispersion , detected in about one third of the sample ; that the ionized gas is almost ubiquitous and often presents more irregularities in its kinematics than the stellar component ; and that the line ratio [ oiii]/h@xmath0 assumes often low values over most of the field , possible indication for wide - spread star formation , in contrast to early - type galaxies . + as a following step in our project , we subtracted the emission - lines from our sauron spectra , measured as explained in @xcite , and on the emission - cleaned spectra derived absorption line - strength indices in the lick / ids system ( @xcite , @xcite , @xcite ) , which allows an easier comparison with existing data and with other classes of objects . here
we present and analyse the absorption line - strength maps .
the data and maps presented in this paper will be made available via the sauron web page http://www.strw.leidenuniv.nl / sauron/. + the paper is structured as follows .
section 2 briefly describes the sample selection , observations and data reduction .
section 3 reviews the analysis methods . in section 4 we present the line - strength maps for all the galaxies in our sample .
in section 5 we investigate the correlations between different indices and between indices and other galactic parameters , such as the morphological type and the central velocity dispersion , mainly focussing on the central values . in section 6
we present our estimates for the ages and metallicities , obtained by comparison of our observed line - strength indices with model predictions ; we also explore the star formation history and give estimates for the time - scale @xmath5 , in an exponentially declining star formation rate scenario .
section 7 summarises the main results . in appendix
a we deal with the radial variations of the indices and the stellar population parameters , presenting azimuthally averaged radial profiles .
appendix b shortly describes the individual galaxies and their peculiarities , giving a brief summary of relevant literature information . for more literature reviews on these objects
, we suggest the reader to check section 6.4 in @xcite .
the galaxies were optically selected ( @xmath9 @xmath8 12.5 , according to the values given in @xcite , hereafter rc3 ) with hst imaging available from wfpc2 and/or nicmos .
their morphological type ranges between sb and sd , following the classification given in ned ( from the rc3 ) .
galaxies in close interaction and seyferts were discarded .
the resulting sample contains 18 nearby galaxies , whose main properties are listed and illustrated respectively in table 1 and fig . 1 in @xcite .
the ngc numbers and some characteristics are reported also in the tables in the present paper .
observations of the 18 sample galaxies were carried out during 6 nights in january 2004 , using the integral - field spectrograph sauron attached to the 4.2-m wht .
the exposure details are listed in table 2 in @xcite .
+ we used the low spatial resolution mode of sauron , giving a field - of view ( fov ) of 33 @xmath10 41 .
the spatial sampling of individual exposures is determined by an array of 094 @xmath10 094 square lenses .
this produces 1431 spectra per pointing over the sauron fov ; another 146 lenses sample a region 19 away from the main field in order to measure simultaneously the sky background .
sauron delivers a spectral resolution of 4.2 fwhm , and its maximum spectral coverage is 4800 - 5380 ; the wavelength interval common to all lenses is the narrow spectral range 4825 - 5275 ( 1.1 per pixel ) .
this wavelength range includes a number of important stellar absorption lines ( e.g. h@xmath0 , fe5015 , mg_b _ ) , and potential emission - lines as well ( h@xmath0 , [ oiii ] , [ ni ] ) . for a more exhaustive description of the instrument , see @xcite
. + data reduction was carried out using the dedicated software xsauron developed at centre de recherche astronomique de lyon ( cral ) .
for details on the data reduction we refer the reader to @xcite and @xcite . for an explanation of the spectrophotometric calibration we recommend the reader to study paper vi , in particular its section 3.1 .
the fully reduced and calibrated spectra were first spatially binned using a voronoi tessellation @xcite in order to reach a minimum signal - to - noise ratio of @xmath2 60 per spectral resolution element per bin .
the kinematics of stars and gas were then extracted from the binned spectra via the pixel - fitting methods extensively described in @xcite , emsellem et al .
( 2004 , hereafter paper iii ) , and sarzi et al .
( 2006 , hereafter paper v ) , that we already applied to calculate the kinematical maps shown in @xcite .
the spectra were fitted using the ssp models from @xcite .
we selected a library of 48 models evenly sampling a wide range in both age and metallicity ( 0.50 @xmath3 age @xmath3 17.78 gyr , @xmath61.68 @xmath3 [ fe / h ] @xmath3 + 0.20 ) .
we used this ssp library and additional gaussian templates reproducing the emission - lines to fit the observed spectra , separating the emission - lines from the underlying absorption line spectrum ( see @xcite for details on the procedure ) .
we notice that for our objects this separation is crucial if we want to measure absorption strengths , since in spiral galaxies the gas is almost ubiquitous and emission in several cases dominates the spectrum , filling in completely the absorption features , as shown for example in the cases of some central aperture spectra in the top row of fig .
[ spec_864 ] ( black lines ) ; in the same figure we overplot the best - fit to the spectra obtained with our method ( red lines ) and the emission - removed spectra ( blue lines ) , illustrating the reliability of the applied gas - cleaning procedure .
the bottom row of fig . [ spec_864 ] presents similar plots for bins located @xmath2 10 from the galaxy s centre , showing that our fits and emission correction are reliable also outside the central region , given the fact that a minimum signal - to - noise of 60 is garanteed by the spatial binning .
the reader can find more examples of this in fig . 1 in @xcite , fig . 3 in @xcite , and fig . 3 in falcn - barroso et al .
( 2006 ) , hereafter paper vii .
+ we also notice that in two cases ( namely , ngc2964 and ngc4102 ) we detected regions of activity ( see @xcite for a description ) , where the spectra are characterised by double - peaked emission - lines due to the presence of multiple velocity components in the gas , because of the presence of a jet .
in those regions our method of fitting the lines with single gaussians fails to reproduce accurately the line profile , but since this is limited to only a few spectra , we did not modify the method . in any case , this does not have any impact on our following analysis , since the double - peaked emission - lines do not contaminate the spectra defining the central aperture ( see section [ correlationsec ] ) on which most of this study is focused on . on the emission - removed spectra we calculated line - strength indices on the lick / ids system . for details on the actual computation of the line - strengths , we refer to @xcite .
the calibration to the lick / ids system takes into account several effects : the difference in spectral resolution between the lick / ids system and the sauron instrumental setup , the internal velocity broadening of the observed galaxies and small systematic offsets rising from differences in the shape of the continuum , due to the fact that the original lick / ids spectra have not been flux calibrated . in our observing run , we obtained spectra for 21 different stars in common with the lick / ids stellar library @xcite , for a total of 39 repeated observations . fig . [ lickoffsets ] shows the difference between the lick / ids measurements and ours ( for stars with repeated observations , we considered the average of the different values available ) .
we calculated the mean offsets to the lick / ids system and their dispersion using bi - square weighting in order to minimize the influence of the outliers , after removing from the sample two stars of spectral type m6iii and m7iii , whose measured metal indices differ more than 2 from the lick measurements .
the adopted offsets that we applied to all our measurements are : @xmath60.06 ( @xmath11 0.06 ) for h@xmath0 , + 0.32 ( @xmath11 0.15 ) for fe5015 and + 0.15 ( @xmath11 0.05 ) for mg_b_. + from a visual inspection of our maps of the fe5015 index , we saw that our measurements tend to assume high values in the outer parts , especially in the bottom - right corner , revealing a systematic effect that makes the maps look asymmetric .
the most striking manifestation of the problem is this ` high - iron corner ' , but there could be as well regions where the index is too low .
a detailed inspection of the spectra showed that the cause is a wrong shape of the continuum in the fe5015 spectral region , as proved by the fact that we are unable to fit our twilight spectra using our ssp template library : the fe5015 region is not matched by the templates , and the fe5015 map calculated on the twilight spectra is not flat over the field , showing a range of about 1.6 from minimum to maximum value .
this continuum problem could be due to instrumental instabilities , to the high sensitivity of the grism to the incidence angle of the light , to the imperfect extraction of the spectra , to the misalignment of the spectra with respect to the columns of the ccd ( see @xcite ) .
we can exclude , instead , relations with problems in the gas removal , since the fe5015 measurement is affected also in the twilight frame , where the spectra are gas - free .
in addition to this , continuum problems should not influence the gas - cleaning procedure : in the fitting , a polynomial is included in order to account for differences in the flux calibration between the observed spectra and the model ones .
+ unfortunately , we could not take this effect into account properly at the data reduction stage , and decided to apply a correction to the affected spectra before measuring the fe5015 index .
a problem related to the shape of the continuum was experienced also for the early - type spirals of the sauron survey and affected the h@xmath0 measurements , due to the closeness of the spectral feature to the edge of the spectral range ; despite the fact that the problem is different , we worked out a correction similar to the one that was there applied in the case of h@xmath0 and described in @xcite ( section 3.1.2 ) .
+ for each galaxy , we averaged the spectra within an aperture of 10 radius , obtaining a ` global spectrum ' for that galaxy .
we then fitted this global spectrum using a linear combination of ssp templates , masking the spectral regions possibly affected by emission lines . in this way we obtained for each galaxy a ` global optimal template ' that approximates very well the general spectral features of that galaxy .
then we considered the two - dimensional galaxy s datacube , spatially binned and cleaned from emission as explained in the previous sections ; for each bin , we fitted the cleaned spectrum using the ` global optimal template ' together with an 11-th order multiplicative polynomial , over the whole spectral range .
the fit determined the best - fitting polynomial continuum , from which we then removed the linear component ( which does not change the line indices ) ; we considered the residuals and calculated their rms variation over the wavelength range of fe5015 . for bins covering four or more single - lens spectra and with an rms greater than 0.01
, we used the fitted polynomial as a correction to the spectral shape by dividing it through the emission - cleaned spectrum .
for these bins , we finally measured the fe5015 index on the corrected spectrum .
+ this correction depends critically on the assumption that the optimal template is an accurate representation of the true galaxy spectrum . whilst this is generally true , in the central ,
sometimes bulge - dominated , regions of our galaxies , non - solar abundance ratios can create a fit residual similar in size to the systematic effect we wish to remove . by correcting only bins containing four or more lenses , we avoid these central parts , since the data there are unbinned . correcting large bins is also important because of their larger contribution to the spatial appearance of the maps . + we verified the fe5015 correction using exposures of the twilight sky .
as we mentioned above , in the twilight maps of fe5015 , a spatial gradient was also visible , as is apparent in the galaxy data . in the uncorrected twilight , the field centre gave values of fe5015 close @xmath2 4 , consistent with a typical g2v - g4v sun - like spectrum , showing that the field centre is not significantly affected by the problem .
after applying our correction technique , the gradient was reduced , and the global level remained around 4 . we show this in fig .
[ twilightcorrection ] , where we display the fe5015 map calculated on the twilight frame , before and after applying the correction .
similarly , in fig . [ correctionexample ] we show the fe5015 map for ngc488 before and after correction .
white asterisks mark the centroids of the bins that have been corrected .
the correction clearly removes most of the asymmetry , proving the usefulness of the method applied .
+ figure [ correctedregions ] presents for each galaxy the map of the rms deviations of the fitted continuum after removal of a linear slope .
we plot in black the bins that do not meet the requirements for the correction , for which we have artificially put the rms to 0 ( for plotting purposes only ) ; in these bins the fe5015 index is calculated on the original emission - cleaned spectra .
the maps are oriented with the horizontal and vertical axis aligned , respectively , to the long- and to the short- axis of the sauron field .
the extension of the corrected area is mainly related to the signal - to - noise of the data : galaxies with poor signal - to - noise will require higher binning and the big bins will undergo the correction . + as for the uncertainties on the measured indices , the most worrying factor is the separation of emission and absorption lines .
simulations presented in appendix a of paper v show that the accuracy in recovering the emission - line fluxes does not depend on the a / n ratio between the line amplitude and the noise level in the stellar continuum , but only on the signal - to - noise level in the stellar continuum itself .
these simulations were meant to investigate the accuracy of emission - line fluxes measurements for ellipticals and lenticulars .
in the case of signal - to - noise = 60 , the typical uncertainties in the line fluxes obtained in appendix a of paper v translates in errors in the equivalent width of the emission line of @xmath2 0.08 , and in similar errors in the absorption line indices .
@xcite presents absorption line - strengths for early - type spirals observed with sauron ; in some of the galaxies studied there , the emission lines are stronger by a factor up to 100 , relative to the absorption lines , than in the ellipticals or lenticulars of paper v with the strongest emission - lines .
therefore , for the objects in paper xi the situation is much closer to what happens in our spirals .
there , the authors report results of similar simulations of the reliability of the gas cleaning , exploring a range in a / n ranging up to 100 .
the results are such that the errors in the absorption line indices do not vary appreciably with increasing a / n .
when coming to the data , one has to consider that the spectra are packed close together , so that the absorption lines are affected by neighbouring spectra ; despite the reduction tries to minimize this contamination , there might be some residual effects , particularly in presence of strong emission lines .
additional errors can derive from template mismatch in the fitting procedure and from the continuum correction described above .
therefore , throughout the rest of the paper we adopt a conservative value of 0.2 for the uncertainties in h@xmath0 and mg_b _ and 0.3 in fe5015 .
the same figures were used in paper xi .
figures [ maps1]-[maps9 ] present the absorption line - strength maps of the 18 galaxies , ordered by increasing ngc number . for each galaxy , in the first row
we show the total intensity obtained by integrating the full sauron spectra in the wavelength direction , the galaxy ngc number ( and the ugc also , when available ) , and the two - dimensional map of the characteristic time - scale of star formation @xmath5 , in gyr ( see section [ continuum ] ) .
overlapped in pink on the intensity map are the contours delimiting the outer limits of the so - called ` bulge ' and ` disc regions ' , defined in section [ correlationsec ] . in the second row we present the two - dimensional maps of h@xmath0 , fe5015 and mg_b _ , where the measured indices have been calibrated to the lick / ids system and are expressed in equivalent widths and measured in .
the maps of fe5015 have been corrected as explained in section [ correctionsec ] ; the corrected bins are marked with a white dot . in the third row , we present the maps of age ( expressed in units of decimal logarithm and measured in gyr ) , metallicity ( in units of decimal logarithm , with the solar metallicity as zero point ) and abundance ratios obtained with the one - ssp approach ( see section [ singlessp ] ) . overplotted on each map are the isophotal contours . the maps are all plotted with the same spatial scale , and oriented with the horizontal and vertical axis aligned , respectively , to the long- and to the short- axis of the sauron field for presentation purposes ; the orientation is the same as in @xcite and as in figures [ correctionexample ] and [ correctedregions ] in this paper .
the relative directions of north and east are indicated by the arrow above the galaxy s name .
the maximum and minimum of the plotting ranges and the colour table are given in the tab attached to each map . to make the comparison among different objects more immediate
, we adopted the same plotting ranges for all of the galaxies , but in appendix [ individuals ] we will enlighten the spatial structures within the single maps , when noticeable , by inspecting the maps plotted on a ` galaxy - by - galaxy range ' .
in this section we analyse the dependence between the line - strength indices and other galactic parameters ; we mainly focus on central properties . for each galaxy we defined three regions : the central circular aperture of 15 radius ( corresponding to the innermost nine single - lens spectra ) ; a so - called ` bulge region ' , defined as the region within which the light due to the fitted exponential disc falls below 50% of the total light , with the exclusion of the central aperture ; and a so - called ` disc region ' , a 3-thick elliptical annulus external to the ` bulge region ' 13 south of the centre . ] .
we averaged the spectra in these regions before spatial binning , and on the resulting spectrum calculated the line - strength indices , after emission removal : in several cases , even the disc region is affected only partially by the correction ; in addition to this , the bulge and disc region measurements are obtained averaging spectra from spatially separated areas , which helps in minimising the impact of the continuum problem in the fe5015 spectral region . ] .
the actual numbers for the definition of the bulge and disc regions ( semi - major axis of the ellipses , position angles and ellipticities ) were taken from our own photometric analysis and bulge - disc decomposition based on archival nir images ( ganda et al .
, in preparation ) .
there we model the galaxies as an exponential disc and a srsic ( 1968 ) bulge ( which often turns out to be very tiny , with effective radius smaller than @xmath2 5 ) and apply a two - dimensional decomposition from which we can estimate the bulge extension ( cfr .
@xcite for the method and ganda et al . , in preparation , for the actual analysis of the profiles ) . for clarity , in the top left map in figures [ maps1]-[maps9 ] we overplot in pink the elliptical contours defining the outer limits of the ` bulge region ' and of the ` disc region ' . in any case
, we want to warn the reader that the ` bulge ' and ` disc ' regions here introduced are mainly a tool for us to investigate radial variations , and do not refer to distinct components , since the so - called ` bulge ' is affected by a huge contamination from the disc , and an accurate decomposition is beyond the scopes of the present paper , and will be addressed in detail in the mentioned paper in preparation . + in table [ tableproperties ] we list the morphological type ( from the rc3 ) and the stellar velocity dispersion , with its associated error , measured on the central aperture spectra , since we will make use of these quantities in the following sections and figures ; the errors on the velocity dispersion are based on the scatter of the velocity dispersion values that we measure for the single - lens spectra within the central aperture , and do not take into account template mismatch .
table [ indicesvalues ] lists the index values calculated , galaxy by galaxy , on the central aperture and on the bulge and disc regions .
.hubble type ( rc3 through ned ) ; numerical morphological type ( from rc3 , indicated as ` t - type ' in some of the following pictures ) ; stellar velocity dispersion ( in km s@xmath12 ) and its associated error , also in km s@xmath12 , measured on the central aperture spectrum , for all galaxies . [ cols="^,<,^,^,^",options="header " , ] if we focus on the values obtained for the central apertures , we find a correlation between the best - fitting @xmath5 and the central velocity dispersion , as shown in fig . [ tauvssigma ] where we plot log(@xmath5 ) against log(@xmath1 ) .
we see that higher @xmath1 galaxies , which are larger and more massive , tend to have shorter star formation time - scales , more consistent with an instantaneous burst scenario , while at the low-@xmath1 end the smaller galaxies tend to have larger @xmath5 , indicating a star formation history spread over time .
therefore , the time - scale @xmath5 represents a physically meaningful parameter , parametrizing a smooth transition from ssp - like to constant sfr .
+ figure [ tauvsindices ] shows that rather tight correlations exist also between the central @xmath5 and the central line - strength indices .
these plots tell us that old , metal - rich galaxies ( low h@xmath0 , high fe5015 and mg_b _ values ) have shorter star formation time - scales , while young and more metal poor galaxies ( higher h@xmath0 , lower fe5015 and mg_b _ ) have a star formation history more extended in time : we find again a smooth transition between a ssp - like and constant star formation .
the central aperture values of @xmath5 used in figures [ tauvssigma ] and [ tauvsindices ] are listed in table [ tauvalues ] .
+ we also investigated the constant star formation history scenario , by selecting the corresponding option when using the programs from bruzual & charlot , building a grid with the model indices for the exponentially declining star formation rate and adding those for constant star formation rate , for the various metallicity values selected and for @xmath13 = 1 and age fixed to 10 gyr .
when applying our @xmath14 minimisation procedure , we found that the constant star formation was preferred in the regions where the fit with the exponential star formation only had selected the highest values allowed for the time - scale @xmath5 .
this confirms our interpretation of @xmath5 as a parameter describing a continuous transition from constant sfr to ssp .
+ we then repeated the analysis for different values of the fixed age of the evolved models , spanning the range 1 - 18 gyr and found that the @xmath5 values depend on the fixed age , given the existence of a strong degeneracy between age and @xmath5 , while the metallicities are more robust .
+ the main interpretative difficulty of this method is indeed due to the degeneracies in the parameter space , which are difficult to break with the few observed lick indices that we measure .
nevertheless , by inspecting the contours of @xmath15 in the ( @xmath5-metallicity ) plane , for several choices of the fixed age of the evolved models , we can in some cases exclude certain star formation scenarios in favour of others , even though we can not give a precise , quantitative estimate of the fitted parameters .
for example , in the case of ngc488 , if we look at the two - dimensional @xmath15 space ( metallicity , @xmath5 ) , for a fixed age of the evolved models ) , and look at the contours in the ( @xmath5-metallicity ) plane representing the 3-@xmath1 confidence level , based on the @xmath15 for one degree of freedom , we notice that it lies in a region far from the high-@xmath5 end , and this happens independently on the chosen values of the fixed age .
even if we perform the minimisation by allowing free age , @xmath5 and metallicity , and examine the projection of the @xmath15 contours in the ( @xmath5-age ) plane , we see that for all ages we can exclude large @xmath5 values at a high confidence level .
this must be regarded as an exercise , since we are not interested in fitting simultaneously age and @xmath5 , but it gives us insight on the impact of the degeneracies and of our assumptions on the robustness of our conclusions .
thus , for ngc488 we can safely state that we can exclude a constant star formation scenario in favour of an exponentially declining star formation rate , with a short time - scale @xmath5 ( closer to an instantaneous burst than to a constant sfr ) .
this is illustrated in the first column in fig .
[ chi2contours1 ] .
similar hints for the star formation history of other galaxies will be given , whenever possible , in appendix [ individuals ] , dedicated to the description of the single galaxies individually . for all galaxies ,
the @xmath15 contours for the fit at different fixed ages and with free age are presented in figures [ chi2contours1]-[chi2contours3 ] .
in this paper we presented the first integral - field measurements of line - strength indices and stellar population parameters for a sample of late - type spiral galaxies .
this work is based on data of 18 spirals spanning a range in type between sb and sd , observed with sauron , which provides measurements of the lick indices h@xmath0 , fe5015 and mg_b _ over a field of view covering the nuclear region of our objects .
+ here we summarise the main achievements of the paper : * we presented the two - dimensional maps of the line - strength indices and analysed the values extracted on a central aperture of 15 radius .
* we looked at the relations with morphological type : we found that late - type spirals do differ from earlier - type galaxies in terms of their populations , having globally higher h@xmath0 values and lower metal indices than ellipticals , lenticulars and early - type spirals .
this suggests younger ages and lower metallicities .
* we investigated the relations between indices and central velocity dispersion : late - type spirals do not obey the index - @xmath1 relations found for early - type galaxies : our galaxies lie above the h@xmath0 - @xmath1 and below the mg_b _ - @xmath1 relations determined for the elliptical galaxies , that correspond to old populations .
* we determined ssp - equivalent ages , metallicities and abundance ratios by comparison of our observed indices with models : ages are mainly young , metallicities low and abundance ratios close to solar ; the range spanned in the parameters is narrower than for early - type spirals .
* we investigated the possibility of recovering the star formation history : we assumed that the galaxies formed 10 gyr ago , and that they experienced an exponentially declining or constant star formation rate since then , and compared our indices with the evolved models of @xcite ; the fit returned an estimate for the @xmath4-folding time - scale for star formation @xmath5 , which parametrizes a smooth transition from a ssp - like star formation history to a constant sfr : small @xmath5 approximate an instantaneous burst and very large @xmath5 reproduce a constant sfr ; for the central apertures of our galaxies , we found a large range in @xmath5 values , from 1 to 15 gyr , covering basically the whole parameter interval that we explored . *
interestingly , we found a trend between the fitted @xmath5 and the central velocity dispersion , in the sense that galaxies with large @xmath1 tend to have small @xmath5 and galaxies with low @xmath1 tend to have large @xmath5 .
our observations and results nicely fit in the scenario summarised by @xcite : there are two modes of star formation ; one takes place in the extended discs of spiral and irregular galaxies , the other in compact gas discs in the nuclear regions of galaxies .
the first has been detected mainly via h@xmath16 surveys and it is strongly dependent on the morphological type : the star formation rate increases by a factor of @xmath2 20 going from sa to sc galaxies .
similar information is derived from observations of the uv continuum and broadband visible colours .
the second mode of star formation can be found in the circumnuclear regions ; the physical conditions in these regions are in many cases distinct from the more extended star forming discs .
the circumnuclear star formation is characterised by the absolute range in sfr , much higher spatial concentration of gas and stars and its burst - like nature @xcite .
circumnuclear star formation is largely decoupled from hubble type .
studies on the dependence of nuclear h@xmath16 emission in star forming nuclei as a function of galaxy type ( @xcite , @xcite , ho , filippenko , & sargent 1997 ) showed that the detection frequency of hii region nuclei is a monotonic function of type , increasing from 0% in elliptical galaxies to 8% in s0 , 22% in sa , 51% in sb , and 80% in sc - i m galaxies . among the galaxies with nuclear star formation ,
the h@xmath16 luminosities show the opposite trend ; the average extinction - corrected luminosity of hii region nuclei in s0 - sbc galaxies is nine times higher than in sc galaxies .
the conclusion of these studies , as summarised by @xcite , is that the bulk of the total nuclear star formation in galaxies is weighted towards the earlier hubble types , even though the frequency of occurrence is higher in the late types . according to @xcite , the central star formation accounts for 10 - 100% of the total sfr of spiral galaxies .
the highest fractions occur in early - type spiral galaxies , which typically have low sfrs in their outer discs @xcite .
+ the work presented in this paper supports this picture : if we put together the information gathered from the sample of e / s0s of paper vi , the sample of sa galaxies of paper xi and our own sample , and look at the behaviour along the hubble sequence , we find that early - type spirals show a larger range in age than both e / s0s and late - type spirals ; this is consistent with a scenario where the star formation in elliptical and lenticular galaxies is not very important , in early - type spirals it is dominated by short bursts and in late - type galaxies it is more quiescent .
this is also supported by our finding that big spirals tend to have an ssp - like star formation history , while for smaller ones the star formation history is better approximated by a constant over time .
+ the following steps in the analysis of the sauron data here presented and discussed will be a full investigation of the spatial information , with a more detailed study of the radial profiles ( here shown in appendix [ gradientssec ] ) and the gradients , and an interpretation of the connection between features in the line - strengths maps and in the morphological appearance .
the interdependencies between the populations and the kinematical structures @xcite will also be examined .
+ the main limitation of our work is the degeneracy in the parameter space in the different approaches we investigated .
our analysis relies on three data - points only ( the line - strength indices within the sauron spectral range ) , which are too few to break the degeneracies and constrain the fitted parameters .
therefore , it is urgent to complement the data presented here with spectra with a broader spectral coverage .
observations at other wavelengths -both shorter and longer- would help building a complete picture : both the uv and the nir spectral regions can provide population indicators much less affected by the model degeneracies found in the optical .
in addition , different sub - populations contribute to the integrated light in different spectral ranges , therefore a multi - wavelength study would be a step forward towards a full understanding of the formation and evolution of our objects .
we kindly acknowledge isabel prez for very useful discussion and comment .
we thank the referee for interesting comments that improved the paper .
the sauron - related projects are made possible through grants 614.13.003 and 781.74.203 from astron / nwo and financial contributions from the institut national des sciences de lunivers , the universit claude bernard lyon i , the universities of durham , leiden , groningen and oxford , the british council , pparc grant ` extragalactic astronomy & cosmology at durham 19982002 ' , and the netherlands research school for astronomy nova .
kg acknowledges support for the ubbo emmius phd program of the university of groningen .
mc acknowledges support from a pparc advanced fellowship ( pp / d005574/1 ) .
gvdv acknowledges support provided by nasa through hubble fellowship grant hst - hf-01202.01-a awarded by the space telescope science institute , which is operated by the association of universities for research in astronomy , inc .
, for nasa , under contract nas 5 - 26555 .
this project made use of the hyperleda and ned databases .
afanasiev v.l . ,
silchenko o.k . , 2005 ,
a&a , 429 , 825 allard e.l . ,
knapen j.h . ,
peletier r.f . ,
sarzi m. , 2006 , mnras , 371 , 1087 bacon r. , copin y. , monnet g.,miller b.w . , allington - smith j.r . , bureau m. , carollo c.m . , davies r.l . , emsellem e. , kuntschner h. , peletier r.f . ,
verolme e.k .
, de zeeuw p.t . , 2001 , mnras , 326 , 23 ( paper i ) bender r. , burstein d. , faber s.m . , 1993 , apj , 411 , 153 bernardi m. , sheth r.k . , annis j. , burles s. , 2003 , aj , 125 , 1882 bker t. , laine s. , van der marel r. p. , sarzi m. , rix h .- w . ,
ho l. c. , shields j.c . , 2001 ,
aas , 199 , 705 bker t. , laine s. , van der marel r.p . , sarzi m. , rix h .- w .
, ho l.c . , shields j.c . , 2002 , aj , 123 , 1389 bosma a. , casini c. , heidmann h. , van der hulst j.m . ,
van woerden h. , 1980 , a&a , 89 , 345 bruzual c. , charlot s. , 2003 , mnras , 344 , 1000 burstein d. , davies r.l . , dressler a. , faber s.m .
, lynden - bell d. , 1988 , in kron r.g .
, renzini a. , eds , assl vol . 141 . towards understanding galaxies at high
kluwer , dordrecht , p.17 buzzoni a. , gariboldi g. , mantegazza l. , 1992 , aj , 103 , 1814 cappellari m. , copin y. , 2003 , mnras , 342 , 345 cappellari m. , emsellem e. , 2004 , pasp , 116 , 138 carollo c.m . ,
danziger i.j .
, buson l. , 1993 , mnras , 265 , 553 carollo c.m . , stiavelli m. , de zeeuw p.t . , mack j. , 1997 , aj , 114 , 2366 carollo c.m . , stiavelli m. , 1998 , aj , 115 , 2306 carollo c.m . , stiavelli m. , mack j. , 1998 , aj , 116 , 68 carollo c.m . , 1999 ,
apj , 523 , 566 carollo c.m . , stiavelli m. , seigar m. , de zeeuw p.t . , dejonghe h. , 2002 , aj , 123 , 159 carollo c.m . , scarlata c. , stiavelli m. , wyse r.f.g . , mayer l. , 2006 , astro - ph/0610638 chabrier g. , 2003 , pasp , 115 , 763 cid fernandes r. , gonzalez delgado r.m . , schmitt h. , storchi - bergmann t. , martins l.p . , perez e. , heckman t. , leitherer c. , schaerer d. , 2004 , apj , 605 , 105 cid fernandes r. , gonzalez delgado r.m .
, storchi - bergmann t. , martins l.p . , schmitt h. , 2005 , mnras , 356 , 270 colless m. , burstein d. , davies r.l . , mcmahan r.k . ,
saglia r.p . ,
wegner g. , 1999 , mnras , 303 , 813 condon j.j . ,
condon m.a . , grisler g. , puscell j.j . , 1982 ,
apj , 252 , 102 cornett r.h , oconnell r.w . ,
greason m.r .
, offenberg j.d . ,
angione r.j .
, bohlin r.c . , cheng k.p .
, roberts m.s . , smith a.m. , smith e.p .
, talbert f.d .
, stecher t.p . , 1994 ,
apj , 426 , 553 davies r. l. , sadler e. m. , peletier r. f. , 1993 , mnras , 315 , 650 denicol g. , terlevich r. , terlevich e. , forbes d.a .
, terlevich a. , carrasco l. , 2005 , mnras , 356 , 1440 de vaucouleurs g. , de vaucouleurs a. , corwin h.g.jr . , buta r.j .
, paturel g. , fouqu p. , 1991
, third reference catalogue of bright galaxies , springer - verlag , new york ( rc3 ) devereux n.a .
, 1989 , apj , 346 , 126 emsellem e. , cappellari m. , peletier r.f . , mcdermid r.m . , bacon r. , bureaus m. , copin y. , davies r.l . , krajnovi d. , kuntscner h. , miller b.w . ,
de zeeuw p.t . , 2004 , mnras , 352 , 721 ( paper iii ) espada d. , bosma a. , verdes - montenegro l. , athanassoula e. , leon s. , sulentic j. , yun m.s . , 2005 , a&a , 442 , 455 faber s.m . ,
friel e.d . , burstein d. , gaskell c.m . , 1985 , apjs , 57 , 711 falcn - barroso j. , peletier r.f . , balcells m. , 2002 , mnras , 335 , 741 falcn - barroso j. , bacon r. , bureau m. , davies r.l . ,
de zeeuw p.t .
, emsellem e. , fathi k. , krajnovi d. , kuntschner h. , mcdermid r.m . ,
peletier r.f . ,
sarzi m. , 2006 , mnras , 369 , 529 ( paper vii ) fisher d. , franx m. , illingworth g. , 1995 , apj , 448 , 119 fisher d.,franx m. , illingworth g. , 1996 , apj , 459 , 110 ganda k. , falcn - barroso j. , peletier r.f .
, cappellari m. , emsellem e. , mcdermid r. , de zeeuw p.t , carollo m. , 2006 , mnras , 367 , 46 gonzalez delgado r.m . , cid fernandes r. , perez e. , martins l.p .
, storchi - bergmann t. , schmitt h. , heckman t. , leitherer c. , 2004 , apj , 605 , 127 gorgas j. , faber s.m . ,
burstein d. , gonzalez j.j . , courteau s. , prosser c. , 1993 , apjs , 86 , 153 ho l.c .
, filippenko a.v . ,
sargent w.l.w . , 1997 ,
apj , 487 , 579 idiart t.p .
, de freitas pacheco j.a . ,
costa r.d.d . , 1996 , aj , 112 , 2541 jablonka p. , martin p. , arimoto n. , 1996 , aj , 112 , 1415 jablonka p. , 2007
, astro - ph/0701169 jrgensen i. , franx m. , kjrgaard p. , 1996
, mnras , 280 , 167 jrgensen i. , 1999 , mnras , 315 , 607 keel w.c . , 1983 , apj , 269 , 466 kennicutt r.c . , 1983 , apj , 272 , 54 kennicutt r.c . , kent s.m . , 1983 , aj , 88 , 1094 kennicutt r.c . , 1998 , ara&a , 36 , 189 kormendy j. , kennicutt r.c . , 2004 , ara&a , 42 , 603 kuntschner h. , 2000 , mnras , 315 , 184 kuntschner h. , emsellem e. , bacon r. , bureau m. , cappellari m. , davies r.l .
, de zeeuw p.t . , falcn - barroso j. , krajnovi d. , mcdermid r.m .
, peletier r.f . ,
sarzi m. , 2006 , mnras , 369 , 497 ( paper vi ) laine s. , shlosman i. , knapen j.h . , peletier r.f . , apj , 567 , 92 longhetti m. , rampazzo r. , bressan a. , chiosi c. , 1998 , a&as , 315 , 251 macarthur l.a . , gonzalez j. , courteau s. , to appear in proc .
iau symp 241 , stellar populations as building blocks of galaxies , eds .
peletier and a. vazdekis , astro - ph/0701881 matthews l.d . , gallagher j.s . , 2002 ,
apjs , 141 , 492 mcdermid r.m . ,
emsellem e. , shapiro k.l . , bacon r. , bureau m. , cappellari m. , davies r.l .
, de zeeuw t. , falcn - barroso j. , krajnovi d. , kuntschner h. , peletier r.f .
, sarzi m. , 2006 , mnras , 373 , 906 ( paper viii ) moorthy b.k . , holtzman j.a . , 2006 , mnras , 371 , 583 natali g. , pedichini f. , righini m. , 1992 , a&a , 256 , 79 noordermeer e. , 2006 , phd thesis , univ
groningen ocvirk p. , pichone c. , lanon a. , thibaut e. , 2006 , mnras , 365 , 46 ocvirk p. , pichone c. , lanon a. , thibaut e. , 2006 , mnras , 365 , 74 panter b. , heavens a.f .
, jimenez r. , 2003 , mnras , 343 , 1145 peletier et al .
, 2007 , astro - ph/07042839 ( paper xi ) phookun b. , vogel s.n .
, mundy l.g . , 1993 , apj , 418 , 113 proctor r.n . , sansom a.e .
, 2002 , mnras , 333 , 517 reakes m. , 1979 , mnras , 187 , 525 rossa j. , van der marel r.p . , bker t. , gerssen j. , ho l.c . , rix h .- w . , shields j.c .
, walcher c .- j . , 2006 , aj , 132 , 1074 snchez blzquez p.,gorgas j. , cardiel n. , gonzlez j.j . , 2006 ,
a&a , 457 , 787 snchez
blzquez p. , forbes d.a .
, strader j. , brodie j. , proctor r. , 2007 , astro - ph/0702572 sarzi m. , falcn - barroso j. , davies r.l . , bacon r. , bureau m. , cappellari m. , de zeeuw p.t . , emsellem e. , fathi k. , krajnovi d. , kuntschner h. , mcdermid r.m .
, peletier r.f , 2006 , mnras , 366 , 1151 ( paper v ) schulman e. , bregman j.n . ,
brinks e. , roberts m.s . , 1996 ,
aj , 112 , 960 schweizer f. , seitzer p. , 1992
, aj , 104 , 1039 serra p. , trager s.c .
, 2007 , mnras , 374 , 769 srsic j.l . , 1968 , atlas de galaxias australes ( cordoba : obs .
astronom . )
soria r. , wong d.s .
, 2006 , mnras , 372 , 1531 stauffer j.r . , 1982 , apjs , 50 , 517 terlevich r. , davies r.l .
, faber s.m . , burstein d. , 1981 , mnras , 196 , 381 thomas d. , maraston c. , bender r. , 2003 , mnras , 339 , 897 thomas d. , maraston c. , bender r. , mendes de oliveira c. , 2005 , apj , 621 , 673 ulvestad j.s .
, ho l.c . , 2002 , apj , 581 , 925 vazdekis a. , casuso e. , peletier r. f. , beckman j.e . , 1996 , apjs , 106 , 307 vazdekis a. , 1999 , apj , 513 , 224 vollmer b. , huchtmeier w. , van driel w. , 2005 , a&a , 439 , 921 walcher c.j . , van der marel r.p . , mclaughlin d. , rix h .- w . , bker t. , hring n. , ho l.c .
, sarzi m. , shields j.c . , 2005 , apj , 618 , 237 worthey g. , faber s.m .
, gonzalez j.j . , burstein d. , 1994 , apjs , 94 , 687 worthey g. , collobert m. , 2003 , apj , 586 , 17 young j.s . , xie s. , kenney j.d.p .
, rice w.l . , 1989 , apis , 70 , 699 zaritsky d. , kennicutt r.c . , huchra j.p . , 1994 ,
apj , 420.87
we briefly investigated the radial variations of the line - strength indices and age , metallicity and abundance ratios . from the corresponding maps , we calculated radial profiles of h@xmath0 , fe5015 , mg_b _ and age , z and [ mg / fe ] from the one - ssp approach , by averaging the maps on concentric annuli of 08 width .
we present these profiles in figures [ profiles1]-[profiles3 ] .
these plots show that there is a wide variety of radial behaviours among late - type spirals : the line - strength indices can be roughly constant in radius ( h@xmath0 profile for ngc488 ) , rising outwards ( even very steeply , see for example the h@xmath0 profile for ngc5678 ) , rising inwards ( many fe5015 and mg_b _ profiles ) .
in some cases there are also complicated structures , with ` oscillating ' profiles : see for example the fe5015 profile for ngc864 or ngc2805 and several others
. some of the ` oscillating ' profiles might just reflect the noisy nature of the data for some low - surface brightness objects ( for example : the fe5015 profile for ngc2805 ) , but others are probably related to real variations of the indices within the galaxy . despite all this variety , we conclude that in most cases the h@xmath0 absorption strength increases with radius , while fe5015 and mg_b _ decrease . the fact that the metal lines show most often negative gradients
was already known for early - type galaxies , while in the case of h@xmath0 for elliptical galaxies the gradients are consistent with a flat relation with radius , or present a mild positive outwards radial gradient ( see @xcite and references therein ) . + the stellar population parameters show the same heterogeneous behaviours .
roughly , the age shows most often a complex behaviour with radius , reflecting features recognizable in the maps in figures [ maps1]-[maps9 ] , while the metallicity in many cases is decreasing , in agreement with @xcite , who measure mostly negative metallicity gradients for a sample of s0-sc galaxies ; in the majority of the galaxies , the metallicity profile resembles closely the aspect of the fe5015 profile .
looking at the abundance ratio profiles in more detail , we see that the central values are usually very close to solar ; many of the profiles stay approximately constant with radius , after an initial increase in the bulge - dominated region , which can be very tiny ( for a quantitative description of the bulges in these galaxies , we refer the reader to ganda et al .
, in preparation ) .
this is also consistent with the findings of @xcite , that we briefly summarised in section [ indexvsindexsec ] : they suggest that the bulges have positive or null gradients in abundance ratio with increasing radius , while the discs have solar abundance ratios . in early - type galaxies , instead , the radial profiles of abundance ratios are rather flat , as found by kuntschner et al .
( paper in preparation ) for the sauron e / s0 galaxies and by @xcite for a sample of 11 e / s0 galaxies ; the latter authors do find radial abundance ratio gradients , both positive and negative , but rather shallow .
ngc488 is an unbarred sb galaxy and it is the most elliptical - like ( featureless ) object within our sample .
ngc488 has regular stellar and gaseous kinematics , with regular rotation and velocity dispersion smoothly decreasing outwards @xcite .
the line - strength maps present high values for fe5015 and mg_b _ and low values for h@xmath0 over the central regions . as one can notice from the values
reported in table [ indicesvalues ] , this galaxy has the highest fe5015 and mg_b _ in the sample , and the lowest h@xmath0 .
the ages are mainly old and the metallicities high .
this is indeed the most metal - rich galaxy in the sample ( the central metallicity from the one - ssp approach is @xmath2 0.49 , then it decreases outwards ) .
this galaxy is well described by a one - ssp : as shown in fig .
[ chi2contours1 ] ( first column ) on the basis of our continuous star formation analysis , ngc488 is very unlikely to have experienced a constant star formation over time ; the observed indices are instead reproduced by a short starburst , approximating a ssp .
the time - scale of star formation @xmath5 is short over the whole field . from the profile in fig .
[ profiles1 ] we can see that the [ mg / fe ] profile is approximatively constant around the value @xmath2 0.2 : abundance ratios are slightly super - solar , confirming the fact that this galaxy resembles an elliptical more than the later objects in our sample .
other suggestions in this sense come from the position it occupies in the diagrams in figures [ corr_mgfe50 ] , [ corr_index ] , [ 1ssp_fullrange ] , where it lies in the region occupied by the ellipticals.=-2 ngc628 , known also as m74 , is a grand - design sc galaxy .
@xcite concluded that the star formation history of ngc628 varies with galactocentric distance ; natali , pedichini & righini ( 1992 ) suggested that the galaxy could be seen as an inner and an outer disc characterised by different stellar populations ; according to them , the transition between the two regions is located at @xmath2 8 - 10 kpc from the centre .
our observations of ngc628 are disturbed by the presence of a foreground star which falls close to the centre ( @xmath2 13 southern ) .
the stellar kinematics is characterised by slow projected rotation , and velocity dispersion decreasing in the central zones , indicating a cold central region , maybe an inner disc @xcite .
this galaxy is part of the ` control sample ' of non - active objects studied by @xcite , @xcite , @xcite , who analyse the nuclear spectrum , describing it as a mixture of @xmath17-@xmath18 yr - old and older stars . + as can be seen from the profiles reported in fig .
[ profiles1 ] , ngc628 has ages ( one - ssp approach ) below 3 gyr out to @xmath2 15 , and becomes older at larger radius ; the features are roughly in agreement with the above mentioned findings of cid - fernandes and collaborators ; the same trend in age is observed , on the common radial range , also by @xcite , on the basis of gmos data independently analysed .
the metallicity decreases outwards .
the ` bumps ' recognizable in the fe5015 , mg_b _ and ( but less pronounced ) metallicity profiles at @xmath2 13 from the centre are due to the foreground star . from our continuous star formation analysis , as shown in fig .
[ chi2contours1 ] ( second column ) , we can state that this galaxy is not described by a constant star formation scenario.=-2
ngc772 , also called arp78 , is an sb galaxy characterised by a particularly strong spiral arm ; it forms a pair at 33 with the e3 galaxy ngc770 . from the kinematical point of view
, ngc772 is characterised by a clear drop of the stellar velocity dispersion in the central zones , possibly corresponding to an inner disc , as suggested also by the anticorrelation between velocity and the gauss - hermite moment h@xmath19 @xcite . in the line - strength maps we can clearly see an extended central region with high fe5015 and mg_b _ and low h@xmath0
; indeed , from the central aperture values reported in table [ indicesvalues ] we can see that this galaxy is one of those with the highest metal indices , surpassed only by ngc488 . in the h@xmath0 map it is possible to see , surrounding the central depression , regions with higher values corresponding to the spiral arms .
a similar behaviour can be seen in the [ oiii]/h@xmath0 map presented by @xcite , which shows a structure resembling the spiral pattern .
estimating ages and metallicity by comparison with ssp models , we infer that ages are mainly young - intermediate and the central metallicity is among the highest in our sample .
the age and abundance ratio maps display the spiral arms structure as well , like the index maps .
[ chi2contours1 ] ( third column ) demonstrates that for this galaxy constant star formation can be ruled out at a high confidence level : high @xmath5 values are never likely . the @xmath5 map shown in fig .
[ maps2 ] ( for age fixed to 10 gyr in the fitting procedure ) presents a spatial structure that partially resembles the h@xmath0 map and the spiral arms pattern .
ngc772 is part of the sample studied in the series of papers by cid fernandes et al .
( @xcite , @xcite , @xcite ) , who are engaged in a project to examine the stellar populations of low - luminosity active galactic nuclei .
they classify this object as ` young - to ' , i.e. a transition object between a liner and an hii nucleus , with weak [ o i ] @xmath20 emission ; they find that the nuclear spectrum is dominated by @xmath17-@xmath18 yr - old stellar populations , with a contribution up to @xmath2 32% from populations younger than @xmath21 yr and a similar contribution also from older stellar populations ( @xmath22 10@xmath23 yr ) .
their data probe distances up to several hundred parsecs from the nucleus ; they measure in ngc772 significant radial variations of the stellar populations , with absorption line - strengths increasing in the inner 3 indicating a younger population in the very centre , especially referring to the ca ii k line ; the mg i absorption strength presents instead less prominent variations over the spatial range of their data .
our data cover a different spectral range and a different spatial extension , but we also concluded that the populations in the inner parts have intermediate ages , and that the age ( from the one - ssp analysis ) increases in the inner @xmath2 5 ( see fig . [
profiles1]).=-2 this is a barred sc galaxy where a nuclear radio source with the linear size of @xmath2 300 pc has been detected @xcite ; according to @xcite its hi profile is symmetric in velocity but asymmetric in intensity ; they also measure large - scale asymmetries in the two - dimensional kinematics , such as a warp .
the mentioned asymmetries occur in the outer parts , well beyond the extent of the sauron field .
the asymmetries could have been induced by a past encounter with a companion , passing outside the optical disc but within the hi disc ; the companion would have dispersed its stars and dark matter after the merging , consistent with the fact that ngc864 appears to be remarkably isolated .
+ our line - strength maps show a very sharp central dip in both fe5015 and mg_b _ ; in the region surrounding the dip , the fe5015 and mg_b _ values rise .
the h@xmath0 absorption also takes on low values over an extended central region ; as seen also in table [ indicesvalues ] , the indices measured on the central aperture of this galaxy are among the lowest in the sample . from the one - ssp analysis ,
given the structure of the line - strength maps , we derive very low metallicity and old ( @xmath2 11.22 gyr ) ages for the central aperture .
+ the analysis of the @xmath15 contours , in the continuous star formation approach , shows that this galaxy is unlikely to have experienced a constant star formation over time ( see fig .
[ chi2contours1 ] , fourth column ) .
= -2 ngc1042 is an scd galaxy forming a pair with ngc1035 at a separation of 22@xmath24 ( corresponding to 177 kpc ) .
it has a bright , small nucleus and otherwise low surface brightness .
our line - strength maps are quite patchy , due to poor s / n in the data and consequent heavy binning of the spectra .
the mg_b _ index takes on low values in the very centre , and higher ones around it .
h@xmath0 is high in the centre and low outside it , being flat over most of the map ; as can be seen also from the line - strength profiles in fig .
[ profiles1 ] , the line - strength indices stay approximately constant over most of the field , outside a central zone of radius @xmath2 3 .
the age of the stellar populations ( one - ssp analysis ) reaches a minimum in the centre .
the metallicity is everywhere low , and peaks in the centre .
+ the analysis of the @xmath15 contours for the central aperture , in the continuous star formation approach , shows that this galaxy is not well described by an ssp ( see fig .
[ chi2contours1 ] , fifth column ) .
the @xmath5 map for age fixed to 10 gyr ( see fig . [ maps3 ] ) shows a peak in the very centre , corresponding to the peak in the h@xmath0 and to the minimum in the mg_b _
maps.=-2 ngc2805 is an sd galaxy seen nearly face - on and it is the brightest member of a multiple interacting system containing also ngc2814 ( sb ) , ngc2820 ( sc ) at 13@xmath24 and ic2458 ( i0 ) .
hi has been detected ( @xcite , @xcite ) and there are claims that the outer hi layers are warped ( see for example @xcite ) .
the galaxy seems to be also optically disturbed , since the spiral arms appear to be broken up into straight segments .
+ the line - strength maps are rather patchy , displaying though a central peak in all three lines . by comparison of our measured indices with ssp models
, we can state that the ages are young - intermediate over most of the field ; that the metallicity peaks in the centre and that abundance ratios are super - solar . + the analysis of the @xmath15 contours , in the continuous star formation approach , shows that this galaxy is not very well described by an ssp , and that even a constant star formation rate can not be excluded at all ages ( see fig .
[ chi2contours1 ] , sixth column).=-2 ngc2964 , classified as sbc , forms a non - interacting pair with the i0 galaxy ngc2968 at 58 . as discussed by @xcite ,
the sauron data reveal that this barred sbc galaxy hosts an active galactic nucleus .
when we calculate the line - strengths , we find that the mg_b _ index assumes low values in an elongated central region , higher values around it , and low values again towards the edges of the field ; the fe5015 index presents a very similar behaviour ; the h@xmath0 map has instead an opposite appearance , with an extended inner region with low index values and a surrounding area with higher values ; in the very centre a shallow peak might be recognized .
these features are recognizable also in the profiles in fig .
[ profiles2 ] .
we notice also that the central region with low fe5015 and mg_b _ corresponds to the central depression in the emission - line [ oiii]/h@xmath0 map ( see fig . 5 in @xcite ) , possibly indicating ongoing star formation . converting the line - strength maps to age and metallicity via our comparison with the ssp models
, we find young - intermediate ages over most of the field , with a peak in the central region , and low metallicity everywhere , with a minimum in the central region , a positive gradient out to @xmath2 5 , and then a rather constant behaviour out to @xmath2 15 . from the continuous star formation approach
, we can not exclude at all ages the case of constant star formation , but the @xmath5 map for age fixed to 10 gyr ( in the fitting procedure ) presents a large central region with low values , possibly suggesting a @xmath2 5 gyr long starburst.=-2 this scd galaxy is very poorly studied in the literature .
ages of the stellar populations ( one - ssp approach ) are young - intermediate and metallicity low over most of the field . from the continuous star formation approach , we can not exclude at all ages the case of constant star formation ( see fig .
[ chi2contours2 ] , second column).=-2 this sbc galaxy is also poorly known .
both the fe5015 and mg_b _ indices peak in an extended central region and decrease towards the edge of the field ; the h@xmath0 map presents opposite behaviour .
the @xmath5 map ( for age fixed to 10 gyr in the fitting procedure ) displays low values over a large central area , possibly indicating a region where the bulk of star formation happened in a quite short starburst .
= -2 ngc3949 is an sc galaxy belonging to the ursa major cluster .
ngc3949 presents quite regular fe5015 and mg_b _ line - strength maps , peaking in an extended central region and decreasing rather smoothly moving outwards .
the h@xmath0 index is low in the central region and higher in the surroundings ; interestingly , it reaches a minimum in a zone south of the centre where the emission - line ratio [ oiii]/h@xmath0 is particularly depressed ( see relative figure in @xcite ) .
the values measured for the h@xmath0 index are the highest in the sample , as one can see from table [ indicesvalues ] and from fig .
[ profiles2 ] .
ages ( from the one - ssp approach ) are young ( @xmath2 1.0 gyr ) in the central region , somewhat higher around it and decreasing again towards the edge of the field . the metallicity is everywhere low , and decreasing outwards .
+ constant star formation can not be ruled out ; it is likely to characterise the star formation mode over most of the field ( see the @xmath5 map for age fixed to 10 gyr in fig .
[ maps5 ] : @xmath5 assumes high values ) . actually ,
when performing the fit allowing the code to choose among exponentially declining and constant star formation , constant is selected over most of the field.=-2 this sbc galaxy is characterised by very regular kinematics for both the stellar and gaseous components @xcite .
the line - strength maps also have a smooth appearance , with fe5015 and mg_b _ decreasing at larger radius and h@xmath0 increasing .
the h@xmath0 map seems to be asymmetric with respect to the centre , since the western side displays higher values than the eastern one .
the stellar populations are young over the whole field and the metallicity , among the highest in our sample ( the third after ngc488 and 772 ) , decreases moving outwards ; the abundance ratio is approximatively solar . the @xmath5 map for age fixed to 10 gyr in fig . [ maps6 ] shows a large central depression .
the analysis of the @xmath15 contours , in the continuous star formation approach , shows that this galaxy is probably not very well described by an ssp , nor by a constant star formation rate.=-2
ngc4102 is an sb galaxy , classified as a liner in the ned database ; it is known to be a powerful far - infrared galaxy @xcite and also to have a strong nuclear radio source @xcite .
@xcite classified it as one of the most powerful nearby starburst galaxies .
the signs of activity are detectable also in the sauron data : as noted by @xcite , the [ oiii ] maps trace the outflowing gas , with a region of very high gas velocity dispersion corresponding also to high values in the [ oiii]/h@xmath0 line ratio map , as expected for an active object . in that same region , we observe double - peaked line profiles for the [ oiii ] emission - lines , pointing towards the presence of different components in the emitting material . as for the line - strength indices ,
mg_b _ is low in an inner region , increases around it , and decreases again in the outer parts .
the fe5015 map is instead more centrally concentrated .
the map of h@xmath0 absorption looks more complex and not symmetric .
ages ( from our one - ssp analysis ) are young all over the field and metallicity is low .
neither in the line - strength nor in the age and metallicity map is there a clear connection with the mentioned region of activity . on the basis of our continuous star formation analysis ,
this galaxy is probably not very well described by an ssp .
= -2 also known as m99 , ngc4254 is a bright sc galaxy located on the periphery of the virgo cluster , at a projected distance of @xmath25 ( @xmath2 1 mpc ) from the centre of the cluster . its optical appearance is dominated by a peculiar one - arm structure : the arms to the northwest are much less defined than the southern arm .
this kind of spiral structure could be related to an external driving mechanism , but for ngc4254 there are no close companions .
phookun , vogel & mundy ( 1993 ) carried out deep hi observations of ngc4254 , detecting non - disc hi clouds , with velocities not following the disc velocity pattern and an extended low surface density tail northwest of the galaxy ; they interpreted these observational results as infall of a disintegrating gas cloud .
they also noticed that there is an hole in the hi emission at the centre of the galaxy , with a diameter of @xmath2 3 kpc .
vollmer , huchtmeier & van driel ( 2005 ) propose instead a scenario where ngc4254 had a close and rapid encounter with another massive galaxy when entering the virgo cluster : the tidal interaction caused the one - arm structure and the hi distribution and kinematics is due to ram pressure stripping .
recently , @xcite studied the x - ray properties of m99 , noticing that a phenomenon often associated with tidal interactions and active star formation -both documented- is the presence of ultraluminous x - ray sources ( ulxs ) .
the x - ray emission appears approximatively uniformly diffused across the inner @xmath2 5 kpc from the nucleus .
they do find a ulx , amongst the brightest observed , located at @xmath2 8 kpc southeast of the nucleus , close to the position where a large hi cloud seems to join the gas disc . from observations of the radial velocities , they suggest that the cloud is falling onto the galactic disc and try to build a link between this collisional event and the formation of the bright ulx .
our data cover a much smaller spatial extent than the mentioned observations , therefore a comparison is not possible ; within our field , both the fe5015 and mg_b _ maps show a regular structure with a central concentration ; at the northern edge of the field , both fe5015 and mg_b _ seem to increase again ; the h@xmath0 map has opposite behaviour , presenting a central depression .
the ages ( from our one - ssp approach ) are generally young and the metallicity , quite low everywhere , decreases slightly moving outwards and shows the same enhancement at the northern edge of the field that we noticed for the fe5015 and mg_b _ line - strength . the @xmath5 map ( see fig .
[ maps7 ] ) shows a central depression .
the analysis of the @xmath15 contours , in the continuous star formation approach , shows that ngc4254 is probably not very well described by an ssp , nor by a constant star formation rate .
= -2 this scd galaxy forms a pair with the scd galaxy ngc4504 , at a separation of 35@xmath24 ( corresponding to @xmath2 165 kpc ) .
the fe5015 and mg_b _ maps seem to be low in the very centre and higher outside it ; from the profiles in fig .
[ profiles3 ] one can see that outside the inner circle of @xmath2 3 radius , the line - strengths keep an approximately constant value , out to @xmath2 13 .
ages ( one - ssp approach ) are intermediate - young and metallicities very low , with respectively a maximum and a minimum in the centre . constant star formation can not be ruled out .
= -2 ngc4775 is an sd galaxy , for which very little is known from previous studies .
the metal lines ( fe5015 and mg_b _ ) seem to be centrally concentrated , while h@xmath0 is low over an extended central area , and higher in the surrounding region .
the maps are not symmetric with respect to the centre .
the galaxy is young in the nucleus and has older ages in an annular ( extended ) region around the centre ; the metal content is everywhere low , and peaks in the centre .
constant star formation can not be ruled out , and is selected over a large part of the field when allowing to choose among constant and exponentially declining star formation.=-2 ngc5585 is a barred sd galaxy ; together with ngc5204 ( sm ) , ngc5474 ( scd ) , ngc5477 ( sm ) , hoiv ( i m ) and m101 ( scd ) it forms the m101 group .
the fe5015 and mg__b _ _ indices assume values among the lowest in the sample ; h@xmath0 has instead a value among the highest ones ( see table [ indicesvalues ] ) .
the metallicity retrieved from the one - ssp analysis is everywhere very low . in the continuous star formation approach ,
constant star formation can not be ruled out ; indeed , when allowing to choose among exponentially declining and constant star formation , in the @xmath14 minimisation , for age fixed to 10 gyr the constant star formation is selected ( for the central aperture).=-2 ngc5668 is another sd galaxy , which has a high rate of star formation , as indicated by its large far - infrared and h@xmath16 luminosities ( @xcite , and references therein ) .
the metal lines ( fe5015 and mg_b _ ) appear to be centrally concentrated , while h@xmath0 has a more complex appearance .
the stellar populations are young and metal poor all over the field ; the age rises out to @xmath2 8 and then decreases ( see fig .
[ maps9 ] for a full view of the spatial structure of the age map and fig .
[ profiles3 ] for the azimuthally averaged profiles ) .
constant star formation can not be ruled out at all ages , and is selected over a large part of the field when allowing to choose among constant and exponentially declining star formation.=-2 this is a barred and very dusty sb galaxy . fe5015 and mg_b _ have a quite regular and centrally concentrated appearance ; h@xmath0 is depressed in the inner region and increases in a annulus - like structure around it . in the very centre h@xmath0
has a local maximum , as shown also in the azimuthally - averaged profile in fig .
[ profiles3 ] .
ages ( one - ssp approach ) are young over most of the field and the metallicity , rather low everywhere , decreases moving outwards .
the @xmath5 map at age fixed to 10 gyr ( see fig . [ maps9 ] ) shows a large central depression .
the analysis of the @xmath15 contours shows that ngc5678 is probably not very well described by an ssp , nor by a constant star formation rate . like ngc628 and 772 ,
this galaxy is part of the sample studied in the cid fernandes et al papers ( @xcite , @xcite , @xcite ) ; following their classification , it is a ` young - to ' galaxy , with weak emission in the [ o i ] @xmath20 line ; according to those authors , the nuclear spectrum is heavily dominated by @xmath17-@xmath18 yr old stellar populations , with negligible contribution from stars younger than @xmath21 yr .
they detect radial variations of the absorption indices , particularly in the ca ii k line , but less clearly evident than in the case of ngc772 .
+ # 1 @font @ @ plus .1pt 40004000 = 1000 = | we present absorption line - strength maps for a sample of 18 sb - sd galaxies observed using the integral - field spectrograph sauron operating at the william herschel telescope on la palma , as part of a project devoted to the investigation of the kinematics and stellar populations of late - type spirals , a relatively unexplored field .
the sauron spectral range allows the measurement of the lick / ids indices h@xmath0 , fe5015 and mg_b _ , which can be used to estimate the stellar population parameters .
we present here the two - dimensional line - strength maps for each galaxy . from the maps , we learn that late - type spiral galaxies tend to have high h@xmath0 and low fe5015 and mg_b _ values , and that the h@xmath0 index has often a positive gradient over the field , while the metal indices peak in the central region .
+ we investigate the relations between the central line - strength indices and their correlations with morphological type and central velocity dispersion , and compare the observed behaviour with that for ellipticals , lenticulars and early - type spirals from the sauron survey .
we find that our galaxies lie below the mg - @xmath1 relation determined for elliptical galaxies and that the indices show a clear trend with morphological type .
+ from the line - strength maps we calculate age , metallicity and abundance ratio maps via a comparison with model predictions ; we discuss the results from a one - ssp ( single stellar population ) approach and from a two - ssp approach , considering the galaxy as a superposition of an old ( @xmath2 13 gyr ) and a younger ( age @xmath3 5 gyr ) population .
we confirm that late - type galaxies are generally younger and more metal poor than ellipticals and have abundance ratios closer to solar values .
we also explore a continuous star formation scenario , and try to recover the star formation history using the evolutionary models of bruzual & charlot ( 2003 ) , assuming constant or exponentially declining star formation rate ( sfr ) . in this last case , fixing the galaxy age to 10 gyr , we find a correlation between the @xmath4-folding time - scale @xmath5 of the starburst and the central velocity dispersion , in the sense that more massive galaxies tend to have shorter @xmath5 , suggesting that the star formation happened long ago and has now basically ended , while for smaller objects with larger values of @xmath5 it is still active now .
[ firstpage ] galaxies : bulges - galaxies : evolution - galaxies : formation - galaxies : kinematics and dynamics - galaxies : spiral -galaxy : structure |
gamma - ray bursts ( grbs ) are a kind of astrophysical phenomenon so luminous in the universe that their isotropic energies of @xmath0 are usually considered coming from extremely relativistic outflows with bulk lorentz factors as high as @xmath1 .
the connection between long - duration grbs and broad - line sne ib / c ( low - metallicity ) has been supported by some pieces of observational evidence @xcite .
the recently launched fermi gamma - ray space telescope with the on - board gamma - ray burst monitor ( gbm ) and large area telescope ( lat ) instruments ( @xcite ) in conjunction with the _ swift _ narrow field instruments @xcite have opened a new era of understanding physical mechanisms of gev photon emission in very energetic grbs and their relation to lower - energy afterglow emission . up to now ,
several novel yet tricky features have appeared during the whole period of observations and their complexities have challenged the anterior established theoretic models .
the first feature is that the sub - mev and gev photons observed by gbm and lat respectively behave distinctive spectral and temporal properties .
the gbm light curves are nearly flat during the main episode of the prompt emission , then drop extremely rapidly [ e.g. @xmath2 for grb 080916c @xcite ] , and eventually cease abruptly .
this can also be validated from @xmath3 of all bursts detected by the _ swift _
satellite @xcite . for the lat light - curves ,
however , a rise appearing in the early few seconds ( @xmath4 ) was followed by a single power - law decay without any cutoff up to hundreds of seconds after the gbm trigger , till below the monitor sensitivity , and decay of the late lat emission was much shallower than the gbm - detected counterparts at the same times @xcite .
furthermore , the lat emission usually lags the gbm emission from a fraction of seconds to a few seconds .
in addition , the spectral slopes of the gbm and lat emissions are often different , e.g. , the gbm data can be fitted with a band spectrum that is composed of two smoothly - joining power laws , while the lat data are often fitted by a power law with a slope intermediate between the two slopes of the gbm fit .
these properties seem to indicate that the high - energy emission and low - energy emission detected by lat and gbm have different origins .
the temporal properties of the lat emission have been studied , and some explanations differing in the source s dominant component have been proposed .
a prevalent explanation is particle - dominated models .
one inclination is the leptonic interpretation . because of their distinctive light curve properties we mentioned above , it is highly probable that the mev photons may be of an internal origin , while the gev photons may be generated via synchrotron emission of electrons accelerated by an external forward shock that also leads to lower - energy afterglow emission @xcite .
@xcite studied the klein - nishina effect on the high - energy afterglow emission and found that at early times such an effect strongly suppress the inverse compton scattering of those electrons that produce the high - energy emission .
thus synchrotron radiation of the electrons becomes a plausible mechanism .
this conclusion was independently drawn by @xcite , who suggested that the lack of a thermal component in the gbm spectrum of grb 080916c is consistent with a relativistic poynting - flux - dominated outflow .
the explanation of synchrotron radiation of the electrons accelerated by a forward shock is fairly natural to account for both the observed delay of the @xmath5 mev photons and long lasting of their emission .
however , a thermal component deviating from the smooth band spectrum function in some bursts ( i.e. grb090510 , grb090902b , grb090926a ) seems to be beyond a prediction of what is often invoked from the external shock model @xcite .
moreover , a rapid rise during the first few seconds ( e.g. @xmath6 of grb080916c ) is hard to be well explained within such a framework @xcite .
therefore , in addition to the hypothesis of a separate origin , @xcite assumed that gev and mev photons may come from the same region , but the onset of the high - energy may result from anisotropic inverse compton scattering of an optical - thin expanding cocoon , delayed compared with the mev emission .
nevertheless , in their calculation , this up - scattering cocoon is so short - lived that it could not account for the whole high energy emission .
another approach is the hadronic scenario .
@xcite suggested that the mev and gev photons could be interpreted as the radiation of accelerated electrons and cosmic ray protons respectively , as well as the delay between them could come from different cooling time scales in a highly magnetized shock . in their framework , the model of grb 080916c is plausible only when @xmath7 and the jet opening angle @xmath8 .
alternatively , @xcite considered another possibility of the photomeson cascade and proton synchrotron models , and provided their constraint in grb090510 .
notwithstanding , to reproduce the extra component around gev with these models , the isotropic - equivalent proton injection luminosity is required to be larger than @xmath9 .
such a large proton luminosity is a challenge for the hadronic models .
all the works mentioned above inspire us to consider a plausible structured outflow , in which the bulk lorentz factor of the initial shells tends to be lower than that of the late shells .
this energy accumulation , therefore , would lead to an early rapidly rising light curve and transient soft to hard spectrum .
the second feature of the three long grbs with lat data is that some humps or flares appear in the light curves of the @xmath10 s low - energy afterglows ( @xcite ) . neglecting this feature ,
@xcite fitted the multi - band light curves of these grbs by assuming that a relativistic external shock sweeps up an interstellar medium , and obtained reasonable physical parameters or parameter spaces accordingly .
although their fittings somehow favor the external shock model , humps ( or sometimes flares ) have indeed been observed on the late - time optical and x - ray afterglow light curves , which do not completely accord with a simple power - law but instead the decay after the humps is shallower than that before the humps .
these observations call for a more meticulous consideration of the external shock model .
it is noted that a density - jump medium proposed by @xcite has provided us with a clue for some optical and x - ray humps and well fitted to several bursts ( @xcite ) . in this scenario ,
a relativistic jet first expands in a stellar wind , subsequently encounters a density jump , and finally expands in a homogeneous medium .
this interaction can produce an observed light - curve bump . in this paper
, we show that the early - time gev emissions together with the late - time radio , optical , and x - ray afterglows of grbs 080916c , 090902b , and 090926a can be understood as being due to synchrotron emission from an external forward shock caused by structured ejecta propagating in a wind bubble jumping to a homogeneous density medium . in sec.[sect :
obs ] we include a set of observed broadband ( lat , xrt , uvot ) data on these three luminous long bursts .
then , our model is set up in sec.[sect : model ] .
in particular , the structured ejecta can well explain the universal early rising feature of the gev emission from these bursts , and the density - jump medium can account for some certain plateaus and flares in the late afterglows . additionally , for the sake of verifying whether or not the lower energy ( x - ray , optical , radio ) emission originates from the same source as the higher energy ( @xmath11100 mev ) emission
, we discuss the effect of synchrotron self - compton ( synchrotron self - absorption ) on the high energy ( radio ) emission , which , in the constrained parameter space estimated analytically from xrt and uvot light curves , is proved in sec.[sect : fit ] to have a small contribution to the flux density during the observed period . in this section ,
we find a reasonable set of parameters valid for most of the late afterglows .
our conclusions concerning a plausible central engine and ambient environment of bright , long grbs are discussed in sec.[sect : con ] . throughout this work ,
we adopt the convenience @xmath12 in units of cgs .
among the 19 observed fermi - lat grbs during the first 2 years operation , grb080916c , grb090902b and grb090926a are 3 typical brightest long grbs with abundant of spectra information @xcite .
this burst , located at redshift @xmath13 @xcite , is the first grb detected by fermi with high significance of photons at energies @xmath14 gev .
the isotropic energy emitted from prompt emission can be estimated as @xmath15 @xcite . at 00:12:45.613542 on 16 september 2008 @xcite ,
this grb was triggered on by gbm with the duration @xmath16s . before @xmath17s
, the lat light - curve shows an extremely steep rise @xmath18 , followed by a simple power law decay @xmath19 until @xmath20s @xcite .
x - ray and optical photons were detected by xrt and uvot since 17:11:28 16 september 2008 ( @xmath21ks ) .
a steep decay ( to @xmath22ks ) continued with a plateau ( to @xmath23ks ) goes to a slightly shallower decay without break until 1.3ms from the trigger @xmath24 . during the period of @xmath21ks@xmath25ks , the simple power law light curve and spectrum evolution
show the flux density at x - ray band , @xmath26 , as well as for the optical band , @xmath27 @xcite .
besides , agile , rhesso , integral , konus - wind , and messenger all provide a plentiful of data information for the late afterglow @xcite .
many groups have studied this grb with the external shock model ( @xcite ) , and suggested that the late time afterglow data can be used to extrapolate the early lat data as well as be predicted from it , under the preference of circumburst density stratification .
this burst is an exceptional case with redshift @xmath28 , whose speciality lied in both phases of prompt emission and early afterglow @xcite . unlike majority of the other grb events , excess emission exhibited in both low ( @xmath29kev ) and high ( @xmath30mev ) band during the prompt emission
moreover , the soft - hard - soft spectral evolution indicates two components : band function peaking at @xmath31kev + simple power - law with photon index @xmath32 @xcite .
the isotropic energy release @xmath33 @xcite .
there are some explanations for origin of the high - energy power - law emission component observed in the lat energy range ( which accounts for @xmath34 of the total 10kev to 10gev fluence ) , including the hadronic origin [ either proton synchrotron radiation @xcite or photohadronic interactions @xcite ] , or thermal emission from the jet photosphere @xcite .
this burst was triggered and located at 11:05:08.31 on september 2009 , by the fermi - gbm @xcite and suzaku - wam @xcite with multipeaked duration @xmath35 .
after a rapid rise until @xmath36s , the lat band light - curve decays up to @xmath37ks with @xmath38 , including an energetic photon detected as high as @xmath39 gev within the prompt emission phase and another @xmath40 gev at @xmath41s @xcite .
the _ swift _ xrt ( @xcite ) and uvot @xcite began concurrently target of opportunity of the field of grb090902b s fading source from 23:36 on 2 september 2009 and copious of data have been in hand from @xmath42ks . a steep decay ( to @xmath43ks ) simultaneously ended with a slight rise in both bands .
it is highly possible that jet break occurred at @xmath44ks @xcite . during this whole period ,
the x - ray spectrum evolved as @xmath45 , and for the optical band , @xmath46 .
in addition , vla began to observe the afterglow since 3 september 2009 @xcite until 5 months later at 8.5ghz and 4.8ghz . using the forward - reverse shock and constant density model
, @xcite suggested the afterglow of this burst is better fitted in x - ray and radio bands than in the optical / uv band , except for the first point at @xmath47s .
parameter constraints without taking lat data into account suggest a low circumburst density and a large kinetic energy , which is possible in the narrow opening angle @xmath48 .
later , @xcite have provided a meticulous discussion about the two - component forward - reverse external shock in a monotonic circumburst environment and their calculation was well fit to the four observed bands , except for the very early rising part of the lat light curve .
this burst is the grb detected at @xmath49 by fermi with photons as high as @xmath50 gev at 26s after the trigger @xcite .
the prompt emission time - integrated flux indicates its isotropic emission energy of @xmath51 @xcite . since the trigger at 04:20:26:26.99 on 26 september 2009 , gbm and lat started operation and found that multiple pulses with total duration @xmath52 can be well modeled by a band function spectrum in the prompt emission , along with the suzaku - wam @xcite , konus - wind @xcite , and rt-2 on coronas - photon @xcite .
the light curve rose slightly for quite a few seconds ( to @xmath53s ) , and decayed @xmath54 .
a fading x - ray counterpart was observed by xrt since 17:17 on 26 september ( @xmath55ks ) @xcite , then prompt , smarts detected photons in several optical filters but no radio ( 5.5ghz ) source under the limit of 1.5mjy was detected up to 1 october 2009 . as shown on the light curve
, a plateau ( @xmath56ks @xmath57ks ) overlaps the simple power law decay that is slightly steeper than the fading supposed to be extrapolated from the early trace .
no break up to 100ks from the trigger @xmath24 was observed . during the period from @xmath55ks to @xmath58ks , two variabilities , are suggested to be flares ( the first , at @xmath59 with @xmath60 ; the second is slight , at 195ks-260ks , with @xmath61 ) overly on the simple power law light curve in both x - ray band ( @xmath62)and optical band ( @xmath63 ) @xcite .
@xcite and @xcite studied the later afterglow ( x - ray , optical ) of this burst , provided a certain parameter space , confirmed the second flare , and indicated that the jet break occurs around @xmath64d .
however , due to lack of radio data , the constraints can not be narrowed down . meanwhile , as for a common rebrightening plateau in 5 optical bands , they suggested a density - jump circumstance ( e.g. , @xcite ; @xcite ) or a smooth injection of energy into the forward shock from the central engine ( e.g.,@xcite ) , and called for a detailed analysis for the early higher energy emission .
the behavior of the grb afterglows shed light on the external shock model @xcite .
a collimated ( @xmath65 ) , ultrarelativistic outflow of matter and/or radiation is driven by a central engine .
some certain dissipative processes within the outflow give rise to the prompt gamma - ray emission , with a fraction @xmath66 of the total relativistic energy converted to high - energy radiation . a self - consistent result of the multiband ( x - ray , optical , radio ) afterglow data coordinated with the prompt emission relics producing @xmath67 photons , may support the external shock origin of lat photons , which is the same as the later lower - band emission .
meanwhile , due to the difference between the light curve slope of lat and gbm , we take no consideration of the gbm emission in this piece .
moreover , the particularities of grb 090902b we mentioned above ( soft - hard - soft spectral evolution , flares ) may evoke doubts about the conventional external shock + monotonic circumstance model ( we will discuss it in sec.[sect : con ] ) , but enlighten our deliberate consideration on amendment to structured ejecta sweeping up the density - jump medium .
energy can be injected by a central engine continuously during a period ( majority of which is within the first main pulse of prompt emission , before @xmath68 ) .
we here suggest , within the framework of the collapsar model , an increase of the lorentz factor of the ejecta in active time of the central engine may due to two reasons .
first , angular momentum of the accreted fall - back matter would spin up the central compact object and thus its rotational energy loss could give rise to an increase of the ejecting luminosity .
second , earlier - ejected shells , when breaking through the stellar envelope , may suffer from more massive baryon contamination , and thus later - ejected shells may propagate in a tunnel ( the later the cleaner ) . therefore , the lorentz factor of the ejecta head may increase with time . for simplicity ,
we assume the bulk lorentz factor of the front materials distribute as a power - law function of time : @xmath69 for @xmath70 . if @xmath71 , the lorentz factor of the shocked matter increases , being due to energy injection .
we denote @xmath72 as the accumulated mass ejected by the central engine at some certain time , and @xmath73 as the bulk lorentz factor of a blast wave at this certain time , which is defined when the mass of the surrounding matter swept up by the blast wave , @xmath74 , is equal to @xmath75 . after this time
, the bulk lorentz factor presents an obvious deceleration , contributing to a peak of the lat light curve .
this time is called the deceleration time @xmath76 , which is assumed in this paper to be around the stop time of effective energy injection . this assumption is reasonable because energy injection to the forward shock is effected by the circumburst medium .
thus , both the apparent energy - injection time @xmath68 and the deceleration time @xmath76 should be shorter than the central engine ceasing time @xmath77 .
the shock propagates a distance @xmath78 during a small observed time @xmath79 . assuming the proton number density of a medium @xmath80 , thus for a homogeneous medium ( @xmath81 ) , @xmath82 ( @xcite ) or for a wind medium ( @xmath83 ) , @xmath84 ( @xcite ) . in both cases ,
we obtain @xmath85 the deceleration radius can be written as @xmath86 therefore , @xmath87 . according to @xcite , in order to check whether the the unshocked ejecta is assumed to be equal to that of the shocked matter , let @xmath88 represent the thickness of the shell , and @xmath89 be the density ratio between the preshock fluid in the shell and in the circumburst surrounding ( @xmath83 ) , given by @xmath90 this may bring about a correction for the gev rising part under different conditions , i.e. , in the thick shell case , @xmath91 , a reverse shock is relativistic , the lorentz factor of the shocked fluid should be changed and the contribution of the reverse shock to the flux density should be considered , while in the thin shell case , @xmath92 , this change is negligible .
however , no matter how the thickness of the shell is , @xmath93 . in the following analytical calculations , we simplify that @xmath94 ( which can be observed from the figures of temporal flux density ) and assume that the self - similar condition is established after the deceleration time . before the deceleration time , energy is continuously injecting into the former shells .
the mass ejected later catches up with the earlier ones , and @xmath95 can be simplified as the bulk lorentz factor of these materials combination . in the very beginning , it is possible that the bulk lorentz factor is too low for majority of high - energy gamma rays to escape from the mev background . according to predominate hypothesis ( @xcite ) , only when the optical depth of the @xmath96-ray absorption @xmath97 , the bulk lorentz factor is large enough to radiate almost completely . in this piece , we use sketchy in @xcite to check whether the absorption play a dominate role in certain parameter spaces . because of the association of long grbs with star forming regions , it is highly possible that massive stars are still embedded in a cloud when giving birth to grbs .
low density bubbles are created by stellar winds from grb progenitors , whose sizes and densities strongly depend on the initial ambient density
. therefore , a density jump occurs at the boundary between the wind bubble and the outer cloud , which may result in a bump / flare and shallow decay of the light curve during the later period @xcite .
it has been usually assumed that the afterglow emission from the blast wave is nearly adiabatic ( @xcite ) , so the total kinetic energy of the relativistic shock is constant .
however , if we take radiative energy loss into consideration , the radiation efficiency of the blast wave can be given with the electronic fraction @xmath102 by @xcite @xmath103 in such a situation , energy loss is significant in the fast cooling and gradually dwindle to the quasi - adiabatic case ( i.e. , the slow cooling phase ) .
hereafter , a denotation @xmath104 can analytically result in the bulk lorentz factor evolution , @xmath105 consequently , the doppler factor becomes @xmath106@xmath107 ( where @xmath108 is the emitting latitude angle ) .
when the blast wave reaches the density boundary , the observed light curve will not behave as an abrupt jump but a smooth rise or plateau instead , due to the curvature effect ( @xcite ) .
although this is trifle to be accomplished in analytic modeling , a simplified power law distribution of medium density in the period of @xmath109 can account for the same light curve feature ( see fig .
[ fig : parameter].a ) .
succeedingly , the blast wave radius can be written as @xmath110 in the comoving frame , with an electronic fraction @xmath102 and magnetic fraction @xmath111 of the post - shock thermal energy density , which may either vary or not with different media .
denote @xmath112 , the magnetic field intensity @xmath113 .
the cooling , minimum , and maximum electron lorentz factors are @xmath114=6\pi m_ec ( 1+z)(\sigma_tb^2\gamma t)^{-1}/[1+y(\gamma_c)]$ ] , @xmath115 , and @xmath116 , while the synchrotron characteristic frequencies @xmath117 , @xmath118 , @xmath119 can be calculated by @xmath120 $ ] ( where @xmath121 ) , where @xmath122 is the compton factor for electrons with the lorentz factor @xmath123 , defined as the ratio of the self - inverse compton ( ssc ) to synchrotron emission ( @xcite ) .
@xmath124 when klein - nishina ( kn ) effects are unimportant , i.e. , the ic scattering of @xmath123 electrons with synchrotron photons are in the thomson scattering regime , @xmath122 can be simply derived by @xcite as @xmath125 , which is independent with @xmath123 .
however , for high - energy electrons whose kn effect becomes important , @xmath122 depends on @xmath123 and can be expressed as analytical solutions in different cases from @xcite . the total number of shocked accelerated electrons @xmath126 .
the peak spectral power of a single electron is @xmath127@xmath128 , and the peak flux density at certain time of the afterglow is @xmath129 . here , the luminosity distance @xmath130^{-1/2}dz$ ] ( here we adopt a standard @xmath131 cosmology with @xmath132 , @xmath133 , @xmath134 , @xmath135 @xcite ) . for a lower energy band ( e.g. , radio )
, synchrotron self - absorption may play a crucial role . the self - absorption optical depth @xmath136 or ( @xmath137 , according to the definition @xmath138 ,
becomes @xmath139 and @xmath140 , where @xmath141 @xcite .
succeedingly , the self - absorption frequency @xmath142 can be written as @xmath143 then , the ssa power - law light curve yields @xcite @xmath144 here @xmath145 , @xmath146 .
@xmath147 when @xmath148 , while @xmath149 , @xmath150 .
it is noted that before the optical depth becomes thin , the annihilation effect of high - energy photons can affect their flux , so the observed flux can be estimated as @xmath151 .
after @xmath77 , @xmath152 . for a high energy band , synchrotron self - compton ( ssc )
radiation might contribute to the observed flux , the characteristic frequencies @xmath153 , @xmath154 , @xmath155 , @xmath156 can be calculated by @xmath157 ( where @xmath158 , and @xmath159 .
thus the ssc power - law light curve yields @xcite @xmath160 here @xmath161 , @xmath162 , @xmath163 , @xmath162 .
@xmath147 when @xmath164 , while @xmath165 , @xmath150 .
however , due to the kn effects on optically thin spectrums , equations and should be amended as different cases mentioned in equations of @xcite .
this calls for the definition of electron kn lorentz factor : @xmath166 $ ] ( @xmath167 ) .
we have , if the synchrotron photons emitted by electrons with lorentz factor larger than @xmath168 [ i.e. with frequency @xmath169
$ ] , they can not be upscattered efficiently by electrons with lorentz factor larger than @xmath170 ( which is above the kn limit ) . in the case of a jet
, a simple assumption is that the jet decelerates without sideways expansion .
after the jet break time @xmath171 at which the jet angle @xmath172 ( where the jet axis is assumed to be along the line of sight ) , the observed flux @xmath173 . comparing the flux density estimated analytically above with the observational multi - band values , certain physical parameters can be constrained . whether the parameter space is reasonable or not can on some extent verify the feasibility of our external shock model .
based on the analytical model we mentioned above , our procedure of further narrowing down the physical parameter space in the self - similar phase includes the following steps : in the wind environment : \(1 ) according to the expression of @xmath174 in different spectral regimes given by @xcite and the observed high - energy , x - ray and optical temporal and spectral indices , @xmath175 can be roughly estimated ( see tab .
[ tab : data ] ) .
\(2 ) for simplicity , @xmath176 in the fast cooling regime and then @xmath177 decays slowly ( where we denote @xmath178 as the time when @xmath179 , and obtain that @xmath177 decays as @xmath180 in the slow cooling regime ( i.e. , setting @xmath181 , we have , after @xmath178 , @xmath182 , then @xmath183 and @xmath184 , and equation provides @xmath185 , as long as @xmath186 , @xmath187 ) .
\(3 ) because of the definition for the deceleration time [ set @xmath83 in equations and ] , @xmath73 can be expressed with @xmath188 , @xmath189 and @xmath190 ( or @xmath191 ) .
this can estimate whether the shell thickness is thin enough to neglect the reverse shock s contribution to the total flux density .
\(4 ) hereafter the free parameters are @xmath188 , @xmath192 , @xmath193 , and @xmath190 ( or @xmath194 ) .
the observed flux density in different wavebands ( radio , optical , x - ray , @xmath11100 mev ) can be combined and eventually simultaneously be solved . in this paper ,
the temporal indices of the x - ray and optical light curves concurrently evolve according to equation and thus @xmath111 can therefore be expressed with @xmath188 , @xmath195 and @xmath190 ( or @xmath194 ) .
\(5 ) to justify whether the early high energy lat photons come from the same region as the late afterglow , what we need to do is to replace @xmath111 in the form of @xmath188 , @xmath196 and @xmath190 ( or @xmath197 ) from step ( 4 ) . to confirm whether ssc make a significant contribution to the gev afterglow or klein - nishina effect changes the synchrotron and ssc spectra , we need to discuss the value of @xmath198 in the slow cooling regime or @xmath199 in the fast cooling regime @xcite , therefore @xmath200 can be estimated @xcite .
if no solution is found in the range of ( @xmath201 and @xmath202 ) , the @xmath11100 mev photons may come from the other component , vise vesa .
\(6 ) from discussions about the parameter space of ( @xmath203 and @xmath188 ) according to observational statistics , one can obtain several best fittings .
in addition , the ratio @xmath101 of the wind and homogeneous medium density , @xmath102 and @xmath111 at early and late times may have slight changes , which can also be legally adjusted in practise .
we put the above procedure into operation , and best fitting parameters are shown in tab . [
tab : fitting ] .
the evolution of characteristic frequencies for each burst is expected as follows . here
our lat data are taken from @xcite and x - ray data were reported by _ swift _ ( @xcite ) . from the lightcurves , we found that from @xmath76 on , @xmath204 , @xmath205 .
according to the analytical expression from @xcite , before the blast wave reaches the density jump , gev photons at early decaying time ( e.g. , @xmath206s ) has two possibilities : @xmath207 while the lower energy bands ( x - ray , optical ) at @xmath208s , may lie in one of the following slow cooling cases : @xmath209 for 090902b radio band , at @xmath210 when @xmath211 ( i.e.@xmath212 ) , it is similar to the optical spectrum and lightcurve . while @xmath213 , @xmath214 our optical photometry of this burst is taken from @xcite .
first of all , if the x - ray and optical afterglow shares the same source as the high - energy photons detected by lat , their spectral indices @xmath215 and @xmath216 should be consistent with the gev photons , @xmath217 and @xmath218 .
this is preliminarily valid only in the case of @xmath219 $ ] and @xmath220,when @xmath221 and @xmath222 .
consequently , the peak of the lat light curve lying at @xmath223s can be considered as the deceleration time of this burst .
therefore , equations and lead to the relationship among the isotropic kinetic energy @xmath224 , the bulk lorentz factor @xmath225 , and the wind parameter @xmath203 at the very moment : @xmath226 .
equation provides , at @xmath227 , @xmath228 .
thus the reverse shock is nearly newtonian .
subsequently , when the blast wave is embedded in the wind bubble , we assume that the typical value @xmath229 , while @xmath230 , and the characteristic frequencies evolve as @xmath231^{-2 } , \\ \nu_{c}(t_{ii})=1.03\times 10 ^ 9\,{\rm hz } e_{\rm k,55}^{1/2 } a_{35.5}^{-2 } \epsilon_{b,0}^{-3/2}(1 + 2\varkappa)^{-5/2}{t_{ii}}^{1/2}[1+y(\gamma_c , t_{ii})]^{-2},\\ \nu_{m}(t_i)=8.66\times
10^{23}\,{\rm hz } e_{\rm k,55}^{1/2 } \epsilon_{e,0}^{2 } \epsilon_{b,0}^{1/2}(1 + 2\varkappa)^{3/2}{t_i}^{-8/5}{f_p}^{2 } , \\
\nu_{m}(t_{ii})=7.39\times 10^{23}\,{\rm hz } e_{\rm k,55}^{1/2 } \epsilon_{e,0}^{2 } \epsilon_{b,0}^{1/2}(1 + 2\varkappa)^{3/2}{t_{ii}}^{-3/2}{f_p}^{2}.\end{aligned}\ ] ]
setting @xmath232 and @xmath233 as a trail , we only consider the synchrotron emissivity , @xmath234^{-1}.\ ] ] we found that equation is approximately the same under both fast cooling and slow cooling condition , so do with the following equation and .
@xmath235 from equation , we have @xmath236 , so @xmath237s in the case that @xmath238 .
therefore , @xmath239 and @xmath240 , which yields @xmath241 from @xcite , we have , in the wind , if @xmath242 , @xmath243 .
this have been shown in fig .
[ fig : parameter].c .
hence , as long as @xmath244 ( e.g.@xmath245 at 13s or @xmath246 at @xmath247 s ) , @xmath248<0.18a_{35.5}^{3/5}{t_i}^{3/5}(1 + 2\varkappa)^{-3/5}<1,\ ] ] according to @xcite , in the slow cooling , equation leads to @xmath249 , then ssc cooling has no effect on the electron distribution . from the analytical solution of @xcite , @xmath2501 can be estimated . with the assumption that the gev luminosity is dominated by synchrotron emission,@xmath251 provide a stronger constraint , @xmath252.in such parameter space , we have @xmath2531 , so the absorption does not contribute to the early rising .
all the above relations show that @xmath254 , @xmath255 and @xmath203 can be expressed with @xmath256 and @xmath188 , the physical mechanism for the rising part of the lightcurve can be considered as contribution of the structured ejecta . according to observational statistics , the best fitting yields @xmath257 , @xmath258 , @xmath259 , @xmath260 , and @xmath261 . as the result shown in fig .
fig:080916c : before the deceleration time , the bulk lorentz factor of the swept blast wave increases with @xmath262 ( which is self - consistent with our assumption for @xmath188 and thin shell case ) , @xmath178 occurs at @xmath263s then electrons are slow cooling , and deceleration begins from @xmath264 s , the density jump is expected around @xmath265 s ( i.e. , @xmath266 cm ) . after breaking out to the interstellar medium , the surrounding density is around @xmath267 . along with the lat and xrt data mentioned above ,
our optical photometry here is taken from @xcite , as well as the 8.5ghz data reported by the vla @xcite . with the assumption that the late x - ray , optical and radio afterglow and the early lat photons originate from the same region , their spectral indices @xmath268 and @xmath269 should be consistent with the gev photons , @xmath270 . the same as the above analysis , @xmath271 $ ] and @xmath220 , respectively , which result in the plausible light curve and spectrum similar to the above case .
therefore , roughly speaking , @xmath272 and @xmath273 are the available region .
although the peculiar `` soft - hard '' spectral evolution seems tricky in the prompt emission phase , its duration is so short ( @xmath274 s ) and it ends before the deceleration time @xcite , which , is understandable in the frame of structured ejecta acceleration ( we will discuss it in sec.[sect : con ] ) , so the external shock model is still applicable . in addition , although a 33 gev photon was detected at 82 s after the trigger , we consider the 11.2 gev photon at @xmath275s as the highest energy photon of estimation for the lower lorentz factor because the arrival time of the 33gev photon is far beyond the prompt emission phase and may not merely be attributed from the external shock but some other origins .
the turning point of the lat light curve occurs at @xmath2767 s. this indicates that a relationship among @xmath224 , @xmath225 , and @xmath203 becomes @xmath277 .
hence , equation yields that at @xmath227 and @xmath278 , so the thin shell case is applicable in this fitting .
similar to analytical solutions of grb080916c we mentioned above , characteristic frequencies of the photons observed by lat and later multiband ( x - ray , optical , radio ) afterglow can be estimated .
additionally , when we come to the radio band , ssa may be crucial . here
again , we assume that the typical value @xmath279 , while @xmath230 , with model parameters ( @xmath280 and @xmath203 ) , the characteristic frequencies evolve as @xmath281^{-2 } , \\ \nu_{c}(t_{ii})=2.71\times 10^{9}\,{\rm hz } e_{\rm k,55}^{1/2 } a_{35.5}^{-2 } \epsilon_{b,0}^{-3/2}(1 + 2\varkappa)^{-5/2}{t_{ii}}^{1/2}[1+y(\gamma_c , t_{ii})]^{-2},\\ \nu_{m}(t_i)=6.52\times
10^{23}\,{\rm hz } e_{\rm k,55}^{1/2 } \epsilon_{e,0}^{2 } \epsilon_{b,0}^{1/2}(1 + 2\varkappa)^{3/2}{t_i}^{-8/5}{f_p}^{2 } , \\
\nu_{m}(t_{ii})=5.37\times 10^{23}\,{\rm hz } e_{\rm k,55}^{1/2 } \epsilon_{e,0}^{2 } \epsilon_{b,0}^{1/2}(1 + 2\varkappa)^{3/2}{t_{ii}}^{-3/2}{f_p}^{2}.\end{aligned}\ ] ] setting @xmath282 as a trail , the synchrotron emissivity is given by @xmath283^{-1}.\ ] ] @xmath284 from equation , we have @xmath285 , so @xmath286s in the case that @xmath287 .
therefore , @xmath288 and @xmath289 , which yields , @xmath290 meanwhile , for the radio band at @xmath291 s , because of the light curve index @xmath292 , ssa can be neglected , we have @xmath293 therefore , it should be @xmath294 , and if @xmath295 ( e.g.@xmath296 at 10s or @xmath297 at @xmath247 s ) , @xmath298<0.10a_{35.5}^{3/5}(1 + 2\varkappa)^{-3/5}{t_i}^{3/5}<1.\ ] ] under such a parameter space , equation leads to @xmath299 , ssc makes a little effect because of the klein - nishina suppression , so the estimation that @xmath2501 is self - consistent . with the assumption that the gev luminosity is dominated by synchrotron emission , @xmath300
provide a stronger constraint , @xmath301 and @xmath302 . using the above relations
, we can express @xmath254 , @xmath255 and @xmath203 by @xmath256 and @xmath188 , and fit the data according to observational statistic .
the best fitting requires @xmath303 , @xmath304 , @xmath305 , @xmath306 , and @xmath307 .
fig : parameter shows the circumburst density , radius and characteristic frequency , and compton @xmath308 and @xmath309 factor evolution under such a parameter space . fig .
fig:090902b shows the spectrum and light curve fitting . before the deceleration time
( 7s ) , the bulk lorentz factor of the blast wave increases with a slope of @xmath310 ( which is self - consistent with @xmath188 and thin shell case ) .
the shock - accelerated electrons are cooling fast early before @xmath311 s , the density jump is estimated around @xmath312 s ( i.e. , at radius of @xmath313 cm ) . after breaking out to the interstellar medium , the surrounding density is as low as @xmath314 .
these parameters are consistent with the radio light curve except for the first data and is reliable when the jet break takes place nearly at the same time .
we find that the jet angle is 0.04 rad and collimation - corrected energy is @xmath315 .
v band and r band data of this burst are taken from @xcite and @xcite , respectively .
our analysis procedure is the same as mentioned above . due to the hypothesis of the same source , the temporal index of the lat photons @xmath316 , should be consistent with the spectral indices @xmath317 and @xmath318 .
this leads to a fairly broad plausible range for @xmath319 and @xmath320 ( e.g.under the condition of @xmath321 $ ] , @xmath322 requires for @xmath323 and @xmath324 , while @xmath325 and @xmath326 as long as @xmath327 ) .
however , since the light curve after the density jump is a little bit steeper than the early part , @xmath177 ( the latter the smaller ) should not deviate too much from the typical value ( i.e. , @xmath328 , otherwise , the decrease of loss efficiency will cause the decaying slope much shallower ) , whereas the decay of lat is steep , which requires a large @xmath329 ( i.e. , @xmath330 ) .
besides , in the afterglow phase , we can reasonably believe that the first flare at @xmath331 ks comes from an interaction of the blast wave and density jump .
the second `` flare '' may come from late energy injection or another density jump , and in this paper , we smooth it as the simple power law . because the deceleration time is estimated as @xmath332 s , @xmath225 can be written by @xmath333 . hence , at @xmath334 , @xmath335 , so the reverse shock is negligible in this fitting . with the typical value @xmath229 , while @xmath336 , the characteristic frequencies evolve as @xmath337^{-2 } , \\
\nu_{c}(t_{ii})=2.34\times 10^{9}\,{\rm hz } e_{\rm k,55}^{1/2 } a_{35.5}^{-2 } \epsilon_{b,0}^{-3/2}(1 + 2\varkappa)^{-5/2}{t_{ii}}^{1/2}[1+y(\gamma_c , t_{ii})]^{-2},\\ \nu_{m}(t_i)=7.43\times
10^{23}\,{\rm hz } e_{\rm k,55}^{1/2 } \epsilon_{e,0}^{2 } \epsilon_{b,0}^{1/2}(1 + 2\varkappa)^{3/2}{t_i}^{-8/5}{f_p}^{2 } , \\
\nu_{m}(t_{ii})=5.63\times 10^{23}\,{\rm hz } e_{\rm k,55}^{1/2 } \epsilon_{e,0}^{2 } \epsilon_{b,0}^{1/2}(1 + 2\varkappa)^{3/2}{t_{ii}}^{-3/2}{f_p}^{2}.\end{aligned}\ ] ] setting @xmath338 as a trail , the synchrotron emissivity is calculated by @xmath339^{-1},\ ] ] @xmath340 from equation , we have @xmath341 , so @xmath342s in the case that @xmath343 .
therefore , @xmath344 and @xmath345 , which yields , @xmath346 hence , as long as @xmath347 ( e.g. , @xmath348 at 36s or @xmath349 at @xmath247 s ) , @xmath350 < 0.015a_{35.5}^{1/2}(1 + 2\varkappa)^{-1/2}{t_i}^{3/5}<1.\ ] ] in the slow cooling case , equation leads to @xmath249 , ssc makes a little effect due to the klein - nishina suppression , because the estimation of @xmath2501 is self - consistent .
meanwhile , during @xmath351 s , in the case of fast cooling ( @xmath352 ) that @xmath353 and @xmath345 , @xmath354 , @xmath355 according to @xcite , this is within the strong kn regime ( case ii ) . analytical discussion with the above confines yield to the possibilities of case iib and iic , both leads to @xmath2501 .
hence , the contribution from ssc could be neglected , too . in
both fast and slow cooling , @xmath356 provide a stronger constraint , @xmath357 and @xmath302 . using the above relations
, we can express @xmath254 , @xmath255 and @xmath203 by @xmath256 and @xmath188 , and fit the data according to observational statistic .
when @xmath358 , @xmath359,@xmath360 , @xmath361 , and @xmath362 , fitting seems plausible ( fig .
fig:090926a ) . before the deceleration time
, the lorentz factor of the blast wave increases with a slope of @xmath363 , @xmath364 s and from @xmath365 significant deceleration appears .
the blast wave enters the ism around @xmath366 s ( i.e. , at radius of @xmath367 cm ) .
the medium density surrounding the bubble is @xmath368
. tab .
[ tab : fitting ] shows all the best fit parameters for the above three grbs .
common results can be found : \(1 ) in all the three burst fittings , the thin shell case is applicable , the very beginning of gev flux densities are forward - shock dominated .
the rising slope comes from the contribution of the power - law increasing bulk lorentz factor of structured ejecta .
these are self - consistent with the thin shell case , indicating that the central engine is very likely to be the core collapse of a massive star ( we will discuss it in sec.[sect : con ] ) .
our model implies that the central object would keep on accreting matter even after the lat trigger .
\(2 ) the remaining isotropic kinetic energy of the fireball after the prompt emission are all @xmath369 ergs , indicating @xmath370 of the total energy has been emitted in the prompt emission .
this is because @xmath178 occurs before the deceleration time for all the three grbs and the following dynamics could be considered as being quasi - adiabatic . as a matter of fact ,
@xmath102 in our sample fittings is @xmath371 , which meets the requirement of radiation efficiency related to the analytic procedure from @xcite .
practically , slow cooling of electrons starts before the deceleration time , and a jet break appears much later , so wu s spherical - like solution is tenable in this piece .
a low value of @xmath111 ( @xmath372 ) indicates a weak magnetic field , especially for grb 080916c and grb 090902b .
these grbs have been studied by @xcite , whose inferred magnetic field ( along with several other lat - detected events ) is consistent with shock compression of a modest circumstellar field ( @xmath373 g ) .
that is to say , no dynamo process is necessary to generate the magnetic field needed for the observed synchrotron afterglow emission . in our best fittings ,
we find @xmath374 g for grb080916c and @xmath375 g for grb090902b even when @xmath376s , being broadly consistent with the results of @xcite and @xcite .
meanwhile , when the blast wave enters the constant - density circumburst medium , the preshock magnetic field can be calculated with our fitting parameters as @xmath377 @xmath378 g for 080916c and @xmath379 g for 090902b , that suggests no magnetic field amplification .
\(3 ) although the ratio of @xmath102 and @xmath111 is large , synchrotron self - compton is not significantly contributed to the observed flux density in the gev band because of the klein - nishina suppression effect . in our sample fittings ,
only synchrotron radiation from the external forward shock is enough to contribute to the high energy emission and the synchrotron self - compton component can be neglected because @xmath380 in this piece . in particular , for grb 090902b , from our best - fitting parameters , the maximum electron lorentz factor at 82s indicates the highest photon energy can be as high as @xmath381 gev , so it is plausible that 33 gev photon may result from synchrotron radiation .
\(4 ) the wind parameter for three longest bursts is @xmath382 , which is reliably low .
according to @xcite , the progenitor star of grb 080916c , grb 090902b , and grb 090926a could have been in a cloud . due to a high pressure of the cloud
, a speedy wind will be slowed down by a pair of shocks ( viz , a reverse shock that propagates into the wind gas and a forward shock that propagates into the cloud ) .
this interaction produces a stellar wind bubble and forms a density - jump at a radius of @xmath383 cm . the homogeneous density outside this jump
is estimated to be in the range of @xmath384 , which is much less than the typical density of an interstellar medium , but could not be excluded from the possibility of existence in areas of active star formation as the interiors of a prexisting superbubble ( @xcite ) .
if confirmed by the superbubble observation , this can provide a piece of evidence for a connection between long - duration grbs and broad - lined sne ib / c . and would be natural to expect that massive progenitors of grbs explode where we observe star formation . meanwhile , in other fitting works ( @xcite )
, they have also confronted with the problem of low density , and suggest the lower metallicity progenitors with minimal pre - explosion mass loss or selection effects @xcite .
hence , although the lat emission at least during @xmath77 , from the spectral perspective @xcite , is likely to connect to the gbm emission of internal origin , the fitting facts tentatively suggest a possibility that the lat observation records an internal - external shock transition and at least in the lat decaying phase the external forward shock emission is dominated .
it is widely believed that long duration gamma - ray bursts ( grbs ) , like hydrogen - deficient type ib / c supernovae ( sne ib / c ) , result from the core collapse of a massive star .
the main characteristic of setting grbs apart from other sne is that a substantial fraction of the explosive energy is coupled to relativistic ejecta .
a compact central engine is responsible for accelerating and collimating a jet - like outflow and driving a sn explosion ( @xcite ) .
the precise nature of the central engine which powers grb - sne , however , remains an open question . in this paper
, we have undertaken extensive broadband ( @xmath11100 mev , radio , optical , and x - ray ) observations of three long - duration grbs ( grb 080916c , grb 090902b , grb 090926a ) detected by the lat instrument on the fermi satellite .
the bulk lorentz factors imply that the fireball is ultra - relativistic , which indicates that hyper - energetic bursts carry as high as @xmath385 ergs in the blast wave .
the temporal indices @xmath386 and spectral indices @xmath387 in tab .
[ tab : data ] , which are consistent with the observed simple power law slopes , show that our detailed density transition consideration can reconcile the plateau and shallower decay in the later afterglow . tab .
[ tab : fitting ] displays the estimated physical parameter space of the best fit , which is all reasonably certificated by the theory mentioned in sec.[sect : fit ] . specifically , the wind parameter @xmath388 shed us light on the central engine of stellar collapsars and their association with sne ib / c ( @xcite ) .
furthermore , the tricky yet popular early rise of the lat ( @xmath11100mev ) light curve may indicate whether high energy photons share the same source as the low energy afterglows . aside from two main kinds of explanation we mentioned above , leptonic or baryonic models ,
an alternative possibility comes from the high - latitude prompt emission @xcite and thus the flux density evolves as @xmath389 .
however , according to @xmath390 in tab . [
tab : data ] , the calculated temporal index is too steep to explain the early high energy emission decay . or else if the gev component results from the same conventional unstructured ejecta as the late x - ray and optical afterglows , before the deceleration time ( @xmath391 , that is , we only consider the thin shell case for long bursts ) , in the early wind bubble @xmath392 , the bulk lorentz factor is nearly constant , and thus the electronic cooling and minimum frequencies evolve as @xmath393 and @xmath394 , the peak flux density @xmath395 .
even when the ssc makes a significant effect , the ic electronic cooling and minimum frequencies and ic peak flux density evolve as @xmath396 and @xmath397 and @xmath398 .
since @xmath399 around the peak of lat light curve , @xmath400 ( for @xmath401 , this slope is not enough ) , as well as @xmath402 or @xmath403 for slow cooling and @xmath404 or @xmath405 for fast cooling .
no matter whether take the kn cutoff into account , this might be inconsistent with the observed rise of the lat light curve .
our explanation for the initial rising part is simply based on the assumption that @xmath406 .
therefore , @xmath407 , @xmath408 , and @xmath409 when @xmath410 .
co - contribution of the optical depth and structured ejecta bring about the rising .
this assumption can be understandable within the framework of of the core collapse of a massive star ( i.e. , sne ib / c ) .
if the central engine ejects matter continuously , a clean tunnel would be left after the early ejecta sweeps up the stellar envelope , and the succeeding ejecta , whose energy is attained via continuous accretion even after the lat trigger and whose lorentz factor is larger , catches up with the early ejecta and injects more energy into the blast wave , giving rise to a rise of the lat light curve . on the other hand ,
the thermal component can also be explained under such assumption that the previous jet lost kinetic energy due to impedance from shock frontier , then turned it into thermal radiation .
this hypothesis have been provided by @xcite that , in the case of red - giant or blue super giants , the powerful jet lose its energy input at its base in the course of overtaking the sub - relativistic weak supernova shock before reaching the surface , and eject little highly relativistic matter ; while in a helium star , the jet breaking out with continuously accretion power receiving at the base will lead its motion become highly relativistic .
later , some other possibilities , such as @xcite s baryon pollution scenario and @xcite s initial envelope choked model , also result in the earlier power - law rising distribution of bulk lorentz factor . in our frame ,
sub - mev photons provide the background annihilation with the later gev ones .
on one hand , photon with the largest energy from the external shock fasten the constraint of lower lorentz factor of the ejecta [ in @xcite , the lower limit of the bulk lorentz factor should be @xmath411 ; on the other hand , even the ejecta is slower than the lower limit , it is possible that the earlier rising is paretically contributed from opacity effect . about the observed time delay of the pulses and variabilities in both lat and gbm bands , some other modulation effects , except for optical depth can account for the concurrent variabilities in both bands , such as inhomogeneous surroundings , energy injection , etc .. these factors indicate that the variability connection between lat and gbm @xcite can not give a sufficient clue to the internal / external origin of the gev photons . in particular , as for grb090902b ,
one of the tricky properties is its soft - hard - soft quasi - thermal spectrum evolution , since the soft - hard evolution appears within @xmath412 , it can be better understood by our model that the initial outflow was dissipated as thermal component by the denser shock front / photosphere , when it breaks out of the envelope , the consequent highly relativistic ejecta in a clean tunnel can be dissipated strong enough to produce energetic non - thermal emission , even the photosphere can be outshined to some extent , therefore , the spectrum evolves from soft to hard .
99 abdo , a. a. , et al .
2009 , science , 323 , 1688 abdo , a. a. , et al .
2009 , , 706 , l138 asano , k. , guiriec , s. , & mszros , p. 2009
, , 705 , l191 atwood , w. b. , et al .
2009 , , 697 , 1071 band , d. , et al .
1993 , , 413 , 281 bissaldi , e. , & connaughton , v. 2009 , grb coordinates network , 9866 , 1 cenko , s. b. , et al .
2011 , , 732 , 29 chakrabarti , s. k. , et al .
2009 , grb coordinates network , 10009 , 1 chandra , p. , & frail , d.a .
2009 , grb coordinates network , 9889 , 1 chevalier , r. a. , & li , z .- y . 2000 , , 536 , 195 dai , z. g. , & lu , t. 1998a , , 333 , l87 dai , z. g. , & lu , t. 1998b , , 298 , 87 dai , z. g. , & lu , t. 2002 , , 565 , l87 dai , z. g. , & wu , x. f. 2003 , , 591 , l21 evans , p. a. , et al .
2007 , , 469 , 379 evans , p. a. , et al .
2009 , , 397 , 1177 evans , p. a. , et al .
2009 , grb coordinates network , 9871 , 1 fan , y .- z .
2009 , , 397,1539 fenimore , e. e. , madras , c. d. , & nayakshin , s. 1996 , , 473 , 998 gao , w .- h . , mao , j. , xu , d. , & fan , y .- z . 2009 , , 706 , l33 gehrels , n. , et al .
2004 , , 611 , 1005 gehrels , n. , ramirez - ruiz , e. , & fox , d. b. 2009 , , 47 , 567 golenetskii , s. , et al .
2009 , grb coordinates network , 9049 , 1 granot , j. , for the fermi lat collaboration , & the gbm collaboration 2010 , arxiv:1003.2452 greiner , j. , et al .
2009 , , 498 , 89 gupta , n. , & zhang , b. 2007 , , 380,78 jin , z. p. , xu , d. , covino , s . ,
davanzo , p. , antonelli , a. , fan , y. z. , & wei , d. m. 2009 , , 400 , 1829 kennea , j. a. , & stratta , g. 2009 , grb coordinates network , 9868 , 1 kumar , p. , & barniol duran , r. 2009 , , 400 , l75 kumar , p. , & barniol duran , r. 2010 , , 409 , 226 kumar , p. , & panaitescu , a. 2000 , , 541 , l51 lazzati , d. , rossi , e. , covino , s. , ghisellini , g. , & malesani , d. 2002 , , 396 , l5 li , z. 2010 , eprint arxiv , 1004 , 791 , 6 pages lithwick , y. , & sari , r. 2001 , , 555 , 540 liu , r.y . , & wang , x.y .
2011 , , 730 , 1 macfadyen , a. i. , & woosley , s. e. 1999 , , 524 , 262 macfadyen , a. i. , woosley , s. e. , & heger , a. 2001 , , 550 , 410 maxham , b.b , zhang , & b.zhang , 2010 , ( arxiv:1101.0144 ) mszros , p. , & rees , m. j. 1997 , , 476 , 232 meegan , c. , et al .
2009 , , 702 , 791 modjaz , m. , et al .
2008 , , 135 , 1136 nakar , e. , ando , s. & sari , r. 2009 , , 703,675 noda , k. , et al . 2009 , grb coordinates network , 9951 , 1 panaitescu , a. , & kumar , p. 2001
, , 554,667 panaitescu , a. , & kumar , p. 2002
, , 571 , 779 pandey , s. b. , et al .
2010 , , 714 , 799 perri , m. , preger , b. , & stratta , g. 2008 , grb coordinates network , 8261 , 1 piran , t. , & nakar , e. 2010 , , 718 , l63 , a. , et al .
2010 , , accepted ( arxiv:1004.3261 ) razzaque , s. , dermer , c. d. , & finke , j. d. 2009 , arxiv:0908.0513 rees , m. j. , & meszaros , p. 1998
, , 496 , l1 ryde , f. , et al .
2010 , , 709 , l172 sari , r. 1997 , , 489 , l37 sari , r. , piran , t. , & narayan , r. 1998 , , 497 , l17 sari , r. , & esin , a. a. 2001 , , 548 , 787 scalo , j. , & wheeler , j. c. , 2001 , , 562 , 664 soderberg , a. m. , et al .
2010 , , 463 , 513 spergel , d. n. , et al .
2007 , , 170 , 377 stratta , g. , delia , v. , & perri , m. 2009 , grb coordinates network , 9876 , 1 swenson , c. a. , et al .
& siegel , m. h. 2009 , grb coordinates network , 9877 , 1 swenson , c. a. , et al .
2010 , , 718 , l14 tam , p. h. , pun , c. s. j. , huang , y. f. , & cheng , k. s. 2005 , , 10 , 535 terada , y.,et al .
2009 , grb coordinates network , 9897 , 1 toma , k. , wu , x .- f .
, & mszros , p. 2009
, , 707 , 1404 uehara , t.,takahashi , h . , & mcenery , j. 2009 , grb coordinates network , 9934 , 1 van der horst , a. j. , kamble , a. p. , wijers , r. a. m. j. , & kouveliotou , c. 2009 , grb coordinates network , 9883 , 1 vetere , l. 2009 , grb coordinates network , 9961 , 1 wang , x. y. , li , z . ,
dai , z. g. , & meszaros , p. 2009
, , 698 , 198 wang , x. y. , he , h n. , li , z . , & dai , z. g. 2010 , , 712 , 1232 woosley , s. e. , & bloom , j. s. 2006 , , 44 , 507 woosley , s. e. , & heger , a. 2006 , , 637 , 914 5 wu , x. f. , dai , z. g.,huang , y. f. , & lu , t. 2005 , , 619 , 968 zhang , w. , woosley , s. e. , & heger , a. 2008 , , 679 , 639 zhang , b. , & peer , a. 2009 , , 700,l65 zhang , b .- b . , et al .
2010 , , 730 , 141 zou , y .- c . , fan , y .- z . , & piran , t. 2009 , , 396 , 1163 zou , y .- c . , fan , y .- z . , & piran , t. 2011 , , 726 , l2 .according to @xmath413 , @xmath414 , @xmath415 , @xmath416
are the estimated temporal indices of the lat band photons , x - ray + optical photons in the wind bubble , and x - ray + optical photons in the homogeneous medium from our best - fit parameters .
@xmath417 , @xmath418 , @xmath419 are the observed temporal indices of gev photons , simple power - law x - ray photons and optical photons . @xmath420 and @xmath421 are the estimated spectral indices of the lat band photons and x - ray + optical photons in either medium from our best - fit parameters .
@xmath390 , @xmath422 , and @xmath423 are the observed spectral indices of gev photons , simple power - law x - ray photons and optical photons .
high energy data are taken from @xcite , x - ray data are taken from _ swift _
group ( @xcite ) .
in addition , the data labeled with superscript @xmath424 are taken from @xcite , the data with @xmath425 are from @xcite , the data with @xmath426 are from @xcite , @xmath427 represents no data available . for 090926a ,
the data labeled with @xmath428 are from @xcite , the temporal slope of the x - ray and optical band in @xcite is : @xmath429 before the first flare , @xmath430 after the second flare , and @xmath431 between , which is consistent with our estimated value . [ cols="^,^,^,^,^,^,^,^,^,^,^,^",options="header " , ] | we present broadband ( radio , optical , x - ray and gev ) fits to the afterglow light curves and spectra of three long - duration gamma - ray bursts ( grbs 080916c , 090902b , and 090926a ) detected by the gamma - ray burst monitor ( gbm ) and large area telescope ( lat ) instruments on the fermi satellite . using the observed broadband data , we study the origin of the high energy emission , and suggest that the early - time gev emission and the late - time radio , optical , and x - ray afterglows can be understood as being due to synchrotron emission from an external forward shock caused by structured ejecta propagating in a wind bubble jumping to a homogeneous density medium .
if the ceasing time for majority of the energy injection is assumed to be close to the deceleration time of the forward shock , the structured ejecta with continuous energy injection to the forward shock can well explain the early rising feature of the gev mission from these burst , and the density - jump medium can account for some certain plateaus or flares in the late afterglows . from our fits , we find that , on one hand , the external shock origin of the gev photons will make the optical depth have not significant contribution to the early lat rising part , which will loosen strong constraint of lower limits of lorentz factor . on the other hand ,
these fermi - lat events preferentially occur in a low - density circumburst environment , in which case the klein - nishina cutoff will significantly suppress the self - synchrotron compton ( ssc ) radiation
. such an environment might result from superbubbles or low - metallicity progenitor stars ( which have a low mass - loss rate at late times of stellar evolution ) of type ib / c supernovae . |
quantum walks ( qws ) were introduced as a quantization of random walks ( rws ) @xcite .
qws are determined by a graph , its induced hilbert space @xmath4 and a unitary time evolution operator on @xmath4 .
the amplitude of quantum walker is obtained by this unitary iteration to some initial state . due to the unitarity of the time iteration ,
the norm of the amplitude is preserved , which implies that the distribution can be defined at each time step .
however it is believed that there are no non - trivial representations of the present distribution of qws by that of past time like markov chain @xcite .
qws are applied to various study fields , for example , a problem of searching marked elements on graphs @xcite , @xcite , @xcite , @xcite , fundamental physics @xcite , @xcite , the limit theorems for its statistical behaviors @xcite , @xcite , spectral analysis @xcite , @xcite , and photon synthesis @xcite . in @xcite , szegedy walk was formulated as a natural quantization of reversible marcov chains , and szegedy showed that in most cases the quantized walk hits the marked set within the square root of the classical hitting time .
grover walk , which is a special case of szegedy walk is widely studied quantum walk .
it is related to the analysis of zeta function and isomorphic problem between two cospectrum strongly regular graphs @xcite , @xcite . in @xcite ,
grover s search algorithm was introduced to search marked elements in database . in this paper
we focus on the periodicity of the grover walk on graphs .
the periodicity means that the quantum state at some time returns to the initial state .
recently , the periodicity of qws has been studied . in @xcite ,
konno , takei and shimizu studied hadamard walk on cycle graph @xmath5 .
they found that only @xmath6 , and @xmath7 induce periodic hadamard walks , whose periodicity is @xmath8 , and @xmath9 , respectively . in @xcite , higuchi , konno , sato and segawa
studied some qws on several finite graphs .
they found the condition of these graphs to induce periodic qws .
their results are as follows : * ( complete graphs ) szegedy walks induced by isotropic random walk with laziness @xmath10 with @xmath11 \cap { { \mathbb{q}}}$ ] on @xmath12 are periodic if and only if @xmath13 , or @xmath14 , whose period is @xmath15 and @xmath16 , respectively . *
( complete bipartite graphs ) szegedy walks induced by isotropic random walk with laziness @xmath10 with @xmath11 \cap { { \mathbb{q}}}$ ] on @xmath17 with @xmath18 and @xmath19 are periodic if and only if @xmath20 or @xmath21 , whose period is @xmath22 and @xmath23 , respectively .
* ( strongly regular graphs ) grover walks on strongly regular graphs with parameters @xmath24 are periodic if and only if @xmath25 whose period is @xmath26 and @xmath27 , respectively . *
( cycle graphs ) szegedy walks induced by random walk with non - revertible probability @xmath28 with which walker moves clockwise on @xmath5 are periodic if and only if @xmath29 , @xmath30 for @xmath31 , whose period is @xmath32 and @xmath23 , respectively , or @xmath29 , @xmath30 for @xmath33 , whose period is @xmath34 and @xmath23 , respectively , or @xmath35 for @xmath36 , whose period is @xmath9 .
on graphs which induce a periodicity , the distribution iterates with some period , which implies that there are no limit distributions of qw .
so we can say that such graphs are special class of graphs from the viewpoint of qws .
our purpose is to characterize such special class of graphs . in the previous results , for fixed graphs
, they found the condition of graphs to induce periodic qws . on the contrary
, we fixed an integer @xmath0 and tried to characterize graphs to induce a @xmath0-periodic grover walk .
we give @xmath37 as @xmath4 , and the @xmath38 unitary operator called the grover transfer matrix as @xmath39 , where @xmath40 is a set of symmetric arcs of a finite graph @xmath41 .
a walker of grover walk on graphs transfers on arcs .
a motion of the quantum walker is interpreted as a dynamics of plane wave on the metric graphs @xcite .
first , we introduce the notations and qws on discrete graphs .
all graphs @xmath41 considered in this paper are finite and simple graph without loops and multiple edges .
let @xmath42 , @xmath43 be a set of vertices and edges of @xmath41 and define @xmath44 , respectively .
the matrix @xmath45 is an @xmath46 transition matrix of isotropic rws , that is , @xmath47 for @xmath48 , @xmath49 , @xmath50 is an arc from @xmath48 to @xmath51 , and @xmath52 .
we denote @xmath53 , @xmath54 .
in addition , @xmath55 is a set of symmetric arcs of @xmath41 .
we consider @xmath56 as a state space . for a square matrix @xmath57
, if we note @xmath58 it implies that the multiplicity of the eigenvalue @xmath59 of @xmath57 is @xmath60 for @xmath61 .
we denote @xmath62 , if @xmath63 is an eigenvalue of @xmath57 . throughout this paper , @xmath64
, @xmath65 denotes the path , complete graph with @xmath10 vertices , respectively .
the graph @xmath66 implies a cycle graph with @xmath10 vertices and we call it an even cycle if @xmath10 is even , otherwise an odd cycle . in this paper
, we say that the graph is unicycle if it has only one cycle and the graph removed the cycle becomes a non - empty forest .
so we do not call cycle graphs unicycle although they are unicycle generally .
in addition we call it an even unicycle if the length of cycle is even , otherwise an odd unicycle .
ccc we call the graphs like figure 1 , figure 2 unicycle but do not call the graphs like figure 3 so in this paper .
let @xmath67 be a complete bipartite graph with two partitions with @xmath68 and @xmath69 vertices .
the tree graph is defined as the graph without cycles .
the girth of @xmath41 is denoted by @xmath70 , which means the length of the minimal cycle in @xmath41 . furthermore , we assign vectors @xmath71 to every @xmath72 , where @xmath71 s are the standard basis of the hilbert space @xmath73 , that is , @xmath74 for every @xmath75 .
we construct the quantum state at time @xmath76 , @xmath77 as @xmath78 where @xmath79 , and @xmath80 .
then the finding probability of the walker on an arc @xmath81 at time @xmath76 is @xmath82 . giving a @xmath83 unitary matrix @xmath39
, we determine @xmath84 as @xmath85 hence , we can denote @xmath86 as @xmath87 using the initial state @xmath2 .
next , we introduce grover walk .
the evolution operator of the grover walk is the following @xmath83 unitary matrix @xmath88 : @xmath89 the quantum waves on the arc @xmath90 transmits to the arc @xmath81 with @xmath91 and @xmath92 with a rate of @xmath93 , and reflects to the arc @xmath94 with a rate of @xmath95 .
this @xmath39 is called the grover transfer matrix .
we shall give an example of the grover walk and consider the periodicity .
we will provide an example of graph which induces a periodic grover walk .
+ example : @xmath96 . ]
@xmath97 in fact , this @xmath39 satisfies @xmath98 , that is , @xmath99 for an arbitrary initial state @xmath2 .
we say @xmath100 is the graph to induce a @xmath22-periodic grover walk .
the other examples are given in figure 5 , 6 , 7 .
ccc these graphs induce @xmath101-periodic grover walks , respectively . for a positive integer @xmath0 , inducing a @xmath0-periodic grover walk implies that @xmath102 , while @xmath103 for every @xmath104 with @xmath105 for the grover transfer matrix @xmath39 .
these conditions can be regarded as the following spectral problem .
a graph @xmath41 induces a @xmath0-periodic grover walk if and only if @xmath106 for every @xmath107 , and there exists @xmath108 such that @xmath109 for every @xmath104 with @xmath105 .
so what we have to do is checking whether all the eigenvalues of @xmath39 satisfy the condition of the above proposition . in order to consider it , we need the following theorem .
@xmath110emms , hancock , severini and wilson @xmath111 , [ 4]$ ] @xmath112 + for the grover transfer matrix @xmath39 and the transition matrix @xmath113 on a graph @xmath41 , it holds @xmath114 for every @xmath115 .
so @xmath39 has @xmath116 eigenvalues of the following form @xmath117 where @xmath118 .
the remaining @xmath119 eigenvalues are @xmath120 , which have the same multiplicities .
therefore if these two kind of eigenvalues satisfy the condition of proposition 1.1 , @xmath41 induces a @xmath0-periodic grover walk .
we characterize such graphs for @xmath3 and obtain a necessary condition for an odd @xmath0 .
* theorem @xmath121 the graph @xmath122 is the only graph to induce a @xmath123-periodic grover walk . *
theorem @xmath124 if @xmath41 induces an odd - periodic grover walk , then @xmath41 is an odd cycle or an odd unicycle graph .
* theorem @xmath125 the graphs @xmath126 , @xmath127 are the only graphs to induce @xmath128 , @xmath27-periodic grover walks , respectively . *
theorem @xmath129 the graphs @xmath67 are the only graphs to induce a @xmath22-periodic grover walk for every @xmath130 .
this paper is organized as follows : in section 2 , we mention the @xmath123-periodic case and prove theorem 2.1 . in section 3 , we first give a necessary condition for graphs to induce an odd - periodic grover walk and prove theorem 3.2 with several lemmas .
in section 4 , we prove theorem @xmath131 by using a property of bipartite graphs . at the end of this paper
, we summary our results and make some discussions in section 5 .
here we explain the graph to induce a @xmath123-periodic grover walk , and prove theorem 2.1 with some lemmas .
the graph @xmath122 is the only graph to induce a @xmath123-periodic grover walk . for any @xmath118
, it holds that @xmath132 and @xmath133 .
@xmath134-@xmath135 if @xmath57 is a non - negative matrix , that is , all entries are non - negative , then the eigenvector of the maximal eigenvalue of @xmath57 is a non - negative vector and its multiplicity is @xmath136 .
proof of theorem 2.1 : obviously @xmath122 induces a @xmath123-periodic grover walk .
hence , we will prove that if @xmath41 induces a @xmath123-periodic grover walk , then @xmath41 is @xmath122 . by proposition 1.1 , for any @xmath108
, it should hold that @xmath137 . according to theorem 1.2
, @xmath39 has eigenvalues of the form @xmath138 , and the remaining @xmath119 eigenvalues are @xmath139 .
the latter values satisfy the condition of proposition 1.1 , then the former values should satisfy it , which implies @xmath140 , that is , @xmath141 . since @xmath113 is
a non - negative matrix and @xmath136 is the maximal eigenvalue of @xmath113 , then its multiplicity is @xmath136 by lemmas 2.2 , 2.3 .
moreover the multiplicity of @xmath142 is also @xmath136 because of @xmath143 .
so we can obtain @xmath144 considering the connectivity of @xmath41 and the summation of their multiplicities , we can obtain that @xmath122 is the only graph which leads ( 1 ) as a spectrum of its transition matrix .
indeed , @xmath145 obviously it induces a @xmath123-periodic grover walk .
then such a graph is only @xmath122 .
in this section we show that the graphs inducing an odd - periodic grover walk should satisfy some conditions .
then we prove theorem 3.1 and theorem 3.2 .
if @xmath41 induces an odd - periodic grover walk , then @xmath41 is an odd cycle or an odd unicycle graph .
the graphs @xmath126 , @xmath127 are the only graphs to induce @xmath147-periodic grover walks , respectively .
a graph @xmath41 is a bipartite graph if and only if it holds that @xmath148 for the eigenvalues of its transition matrix @xmath113 . furthermore using lemma 2.2 ,
lemma 2.3 , we can gain the following corollary : a graph @xmath41 is a bipartite graph if and only if @xmath149 .
proof of theorem 3.1 : let @xmath0 be an odd integer .
we assume that @xmath150 . for all of the eigenvalues of @xmath39
, @xmath151 should satisfy @xmath152 .
if @xmath150 , then at least one @xmath142 is an eigenvalue of @xmath39 by theorem 1.2 . since @xmath0 is odd , @xmath142 does not satisfy the condition .
thus it should hold that @xmath153 .
it follows that @xmath154 from the connectivity of @xmath41 .
then such graphs must be trees , which satisfy @xmath155 , or cycles , unicycles , which satisfy @xmath156 . from corollary 3.4 and theorem 1.2 , trees , even cycles , and even unicycles are improper graphs since they are bipartite .
hence , an odd cycle and an odd unicycles can induce a @xmath0-periodic grover walk for an odd @xmath0 .
@xmath146 first , we introduce an easy lemma and obtain a restriction of unicycle graphs which induce an odd - periodic grover walk .
we prove theorem 3.2 with them .
the graphs @xmath157 induce a @xmath0-periodic grover walk .
proof : let @xmath57 be an adjacent matrix of @xmath157 , and @xmath158 be an eigenvalue of @xmath57 . for every @xmath104 with @xmath159 , @xmath160 since @xmath157 is @xmath123-regular , it holds that @xmath161 then @xmath162 for @xmath159 .
hence , the eigenvalues of @xmath39 are @xmath163 thus @xmath157 induces a @xmath0-periodic grover walk .
@xmath146 let @xmath41 be an odd unicycle graph .
if @xmath41 induce a @xmath0-periodic grover walk , then it should hold that @xmath164 .
proof : let @xmath0 be an odd integer and @xmath165 .
we assume that an odd unicycle @xmath41 induce a @xmath0-periodic grover walk .
since @xmath41 is a unicycle , @xmath41 contains the graph on figure 8 as its subgraph .
let the vertices of the cycle @xmath166 be @xmath167 .
we define arcs @xmath168 such as @xmath169 let @xmath81 be the arc @xmath170 . for the grover transfer matrix @xmath39 , @xmath171 can be written by @xmath172 where @xmath173 run over arcs which make the walk @xmath174 with the length @xmath0 in @xmath41 . from its periodicity
, it should hold that @xmath175 .
we consider two cases ( i ) @xmath176 , ( ii ) @xmath177 and show @xmath178 in the both of cases .
\(i ) @xmath176 . in order to obtain @xmath179
, we consider the walks with length @xmath0 from @xmath81 to @xmath81 .
since the distance between @xmath81 and @xmath81 is even and @xmath41 is a unicycle , we have to traverse @xmath166 to go from @xmath81 to @xmath81 with odd steps .
this walk has the length at least @xmath180 .
so there are no walks from @xmath81 to @xmath81 with length @xmath0 since @xmath176 .
therefore we can conclude that @xmath181 , and it gives us a contradiction .
\(ii ) @xmath182 . by the observation of ( i ) , the walks should run through @xmath166 at once .
since @xmath182 , there are only two walks such as @xmath183 hence , @xmath184 it follows @xmath185 and @xmath186 from @xmath175 .
therefore @xmath187 however it holds that @xmath188 due to the choice of the subgraph .
thus it follows that @xmath189 .
@xmath146 proof of theorem 3.2 : from the previous arguments , the graphs inducing @xmath128 , @xmath27-periodic grover walks are @xmath126 , @xmath127 or the other unicycle graphs , respectively .
proposition 3.6 leads the fact that the girth @xmath70 of such unicycle graphs should be @xmath190 for @xmath191 , and @xmath27 .
however @xmath192 .
thus no unicycle graphs induce @xmath128 , and @xmath27-periodic grover walks . therefore @xmath126 , @xmath127 are the only graphs to induce @xmath128 , @xmath27-periodic grover walks , respectively .
in this section we prove theorem 4.1 by using a property of bipartite graphs . the graphs @xmath67 are the only graphs to induce a @xmath22-periodic grover walk for every @xmath130 . similar to the previous section , we introduce some lemmas to prove theorem 4.1 .
a graph @xmath41 induces a @xmath22-periodic grover walk if and only if the spectrum of its transition matrix @xmath113 is of the form @xmath193 proof : first , we show its necessity . in other words , we show that if @xmath41 induces @xmath22-periodic grover walk , then spectrum of @xmath113 is of the form ( 3 ) .
similar to proof of theorem 2.1 , for any @xmath108 , it should hold that @xmath194 , and it follows that @xmath195 thus , @xmath196 .
therefore we can obtain @xmath193 immediately , their multiplicities are determined by lemma 2.2 and lemma 2.3 .
next we show its sufficiency .
if the spectrum of @xmath113 of @xmath41 is of form ( 3 ) , then the spectrum of @xmath39 is of the form @xmath197 from proposition 1.1 , @xmath41 induces a @xmath22-periodic grover walk .
@xmath146 + + in fact this spectrum induces complete bipartite graphs immediately .
a graph @xmath41 is a complete bipartite graph if and only if the spectrum of its transition matrix @xmath113 is of the form @xmath193 proof : first , we show its necessity .
if @xmath41 is a complete bipartite graph , then its transition matrix @xmath113 can be written by @xmath198 where @xmath199 , @xmath200 denote the @xmath201 matrix with all @xmath202 entries , and the @xmath201 matrix with all @xmath136 entries , respectively .
then it follows that @xmath203 , and @xmath204 .
it implies that the multiplicity of an eigenvalue @xmath202 of @xmath113 is @xmath205 since @xmath113 is a diagonalizable matrix .
furthermore @xmath136 is an eigenvalue of @xmath113 with multiplicity @xmath136 , and @xmath142 is so from corollary 3.4 .
therefore we can obtain ( 4 ) as the spectrum of the transition matrix of @xmath41 .
next we show its sufficiency .
if the spectrum of @xmath113 is of the form ( 4 ) , then @xmath142 is an eigenvalue of @xmath113 .
thus @xmath41 is a bipartite graph by corollary 3.4 .
then its transition matrix @xmath113 can be represented by @xmath206 with some @xmath207 matrix @xmath208 , and @xmath209 matrix @xmath210 .
these matrices are the transition matrices from a partition @xmath211 to another partition @xmath212 , and from @xmath212 to @xmath211 , respectively , where @xmath213 and @xmath214 , @xmath215 .
in addition each summations over the row of @xmath113 are @xmath136 .
we show that all of the entries of @xmath208 , and @xmath210 are not @xmath202 . from the assumption of @xmath204
, we can see that @xmath203 .
therefore @xmath208 and @xmath210 can be represented by @xmath216 @xmath217 where @xmath218 is a @xmath219 rational vector and @xmath220 is a @xmath221 rational vector , and @xmath222 are rational numbers for @xmath223 , @xmath224 .
we can obtain @xmath225 , @xmath226 for @xmath223 , @xmath224 since the summations over the row of @xmath113 are @xmath136 .
if there exists @xmath227 such that the @xmath227-th entry of @xmath218 is @xmath202 , then the @xmath227-th column vector of @xmath113 is the zero vector .
it contradicts the connectivity of @xmath41 .
thus the @xmath208 does not contain @xmath202 as its entries .
similarly the @xmath210 does not so
. therefore any vertices in @xmath211 are adjacent to the vertices in @xmath212 , and the vertices in @xmath212 are so .
we can conclude that @xmath41 is a complete bipartite graph .
@xmath146 proof of theorem 4.1 : from lemma 4.2 , @xmath41 induces a @xmath22-periodic grover walk if and only if the spectrum of its transition matrix is of the form ( 4 ) . having such a spectrum of @xmath113 means that @xmath41 is a complete bipartite graph by lemma 4.3 .
then we can show that the complete bipartite graphs @xmath228 are the only graphs to induce a @xmath22-periodic grover walk .
in this paper , we have given some characterization of the graphs to induce a @xmath0-periodic grover walk for @xmath3 .
we proved that @xmath229 are the only graphs to induce @xmath230-periodic grover walks , respectively .
one of what we want to do is to determine such graphs for an integer @xmath0 with @xmath231 .
the main method used in this paper to characterize such graphs is to analyze the spectrum of its transition matrix , that is , we found graphs whose spectrum of its transition matrix is occupied by the real part of the @xmath0-th root of @xmath136 and have an eigenvalue that is not the real part of @xmath104-th root of @xmath136 for every @xmath104 with @xmath232 . generally speaking ,
it is difficult to characterize graphs with any given spectrum . for the cases of @xmath231
, we might take another method to solve it . using this lemma and lemma 3.5
, we can conclude that @xmath235 and @xmath236 induce @xmath16-periodic grover walks .
in addition both of the graphs on figure 9 , 10 also induce @xmath16-periodic grover walks .
these graphs are made by identifying two endpoints of some @xmath235 s. they include @xmath235 and @xmath236 .
furthermore both of the graphs on figure 11 , 12 induce @xmath237-periodic grover walks .
@xmath241 induce @xmath242-periodic grover walks , respectively .
we can regard the subdivision as an operator which conserves the periodicity of the grover walk between graphs . to find such operators between graphs
is also our interest .
the grover walk on the graphs are determined by the grover transfer matrix .
we have to investigate the periodicity of another qws on the graphs determined by a unitary matrix except the grover transfer matrix .
m.j.cantero , f.a.gr@xmath243nbaum , l.moral , l.vel@xmath244zquez , matrix - valued szeg@xmath245 polynomials and quantum random walk , communications on pure and applied mathematics , * 63 * , pp464 - 507 , ( 2010 ) | recently , a research on quantum walks has been developed in various areas . in this paper
we focus on the periodicity of the grover walk which is one of quantum walk on the discrete graphs .
then we found some special graphs to induce a periodic grover walk : at some time @xmath0 , the quantum state @xmath1 returns its initial quantum state @xmath2 .
our purpose is to characterize graphs which induce a @xmath0-periodic grover walk for a fixed integer @xmath0 .
we do it for @xmath3 and obtain a necessary condition for odd @xmath0 . |
the essential optical and electro - optical components that define and stabilize the science cavity resonance frequency , drive the cqed system , detect individual atom transits , and perform homodyne detection of the cqed transmitted field , as explained in the text.,scaledwidth=85.0% ] the experiment consists of a standard cqed setup involving laser cooled @xmath0cs atoms and a high finesse fabry - perot optical resonator @xcite . through frequency and polarization selectivity ,
we attempt to drive only the @xmath41 atomic cycling transition at 852 nm in order to approximate the atom as a two - level system .
although the cavity used in the experiment was constructed to optimize signatures of spontaneous dressed - state polarization at high drive amplitudes @xcite , it proved well - suited to study the amplitude ` bistabiliy ' dynamics presented in the main article . inside a uhv ( @xmath42 torr ) chamber and placed on a multi - stage vibration - isolation stack , the fabry - perot optical resonator is formed by two high - reflectivity ( 8ppm transmission , 2ppm loss ) , 10 cm radius of curvature dielectric mirrors with roughly 27@xmath6 m of separation , yielding a 300,000-finesse optical resonator for the standing wave , tem@xmath8 , 18@xmath6m - waist transverse spatial mode with a field decay rate of @xmath43mhz .
we took particular care to mount the mirrors in a rotationally - symmetric manner to minimize stress - induced birefringence in the mirror coatings , allowing for full polarization - selectivity of the atomic transitions .
the cavity length is tuned and actively stabilized by two shear - mode piezoelectric plates underlying the two mirror mounts .
the precise cavity length and resonance frequency is continually stabilized by the pound - drever - hall ( pdh ) @xcite method using an additional laser probe detuned by the desired probe / cavity resonance frequency by two cavity free spectral ranges ( at an optical wavelength of roughly 826 nm , which interacts negligibly with cs ) .
a doppler - limited , magneto - optically trapped ensemble ( mot ) of @xmath44 atoms is formed roughly 1 cm above the cavity mode in the uhv chamber . after cooling ,
the ensemble trap is switched off , allowing the cold atoms to fall under gravity towards the cavity mode and by the time they reach the cavity mode their free - fall velocity tends to dominate any residual thermal motion . due to the strong coupling between the targeted atomic transition and the cavity mode ( with calculated maximum value @xmath45mhz at the cavity anti - nodes , using the dipole strength of the atomic transition and cavity mode volume ) ,
individual atom transits are detected by monitoring the ( @xmath11-dependent ) cavity transmission amplitude using a relatively weak and near - resonant probe @xcite , a free space balanced photodetector , and an actively phase - locked optical local oscillator ( lo ) .
although multiple atom transits per drop may be visible ( apparent whenever @xmath23 ) , the atomic ensemble is sufficiently diffuse such that no more than one strongly - coupled atom is simultaneously present in the cavity mode and we acquire data from only one transit per ensemble drop .
it is in principle possible that a weakly coupled `` haze '' of background atoms are also coupled to the mode , but there is now a considerable literature ( starting with @xcite ) that demonstrates that dropped - atom cavity qed systems quantitatively reproduce the predictions of single - atom theory .
once a strongly coupled atom has been detected , the probe power and frequency shift to the desired experimental levels and data acquisition is initiated .
[ fig : exp ] depicts many aspects of the resonance lock , optical drive , transmission measurement , and atom - triggering in the experiment ( the laser cooling optical system is standard and remains tacit ) . a diode laser is pdh frequency locked 253mhz to the blue of a large , mechanically stable ` transfer ' cavity mode , which is itself locked 253mhz to the red of the @xmath46 , 852 nm hyperfine transition ( so that the diode laser is on atomic resonance ) .
the purpose of this @xmath47 finesse , 10khz - linewidth transfer cavity is two - fold : provide a stable , cs - locked frequency reference for all the lasers and act as a cleaning cavity for the diode laser , producing a more narrow linewidth 852 nm laser source in transmission than is easily achievable in the laser lock . the 852 nm beam transmitted by the transfer cavity seeds a high power slave laser , which sources the light used in both the science cavity drive and optical local oscillator .
the ` drive ' arm of the slave output is frequency shifted to the red of the atomic transition again before entering an electro - optic modulator ( eom 1 ) , which adds the optical carrier sideband near atomic resonance that serves as the science cavity s dynamic drive ( the carrier and other sidebands are sufficiently detuned from the cqed system and have no measurable effect ) .
the science cavity is actively stabilized by a second , titanium - sapphire ( ti : s ) laser at roughly 826 nm and frequency locked relative to the 852 nm diode laser and the atomic transition via a frequency lock of the ti : s laser to the transfer cavity : an optical sideband exactly two free spectral ranges to the red of the desired science cavity resonance near the atomic resonance is defined by a second eom ( eom 2 ) and used to pdh resonance lock the science cavity in transmission .
the power of the locking laser was kept at @xmath30100nw so that the ac stark shift it induced in the atoms could be neglected in our analysis . as implied by the main article ,
the effects of the uncertainties in system parameters ( e.g. laser noise , cavity resonance frequency instabilities , ac stark shifts of the atom ) are dominated by the effects of the variations in the atom - cavity coupling rate . for detecting and triggering off of strongly coupled atoms during their @xmath29s transit through the mode , the drive is initially tuned several mhz from the desired frequency for the experiment and its amplitude is set below the level of atomic saturation , while the several mw optical lo is tuned to the experimental drive frequency .
both heterodyne quadratures of the transmitted atom - probe are detected via their interference with the optical lo .
the heterodyne phase quadrature is used to stabilize the relative phase of the free - space lo against slow drifts in the signal and lo path lengths via a 1khz bandwidth optical phase lock loop .
the amplitude quadrature is monitored by a schmitt trigger that fires when the measured field amplitude drops below a threshold indicating a strongly coupled atom in the cavity . this trigger shifts the drive amplitude , frequency and lo - relative phase to the desired experimental configuration and initiates data acquisition of the homodyne photocurrent at 200ms / s .
although collected at full bandwidth , the data presented in the main article has been additionally filtered for clarity using a 20mhz two - pole low pass , filtering out as much of the high frequency shotnoise as possible while still preserving the visibility of the fast , quantum fluctuations .
the driven jaynes - cummings hamiltonian @xcite is the standard quantum model of the internal dynamics of a two - level atom coupled to an optical resonator mode in a frame rotating with the external drive ( @xmath48 ) @xmath49 where @xmath50 is the cavity mode annihilation operator , @xmath51 is the atomic lowering operator and @xmath52 denotes the hermitian conjugate . from left to right , the rhs terms correspond to the atom - drive detuning , the mode - drive detuning , the atom - mode coupling , and the external drive , respectively .
the complete quantum model of this system comes from extending these hamiltonian dynamics to include processes associated with the dissipation of photons through the cavity mirrors at mean rate @xmath53 per intra - cavity photon and excited atomic state spontaneous emission at mean rate @xmath54 , as modeled by @xmath50 and @xmath51 operator - coupling to external quantum fields @xcite .
only one of these fields , the transmitted mode , is monitored by our homodyne detection setup . for many applications ,
the entire model may be effectively represented by a _ master equation _ that describes the unconditional evolution of any mode and/or atom operator @xmath55 ( in the heisenberg picture ) @xcite @xmath56 + 2\kappa\left(a^\dag oa - \frac12a^\dag ao-\frac12oa^\dag a\right ) + 2\gamma_\perp\left(\sigma^\dag o\sigma - \frac12\sigma^\dag\sigma o-\frac12o\sigma^\dag\sigma\right)\nonumber\\ & \equiv&\mathcal{l}o.\end{aligned}\ ] ] for example , the ( one dimensional ) null space of the analogous , schrdinger picture ` liouvillian , ' @xmath57 , corresponds to the steady state density matrix utilized in figs .
[ fig : mf](b ) and [ fig : ss ] in the main article .
this model may be trivially extended to describe the interaction of any number of two level atoms with the mode ( or even with multiple modes ) .
starting from this quantum model , the maxwell - bloch equations ( mbes ) may be derived by first assuming the approximation that atom - mode operator expectations factor @xcite , _
e.g. _ @xmath58 ( see also @xcite ) .
this approximation is equivalent to modeling the cavity mode as if it were a noiseless classical field coupling to a two - level atom or to an ensemble of atoms ( as the case may be ) .
the mbes can take the form of a set of five real , first - order , non - linear differential equations of motions for @xmath59 and other expectations .
steady state solutions for @xmath59 ( proportional to the mean field transmitted by a fabry - perot cavity ) and their dynamical stability may be found .
properly scaled , these mean - field solutions depend on the atom(s)-mode coupling only through a dimensionless parameter known as the ` cooperativity , ' @xmath60 , where @xmath61 is the number of coupled atoms and @xmath11 the rate of coupling to a single atom @xcite .
thus , as long as the cooperativities are equal in both cases , this model predicts equivalent steady states for a coupled atom and for a coupled macroscopic atomic ensemble . for highly non - linear , @xmath62 systems ,
however , the correlations between discrete excitations in the mode ( photons ) and the atom(s ) that are ignored in the mean - field model may play a significant role in the overall dynamics when @xmath63 , as demonstrated in the main article , with our experimental @xmath6467 , @xmath61=1 system .
measurements of the external field transmitted by the cqed system are modeled using a quantum stochastic differential equation ( qsde ) framework @xcite .
loosely speaking , this model represents observables of the transmitted field measured by our homodyne detection setup as a linear combination of cavity mode observables and ` quantum white noises . ' for example , expected photocurrent distributions and correlations may be calculated from the characteristic functional @xcite @xmath65 = \langle \mathcal{t}\exp\left\{\int_0^t\beta_sdb^\dag_{out , s}-\int_0^t\beta_s^\ast db_{out , s}\right\}\rangle\ ] ] where @xmath66 is some complex - valued scalar function of time , @xmath67 denotes complex conjugation , @xmath68 is the qsde annihilation process of the measured , transmitted field , @xmath69 is the time - ordering operator , and the expectation is taken over both the system and external field degrees of freedom . assuming a boxcar averaging photocurrent filter of width @xmath70 ( so that @xmath71 for @xmath72 and @xmath73 otherwise ) , the fourier transform of eq .
produces a representation of the field analogous to a wigner quasi - probability distribution @xmath74\ ] ] where @xmath75 is the reduced density matrix for the cqed system alone at time @xmath76 ( may be the steady state density matrix , for example ) , @xmath77 is a ` measurement superoperator , ' and @xmath78 is the efficiency with which photons decaying from the cavity are measured by the homodyne detector . as with a wigner function , @xmath79 , the probability that a homodyne measurement of quadrature @xmath80 at time @xmath76 will take the ( real ) value @xmath81 , is obtained by integrating @xmath82 over the complementary quadrature , @xmath83 in practice , we approximate @xmath82 by assuming short photocurrent integration times so that we may ` freeze ' the relevant internal dynamics over the sample interval , taking @xmath84 .
this approximation also implies the assumption that the gaussian fluctuations in the input vacuum field are not correlated with system state over the effective measurement time interval . as a result ,
in this approximation @xmath82 is equivalent to a convolution of the wigner function for the system state and a symmetric , mean - zero gaussian distribution with a variance set by the effective integration time ( which represents the contribution of vacuum field fluctuations to the measurement distribution ) .
these methods and approximations were employed in fig . [
fig : ss ] of the main article .
similarly , integrating eq . over
some @xmath85-quadrature produces the characteristic functional @xmath86 = \langle \mathcal{t}\exp\left\{i\int_0^tk_{s}dy^\dag_{\theta , s}\right\}\rangle,\ ] ] where @xmath87 is an aribtrary real - valued scalar function of time and @xmath88 is the homodyne photocurrent operator of quadrature @xmath80 .
this functional may be used to calculate moments of instantaneous photocurrent measurements @xcite @xmath89\vert_{k=0}.\ ] ] for instance , the mean amplitude - quadrature photocurrent at time @xmath76 is calculated to be @xmath90\ ] ] where @xmath91 , while the two - time photocurrent correlation is ( @xmath92 ) @xmath93.\ ] ] note that the first term on the rhs of eq .
may be identified as the shotnoise contribution to the photocurrent correlation function , while the second is the contribution from the system .
as the detector - filtered photocurrent operator is @xmath94 where @xmath95 is some filter function imposed by the detection at time @xmath76 ( e.g. a low - pass filter initiated at time @xmath76 ) , the above two equations may in principle be used to calculate the mean and variance of filtered photocurrent measurements . in practice , though , we again invoke a small integration time approximation , justified by our high bandwidth detection .
if we approximate the state @xmath75 as static over the effective integration time ( as far as detection is concerned ) , we may take @xmath75 in the rhs of eqs . and as independent of time over the effective integration interval and approximate @xmath96 , greatly simplifying these calculations .
these approximations again allow us to model the detector filter function by a simple time - averaging filter of width @xmath97 , where @xmath98mhz is the cut - off frequency of the two - pole low pass filter that sets the bandwidth of the data presented in the main article .
thus the mean and variance of the photocurrent measurements expected from our detector when the system is in the state @xmath75 may be calculated from @xmath99\nonumber\\ \langle\mathcal{i}_t^2\rangle & = & \tau -\tau^2{\text tr}\left [ \mathcal{m}_0 ^ 2\rho_t\right]\end{aligned}\ ] ] these methods and approximations were used to calculate the theory curves in fig .
[ fig : am ] of the main article . integrating the schrodinger - picture master equation for
am drive ( initiating system and drive in the ground state ) returns an expected ensemble system state @xmath75 at each point in a drive cycle .
these states were then used to calculate the expected ensemble mean and standard deviation of the photocurrent measurements using eqs . , given the bandwidth of the data .
relating the master equation s quantum ensemble , single cycle predictions to the several cycle , single shot data in fig .
[ fig : am ] is justified by an assumption of the system s ergodicity ( see below ) . finally , we relied on quantum trajectory methods @xcite , which compliment the above master equation - based approaches to simulate typical measurements expected from our apparatus in figs .
[ fig : singles ] and [ fig : corr ] . while the master equation may be used to model ensembles of experimental realizations , a simulated quantum trajectory may be used to construct potential experimental homodyne measurement sequences , correctly sampled from the space of all possible sequences .
for example , simulation of an amplitude quadrature photocurrent given a set of experimental parameters first involves the calculation of a possible trajectory for the internal quantum state vector @xmath100 by numerically integrating the stochastic schrodinger equation @xcite @xmath101 where @xmath102 are randomly generated , independent wiener increments , @xmath103 denotes expectation with respect to @xmath100 and the state vector is forcibly re - normalized after each recursive update .
the simulated photo - increment @xmath104 may then be obtained using this state trajectory and calibrated detection efficiency @xmath78 by @xmath105 where @xmath106 is a third , independent wiener increment .
reproduction of several figures in the main article replacing experimental data with simulated photocurrent data produced by quantum trajectory methods .
( a ) , simulated amplitude quadrature photocurrent data with @xmath25 , 20mhz analog bandwidth and perfect detection efficiency .
( b ) , same quantum trajectory realization as ( a ) , but with the 20mhz bandwidth photocurrent simulated with @xmath107 efficiency .
( c ) , histogram of simulated inefficient photocurrents in comparison to the expected distribution derived from master equation based techniques .
( d ) , sample mean and sample standard deviation of simulated photocurrents from fifty 1mhz am cycles , overlaying master equation - based predictions.,scaledwidth=85.0% ] again , quantum trajectory simulations more directly reproduce time - series , experimental measurements , while master equation - based simulations represent ensembles of trajectories or measurements . in fig .
[ fig : sims ] we present many of the same figures in the main article using simulated photocurrent data based on quantum trajectories in comparison with the master equation predictions . figs .
[ fig : sims](a - b ) depict simulated amplitude quadrature photocurrents with the same experimental parameters as in figs . [
fig : mf](c ) , [ fig : singles ] and [ fig : ss](d ) in the main article . in fig .
[ fig : sims](a ) , the measurement of the the mode leaking from the cavity is simulated as detected with perfect efficiency , while fig .
[ fig : sims](b ) simulates the same realization , but with our experimentally typical 20% efficiency , as in fig .
[ fig : singles](e - f ) . fig .
[ fig : sims](c ) histograms a simulated inefficient photocurrent trajectory in comparison with the expected amplitude quadrature distribution using eq .
( _ i.e. _ reproducing the amplitude quadrature component of fig .
[ fig : ss](d ) in the main article using simulated data )
. finally , fig .
[ fig : sims](d ) compares the sample mean and sample standard deviation of simulated amplitude quadrature measurements during 50 cycles of a 1mhz am external drive , for comparison to fig .
[ fig : am](d ) in the main article . both figs .
[ fig : sims](c ) and ( d ) utilize several times more aggregate data than the experimental data presented in the main article in order to articulate the convergence of predictions from quantum trajectories to those of master equation simulations . as we expect both types of systems to be stationary ergodic processes , we have empirically confirmed that the main discrepancies between quantum trajectory and steady state master equation based simulations observed in fig . [
fig : sims ] arise from the marginal appropriateness of the @xmath108 approximation when modeling 20mhz bandwidth signals derived from our system .
this work is supported by darpa - mto under award no . fa8650 - 10 - 1 - 7007 . | broadband homodyne detection of the light transmitted by a fabry - perot cavity containing a strongly - coupled @xmath0cs atom is used to probe the dynamic optical response in a regime where semiclassical theory predicts bistability but strong quantum corrections should apply .
while quantum fluctuations destabilize true equilibrium bistability , our observations confirm the existence of metastable states with finite lifetimes and a hysteretic response is apparent when the optical drive is modulated on comparable timescales .
our experiment elucidates remnant semiclassical behavior in the attojoule ( @xmath1 photon ) regime of single - atom cavity qed , of potential significance for ultra - low power photonic signal processing
. 99 j. l. obrien , a. furusawa , and j. vukovi , `` photonic quantum technologies , '' nature photonics * 3 * , 687 - 695 ( 2009 ) .
t. aoki , a. s. parkins , d. j. alton , c. a. regal , b. dayan , e. ostby , k. j. vahala , and h. j. kimble , `` efficient routing of single photons by one atom and a microtoroidal cavity , '' phys .
rev .
lett . *
102 * , 083601 ( 2009 ) .
l. dicarlo , m. d. reed , l. sun , b. r. johnson , j. m. chow , j. m. gambetta , l. frunzio , s. m. girvin , m. h. devoret , and r. j. schoelkopf , `` preparation and measurement of three - qubit entanglement in a superconducting circuit , '' nature * 467 * , 574 - 578 ( 2010 ) . c. savage and h. .j .
carmichael , `` single - atom optical bistability , '' ieee j. quantum electron .
* 24 * , 1495 - 1498 ( 1988 ) .
g. rempe , r. j. thompson , r. j. brecha , w. d. lee , and h. j. kimble , `` optical bistability and photon statistics in cavity quantum electrodynamics , '' phys .
rev .
lett . * 67 * , 1727 ( 1991 ) .
m. a. armen and h. mabuchi , `` low - lying bifurcations in cavity quantum electrodynamics , '' phys .
rev .
a * 73 * , 063801 ( 2006 ) .
j. kerckhoff , m. a. armen , d. s. pavlichin , and h. mabuchi , `` the dressed atom as binary phase modulator : towards attojoule / edge optical phase - shift keying , '' opt .
express * 19 * , 6478 - 6486 ( 2011 ) .
d. a. b. miller , `` are optical transistors the logical next step ? , '' nature photonics * 4 * 3 - 5 ( 2010 ) .
s. h. strogatz , _ nonlinear dynamics and chaos : with applications to physics , biology , chemistry and engineering _ ( perseus , cambridge , ma , 1994 ) .
h. mabuchi , `` coherent - feedback control strategy to suppress spontaneous switching in ultralow power optical bistability , '' appl .
. lett . * 98 * , 193109 ( 2011 ) . k. srinivasan and o. painter , `` linear and nonlinear optical spectroscopy of a strongly coupled microdisk - quantum dot system , '' nature * 450 * , 862 - 865 ( 2007 ) .
a. faraon , i. fushman , d. englund , n. stoltz , p. petroff , and j. vukovi , `` coherent generation of non - classical light on a chip via photon - induced tunnelling and blockade , '' nature physics * 4 * , 859 - 863 ( 2008 ) .
l. a. lugiato , in _ progress in optics _ , edited by e. wolf ( north - holland , amsterdam , 1984 ) , vol .
xxi .
a. szke , v. daneu , j. goldhar , and n. a. kurnit , `` bistable optical element and its applications , '' appl .
. lett . * 15 * , 376 ( 1969 ) .
s. d. smith , `` optical bistability , photonic logic , and optical computation , '' appl .
opt . * 25 * , 1550 - 1564 ( 1986 ) .
s. ya .
kilin and t. b. krinitskaya , `` single - atom phase bistability in a fundamental model of quantum optics , '' j. opt .
soc . am .
b * 8 * , 2289 ( 1991 ) . c. j. hood , m. s. chapman , t. w. lynn , and h. j. kimble , `` real - time cavity qed with single atoms , '' phys
. rev .
lett . *
80 * , 4157 - 4160 ( 1998 ) .
x. yang , c. husko , c. w. wong , m. yu , and d. l. kwong , `` observation of femtojoule optical bistability involving fano resonances in high - q / vm silicon photonic crystal nanocavities , '' appl . phys .
lett . * 91 * , 051113 ( 2007 ) .
k. nozaki , t. tanabe , a. shinya , s. matsuo , t. sato , h. taniyama , and m. notomi , `` sub - femtojoule all - optical switching using a photonic - crystal nanocavity , '' nature photonics * 4 * , 477 - 483 ( 2010 ) .
h. j. carmichael in _ frontiers in quantum optics _ , edited by e. r. pike and s. sarkar ( adam hilger , bristol , 1986 ) .
h. gang , c. z. ning and h. haken , `` codimension - two bifurcations in single - mode optical bistable systems , '' phys .
rev .
a * 41 * , 2702 ( 1990 )
. h. gang , c. z. ning and h. haken , `` distribution of subcritical hopf bifurcations and regular and chaotic attractors in optical bistable systems , '' phys .
rev .
a * 41 * , 3975 ( 1990 ) .
h. mabuchi , `` derivation of maxwell - bloch - type equations by projection of quantum models , '' phys .
rev .
a * 78 * , 015801 ( 2008 ) .
m. a. armen , a. e. miller , and h. mabuchi , `` spontaneous dressed - state polarization in the strong driving regime of cavity qed , '' phys .
rev .
lett .
* 103 * 173601 ( 2009 ) .
a. c. doherty , a. s. parkins , s. m. tan , and d. f. walls , `` motion of a two - level atom in an optical cavity , '' phys .
rev .
a * 56 * , 833 ( 1997 )
. s. m. tan , `` a computational toolbox for quantum and atomic optics , '' j. opt .
b : quantum semiclass
. opt .
* 1 * , 424432 ( 1999 ) .
j. kerckhoff , h. i. nurdin , d. s. pavlichin , and h. mabuchi , `` designing quantum memories with embedded control : photonic circuits for autonomous quantum error correction , '' phys .
rev .
lett . * 105 * 040502 ( 2010 ) .
h. mabuchi , q. a. turchette , m. s. chapman , and h. .j .
kimble , `` real - time detection of individual atoms falling through a high - finesse optical cavity , '' opt .
lett .
* 21 * , 13931395 ( 1996 ) .
r. w. p. drever , j. l. hall , f. v. kowalski , j. hough , g. m. ford , a. j. munley , h. ward , `` laser phase and frequency stabilization using an optical resonator , '' appl .
phys .
b * 31 * 97 ( 1983 ) .
p. berman . , ed . ,
_ cavity quantum electrodynamics _
( san diego : academic press , 1994 ) .
h. j. carmichael , _ an open systems approach to quantum optics _ ( springer , berlin , 1993 ) . c. w. gardiner and p. zoller , _ quantum noise _
( springer - verlag , berlin , 2004 ) .
a. barchielli , `` direct and heterodyne detection and other applications of quantum stochastic calculus to quantum optics , '' quantum opt .
* 2 * 423 - 441 ( 1990 ) .
research over the past decade in single - atom cavity quantum electrodynamics ( cqed ) has largely focused on the generation of non - classical states of light and on the development of devices for quantum information protocols @xcite .
however single - atom cqed with strong coupling is also a natural context in which to study the interplay of nonlinear mean - field dynamics and quantum fluctuations @xcite , which is a topic of equally broad fundamental interest and with potentially greater near - term relevance for information technology @xcite .
dynamical systems theory @xcite provides key methods and insights for nonequilibrium statistical mechanics , as well as essential tools for analyzing and engineering nonlinear phenomena .
substantial work will be required to generalize these methods to encompass quantum coherence and fluctuations for applications in the emerging disciplines of nanoscale and quantum engineering . absent a systematic understanding of quantum stochastic nonlinear phenomena in the strong coupling regime , intuitive connections to semiclassical theory
provide important guidance for analyzing and engineering the behavior of open quantum systems @xcite . here
we show that the optical response of a strongly driven single - atom cqed system clearly displays competing influences of semiclassical bistability and quantum fluctuations , in quantitative agreement with theory .
our experiment utilizes a single gas - phase @xmath0cs atom as the nonlinear medium in a fabry - perot optical resonator , but we note that the same physics should also be relevant to quantum nonlinear dynamics in strongly coupled nanophotonic systems @xcite .
we study the dynamic input - output properties of the atom - cavity system in a parameter regime that semiclassically @xcite would be expected to exhibit absorptive bistability with a hysteresis loop @xcite suitable for attojoule ( @xmath1 photon ) optical switching @xcite .
theoretical studies have shown @xcite however that quantum fluctuations of the cavity field and atomic dipole should induce spontaneous switching between low- and high - transmission states , destroying true optical bistability , when the energy separation of the semiclassical dynamical attractors reaches the few - photon scale .
previous experimental studies have proven the need to utilize fully quantum models for predicting the steady - state optical response in single - atom cqed with strong driving @xcite . here
we take an important step further by recording dynamic signals that reaffirm the quantum mechanical model , but nonetheless reveal remnant behavior reminiscent of the semiclassical picture even in the deep quantum regime . as research on nanophotonic logic devices pushes towards attojoule - scale switching energies @xcite ,
functionalizing such remnant behaviors will become a definitive challenge .
the maxwell - bloch equations ( mbes ) represent a semiclassical mean - field approximation to the cqed master equation ( see appendix [ sec : models ] ) in the limit of weak coupling and large atom number @xcite and have been used in research on optical bistability with atomic ensembles as the intracavity nonlinear medium @xcite . in single - atom cavity qed with strong coupling and saturating driving fields , the mbes should not apply _ a priori _ as they ignore correlations between the atom and intra - cavity photons .
however , the mbes do retain some relevance as a projection of the quantum master equation onto the sub - manifold of semiclassical atom - field states @xcite and numerical studies have shown @xcite that solutions of the full quantum model often mimic those of the mbes qualitatively .
the mbes predict absorptive optical bistability for the parameters of our experiment , as shown in fig .
[ fig : mf](a ) .
the vertical axis represents input drive amplitude ( with @xmath2 per unit on the vertical axis ) , and the solution curve passes through horizontal planar coordinates propotional to the real and imaginary parts of the output steady - state complex field amplitude(s ) .
the projection of the curve onto the bottom face of the coordinate box indicates the phases of the steady - state solutions ( relative to the optical phase of the laser drive ) , while the projection onto the back face illustrates the input - output relation for amplitude - quadrature homodyne detection .
in this representation , a characteristic ` s - curve ' appears with two stable solutions and one unstable solution co - existing for @xmath3 in the range of @xmath4\kappa$ ] . using identical parameters
, the steady - state solution of the quantum master equation can be determined for each value of the drive strength .
contours of the corresponding intracavity field wigner functions ( tracing over the atomic states ) are displayed for several drive strengths in fig .
[ fig : mf](b ) . a double - peaked structure emerges along the amplitude quadrature in a range of @xmath5 similar to the bistable region of the mbes .
the bimodal steady states correspond to an incoherent mixture of two states : a weakly excited atom and low - amplitude field state , and a fully saturated atom and a high - amplitude field state .
thus , the two peaks of the distribution may be qualitatively associated with the low- and high - amplitude branches of the semiclassical absorptive bistability curve , but in the quantum model neither is truly stable @xcite .
it can be seen that in any single trial the cavity field ( and thus the output power ) spontaneously switches between low and high states , as in the experimental data of fig .
[ fig : mf](c ) ( described below ) .
this stochastic switching is a dramatic consequence of quantum fluctuations in this non - linear , attojoule - scale optical system . in our experiment @xcite ,
laser cooled @xmath0cs atoms are dropped into a fabry - perot optical resonator ( length 27@xmath6 m , 10 cm radius of curvature mirrors , field decay rate @xmath7 9.3mhz ) supporting a circularly - polarized , 852 nm tem@xmath8 mode actively frequency - stabilized relative to the @xmath9 atomic cycling transition ( with dipole decay rate @xmath10mhz ) . as an atom falls through the cavity mode it experiences a position - dependent coupling rate @xmath11 ( maximum @xmath1256.8mhz at the cavity anti - nodes ) that can be monitored via the transmission of a weak and detuned circularly - polarized optical probe . once a strongly coupled atom is detected , the probe power and detuning are adjusted for optical homodyne detection and 200ms / s data acquisition ( first arrow at @xmath13s on the time axis in fig . [
fig : mf](c ) . fig .
[ fig : mf](c ) depicts a representative signal ( all data has been post - filtered at 20mhz bandwidth for clarity ) ; with the drive amplitude held at a fixed value of @xmath14 , the amplitude quadrature of the transmitted field fluctuates with a large variance until @xmath15s ( red arrow ) when the atom is abruptly lost and and the measured transmission settles to the shot noise - variance signal with intermediate mean amplitude expected for our cavity when empty ( perturbed only by slight @xmath161 - 10khz mechanical instabilities ) .
see appendix [ sec : exp ] for more explanation of the apparatus . zooming in on segments of the trace in fig .
[ fig : mf](c ) reveals significant dynamics on @xmath6s time scales .
while the signals after the atom has left the cavity have white noise , gaussian statistics as expected ( e.g. the @xmath17s segment from fig .
[ fig : mf](c ) is shown in fig . [
fig : singles](a ) ) single shot signals observed when the drive is first turned on are more volatile ( the @xmath18s segment from fig .
[ fig : mf](c ) is shown in fig . [ fig : singles](b ) ) , with suggestive , sharp transitions between low and high outputs and a seemingly bimodal distribution . figs .
[ fig : singles](c - d ) depict two more high variance segments from two more experimental trials .
the significance of these observations may be understood from simulated measurement signals using quantum trajectory techniques ( see appendix [ sec : models ] ) .
fig .
[ fig : singles](e ) shows a simulated amplitude quadrature measurement segment , assuming perfect detection efficiency of all photons that decay from the cavity .
this simulated signal randomly switches between two , roughly gaussian - distributed output states , one with a low mean and one with a high mean .
this signal s bimodal distribution may be directly related to the bimodal , steady state wigner function for the intracavity field calculated in fig .
[ fig : mf](b ) ( discussed futher below ) .
re - simulating the same realization of the signal accounting for the calibrated overall photon collection efficiency ( 20% in the particular data set from which traces in fig .
[ fig : singles](a - d ) were produced ) , yields a signal in which binary switching is much less distinct in fig .
[ fig : singles](f ) , but resembles the single shot data shown in figs . [ fig : singles](b - d ) in both the visibility and apparent time scales of the large fluctuations .
( a - c ) wigner function representations of the expected steady state photocurrent distributions , using calibrated detection efficiencies and bandwidths , for three sets of detuning and drive parameters .
marginal distributions of homodyne measurements of any quadrature may be obtained by integrating these representations over the perpendicular quadrature ( see appendix [ sec : models ] ) .
( d - f ) histograms represent the phase- ( @xmath19 ) and amplitude - quadrature ( @xmath20 ) photocurrent distributions from ensembles of the highest - variance segments .
the histograms are compared with theoretically expected distributions obtained from corresponding wigner functions ( a - c ) , respectively.,scaledwidth=85.0% ] to confirm the steady state quantum mechanical model in the high detection bandwidth regime more quantitatively , in fig .
[ fig : ss ] we compare amplitude- and phase - quadrature photocurrent distributions obtained from three different sets of mode - drive , atom - drive detunings and drive amplitudes ( @xmath21 , @xmath22 , and @xmath3 , respectively ) , with @xmath1615% overall cavity photon detection efficiency ( independently calibrated for each parameter set , with variations attributed to thermal drifts in the signal - local oscillator optical mode matching ) .
each histogram is computed from an aggregate of several few-@xmath6s - long segments of the highest variance photocurrent data , such as those presented in figs .
[ fig : singles](b - d ) .
while enhanced amplitude noise is always visible when the drive is first triggered , a minority of events exhibit fluctuations of near - maximal magnitude since our atom detection scheme is triggered by strongly coupled atoms ( @xmath23 ) that may still be significantly less than near - maximally coupled ; a maximum of @xmath24 is achievable with our apparatus ( see appendices [ sec : exp ] and [ sec : models ] ) . because of this post - selection to isolate near - maximally coupled atom transits , useful segments of photocurrent are rather sparse in our overall data set .
we are thus limited to the presentation of small number statistics , but given the very distinctive features of the bistability - related signals ( to be discussed below ) and the straightforward nature of our selection criterion ( highest atom - induced photocurrent variance ) , we are confident that our analysis procedure enables us to draw physically meaningful conclusions . for a near - detuned system at the onset of ` bistability ' with @xmath25 , wide / bimodal and narrow / normal distributions are apparent in the amplitude- and phase - quadrature distributions , respectively ( fig .
[ fig : ss](d ) ) .
however , when the atom - drive detuning is increased to @xmath26 for the same drive amplitude , the low - amplitude transmitted field dominates ( fig .
[ fig : ss](e ) ) as the drive threshold for ` bistability ' increases with @xmath27 .
the bimodal amplitude distribution reemerges when the drive amplitude is increased to @xmath28 in fig .
[ fig : ss](f ) .
the data are in agreement with quantum theoretical predictions ( see appendix [ sec : models ] ) , despite the use of a somewhat idealized model that assumes a static coupling rate @xmath11 . whereas @xmath11 actually depends upon atomic position and zeeman sub - state , and can vary within a photocurrent segment because of complex atomic motion and imperfect optical polarization after many @xmath6s @xcite , laser cooling and the cavity aperture ensures that atoms fall transversely through the mode in @xmath29s , stochastic heating should induce diffusion @xmath30100 nm during the first 5@xmath6s of strong probing , and optical dipole forces should be minimal for near - maximally coupled atoms .
while technical noises like fluctuations in the atomic coupling during a transit , cavity instability and laser noise could also give rise to signals with super - shot noise variance , such sources may be ruled out by a combination of being too small to be of importance , of occurring only on significantly longer time scales , and/or the fact that they would induce super shot - noise distributions in both quadratures roughly equally due to our @xmath31 probe detuning ( such quadrature - independent fluctuations appear to be minimal in fig . [
fig : ss ] , whereas the amplitude quadrature - only ` quantum switching ' is prominent ) .
although the complexity of the measurements dependence on @xmath11 precludes statistically rigorous parameter estimation , we find that a fixed effective value @xmath32 in our analysis provides a good visual fit to the distributions of these short - time datasets ( corresponding to a near - optimal @xmath33 if one makes the gross simplification of fitting the experimental histograms to theoretical marginal distributions with fixed @xmath11 , assuming that the uncertainties of the histogram data points are constant and uncorrelated ) in all three parameter sets with @xmath34 uncertainty in @xmath11 and no other free parameters .
we believe that this approximation of fixed @xmath11 and the finite ( 20 mhz ) bandwidth of our presented homodyne signals account for slight mismatches between theory and experiment in the amplitude quadrature splitting and phase - quadrature mean in the six data sets .
the autocorrelation function for the same aggregated amplitude quadrature photocurrent data presented as a histogram in figure [ fig : ss](d ) is displayed in black .
blue crosses represent the autocorrelation of photocurrents simulated by quantum trajectory methods @xcite for identical parameters , as in fig .
[ fig : singles](f ) .
the stability of these quasi - bistable signals is enhanced relative to linearly scaled empty cavity transmission data taken after the atom is lost ( red pluses , also presented with a 20mhz analog bandwidth and scaled to match the @xmath35 and @xmath36s autocorrelation of the high - variance ` experiment ' signal ) , and the @xmath37ns cavity decay time ( dashed orange line ) characterizing the intracavity field relaxation of the empty resonator .
we attribute the elevated and noisy autocorrelation of the atom - cavity transmission data at @xmath38s timescales to dynamic fluctuations in @xmath11 , as in fig [ fig : ss].,scaledwidth=60.0% ] ( a - b ) single - shot , amplitude quadrature measurements as the drive amplitude is swept at .25mhz and 1mhz , respectively .
black traces are 20mhz bandwidth photocurrents while the green , dashed traces represent the instantaneous drive amplitude .
( c - d ) dashed red ( blue ) traces portray the same photocurrent data in ( a ) and ( b ) , respectively , as a function of the instantaneous , increasing ( decreasing ) drive amplitude .
error bars represent sample mean and sample standard deviation of the same photocurrents within non - overlapping drive amplitude intervals .
red and blue regions represent theoretically expected photocurrent mean and sample variance as a function of instantaneous drive ( see appendix [ sec : models ] ) .
( e - f ) only the sample means of 3 additional single shot measurements of similar duration are plotted in each figure as a function of the instantaneous drive amplitude , as in ( c ) and ( d ) , overlaying the same theoretically expected photocurrent statistics.,scaledwidth=77.0% ] hence , even in the single - atom , @xmath1610 photon regime , distinct high- and low - amplitude states of the output field are not washed out completely by quantum fluctuations .
remnant signatures of optical bistability are visible in the bimodal output photocurrents of figs .
[ fig : singles ] and [ fig : ss ] .
similarly , as suggested in fig .
[ fig : singles ] , we see in fig .
[ fig : corr ] that although the output field switches spontaneously when an atom is present in the cavity , it remains correlated over timescales much longer than that of light transmitted through an empty cavity .
this atom - induced memory effect can be seen as an additional remnant of mbe - type optical bistability , where classically the high- and low - amplitude states are truly stable ( due to negligible fluctuations in the system ) and would therefore exhibit infinite correlation time .
consequently , it should also be possible to observe the hysteretic amplitude response characteristic of classical optical bistability by modulating the system drive slowly compared to the timescale for relaxation of the intracavity field ( set in our case by the cavity decay time ) but rapidly compared to the ` metastable ' memory timescale indicated in fig . [
fig : corr ] .
accordingly , the data in fig .
[ fig : am ] were obtained by recording amplitude - quadrature homodyne photocurrents while sweeping the drive strength sinusoidally at 0.25mhz or 1mhz .
figs .
[ fig : am](a ) and [ fig : am](b ) depict representative single - shot photocurrent segments with @xmath39 encompassing several cycles of sinusoidal drive amplitude modulation ( am ) spanning the steady state bimodal region .
increases in both the mean and variance of the output photocurrent , largely in phase with the drive amplitude , can be discerned in both of these real - time plots . however , plotting the photocurrent as a function of the instantaneous drive amplitude ( figs .
[ fig : am](c ) and [ fig : am](d ) ) reveals a significant hysteresis in the system response at 1mhz am that is barely noticeable at the more adiabatic .25mhz am rate . whereas the response of the empty cavity is linear and non - hysteretic with fixed ( shot - noise ) output photocurrent variance at these modulation frequencies , nonlinear increases in the signal mean and variance are evident in both traces at mid - sweep . at 1mhz am , a hysteresis loop appears to open between the upward and downward drive amplitude sweeps , with the low ( high ) state persisting over a wider range of increasing ( decreasing ) drive amplitudes than at .25mhz am .
moreover , we emphasize that despite the large variance in the signals when viewed at such high bandwidth , the transition from a non - hysteretic to hysteretic mean response as the am rate increases is significant , as highlighted in figs . [
fig : am](e - f ) . these data are consistent with theoretical predictions ( see appendix [ sec : models ] ) , again assuming the same effective value of @xmath40 determined from the short - time segments analyzed in fig . [
fig : ss ] . in conclusion , we have probed the dynamic optical response of a driven , strongly coupled , single - atom cqed system in the vicinity of atomic saturation .
while our observations further confirm the predictions of a fully - quantum model , qualitative remnants of semiclassical absorptive bistability are clearly visible in the data .
while the quantum fluctuation - limited lifetimes of the low- and high - amplitude states in attojoule optical ` bistability ' may be too short for direct use in photonic switching , our results clearly illustrate that existing theoretical models can be used to predict and analyze the dynamic response of real devices in the few photon regime .
such models can be used to make detailed predictions of the impact of quantum effects on ultra - low energy switch performance , which may be of interest to the nanophotonic engineering community . beyond mere simulation and analysis ,
existing theoretical methods can be used to explore new approaches to the suppression of quantum fluctuations in the design of switches and related nanophotonic devices , for example by exploiting embedded coherent feedback control @xcite . even in a purely classical information processing paradigm ,
high spatial - density and ultra - low power nanophotonic circuit design presents intriguing new challenges for the nascent applied physics discipline of quantum engineering . |
* fig . 1 : * the diagram that generates the @xmath28 " entries of the quark and lepton mass matrices .
* fig . 3 : * the diagrams that produce small masses for the first - family quarks and leptons .
the family index @xmath172 on the spinor field takes the values 2 or 3 , giving , respectively , the @xmath38 and @xmath26 terms of the mass matrices .
different @xmath5 vector higgs , denoted with superscript @xmath172 , are exchanged in the two cases . *
4 * a comparison of the model with experiment for the wolfenstein parameters ( @xmath0 , @xmath1 ) .
the axes are @xmath0 and @xmath1 multiplied by the central value of @xmath173 .
the central values allowed by the model lie on the bold dashed circular arc , cf .
the constraints following from @xmath174-mixing and @xmath28 extractions from experimental data are shown in the lightly shaded regions .
the experimentally allowed region is indicated by the heavily shaded central region . a typical unitarity triangle allowed by both data and
the model is shown .
( 360,216 ) ( 60,144)(1,0)30 ( 90,144)(1,0)30 ( 120,144)(1,0)30 ( 180,144)(-1,0)30 ( 180,144)(1,0)30 ( 210,144)(1,0)30 ( 240,144)(1,0)30 ( 300,144)(-1,0)30 ( 120,84)(0,1)12 ( 120,108)(0,1)12 ( 120,132)(0,1)12 ( 180,132)(0,1)12 ( 180,120)(0,-1)12 ( 180,84)(0,1)12 ( 240,84)(0,1)12 ( 240,108)(0,1)12 ( 240,132)(0,1)12 ( 82,158)@xmath175 ( 142,158)@xmath176 ( 202,158)@xmath177 ( 262,158)@xmath178 ( 128,100)@xmath179 ( 188,100)@xmath180 ( 248,100)@xmath181 ( 162,45)*fig .
1 * ( 360,216 ) ( 60,144)(1,0)30 ( 90,144)(1,0)30 ( 120,144)(1,0)30 ( 180,144)(-1,0)30 ( 180,144)(1,0)30 ( 210,144)(1,0)30 ( 240,144)(1,0)30 ( 300,144)(-1,0)30 ( 120,84)(0,1)12 ( 120,108)(0,1)12 ( 120,132)(0,1)12 ( 180,132)(0,1)12 ( 180,120)(0,-1)12 ( 180,84)(0,1)12 ( 240,84)(0,1)12 ( 240,108)(0,1)12 ( 240,132)(0,1)12 ( 78,158)@xmath182 ( 142,158)@xmath183 ( 202,158)@xmath184 ( 262,158)@xmath185 ( 125,100)@xmath186 ( 188,100)@xmath180 ( 248,100)@xmath187 ( 162,45)*fig . 2
* ( 360,216 ) ( 90,180)(1,0)45 ( 135,180)(1,0)45 ( 270,180)(-1,0)45 ( 180,180)(1,0)45 ( 180,120)(0,1)12 ( 180,144)(0,1)12 ( 180,168)(0,1)12 ( 127,194)@xmath188 ( 217,194)@xmath189 ( 190,148)@xmath190 ( 96,120)(1,0)12 ( 120,120)(1,0)12 ( 144,120)(1,0)12 ( 168,120)(1,0)12 ( 180,120)(1,0)12 ( 204,120)(1,0)12 ( 240,120)(-1,0)12 ( 252,120)(1,0)12 ( 124,98)@xmath191 ( 208,98)@xmath192 ( 162,60)*fig . 3 * | a model is presented that fits the quark and lepton masses and mixings wherein five dimensionless parameters and a phase account for fifteen dimensionless observables . among these
are the wolfenstein parameters @xmath0 and @xmath1 , the @xmath2 and @xmath3 mixing angles which are predicted to be small and comparable while the @xmath4 mixing angle is predicted to be large . the model is based on supersymmetric @xmath5 with the form of the mass matrices motivated by simplicity at the level of grand unification . + pacs numbers : 12.15.ff , 12.10.dm , 12.60.jv , 14.60.pq 7.6 in + + + * predicting quark and lepton masses and mixings + * carl h. albright@xmath6 + department of physics + northern illinois university , dekalb , il 60115 + and fermi national accelerator laboratory + p.o .
box 500 , batavia , il 60510 + s.m .
barr@xmath7 + bartol research institute + university of delaware + newark , de 19716 + @xmath8 + @xmath6e - mail : [email protected] ; @xmath7e - mail : [email protected] + in an earlier paper with babu @xcite , the authors have proposed a model based on supersymmetric @xmath5 for the masses and mixings of the heavier two families of quarks and leptons .
one of the guiding principles in constructing the model was that the mass terms should be of a kind that can arise simply in @xmath5 with small representations of higgs fields @xcite .
two critical points emerged from that work : namely , that it is quite natural in unified theories for there to be flavor - asymmetric contributions to the fermion mass matrices , and that these can explain some of the puzzling features of the spectrum .
in particular , it was shown that a flavor - asymmetric term can account in a simple way for the apparent fact that the @xmath9 mixing is very large compared to @xmath10 .
the present work is an extension of that earlier model to include the first family .
we find that it is only necessary to introduce two new parameters , one of which has a complex phase , in order to account for seven quantities that involve the first family . in doing so , several interesting predictions
result , including the values of the @xmath2 and @xmath11 mixing angles , and a relationship between the real and imaginary parts of @xmath12 , _
i.e. _ , between the wolfenstein parameters @xmath0 and @xmath1 . altogether
, as will be seen , the model leads to nine predictions .
* the mass matrices . *
the mass matrices proposed have the following forms : @xmath13 where @xmath14 , @xmath15 , @xmath16 , and @xmath17 stand , respectively , for the up - quark , down - quark , charged - lepton , and the neutrino dirac mass matrices .
the zero superscripts stand here and throughout for quantities defined at the grand unification ( gut ) scale .
the convention we use is that the left - handed fermions multiply these matrices from the right , while the right - handed fermions multiply them from the left .
( thus , it is the entries below the main diagonal that control the quark and lepton mixings that are observed . )
it is immediately apparent that the matrix @xmath18 is rank-2 , corresponding to a massless @xmath19 quark .
we imagine that some small higher - order effects give @xmath19 a mass of several mev .
the fact that @xmath20 must arise at higher order than @xmath21 and @xmath22 is not a flaw but a qualitative success of the model .
the point is that a several - mev @xmath19-quark mass corresponds to @xmath23 .
this is very much smaller than the corresponding ratios @xmath24 and @xmath25 , and so it is reasonable to suppose that it is a higher - order effect . by field redefinitions
one can make all the elements of the matrices in eq .
( 1 ) real except for @xmath26 .
( there can also be a relative phase _ initially _ between @xmath27 and @xmath28 , which was called @xmath29 in @xcite .
after phase redefinitions , however , this parameter survives only in the entries @xmath30 and @xmath31 , which can be approximated by @xmath32 and @xmath33 , respectively , since as will be seen later @xmath34 is a small parameter .
as @xmath35 only appears in terms subleading in @xmath34 , it has only a few - percent effect on predictions . in @xcite
it was found that the fits prefer @xmath36 , so we will assume that value henceforth in this paper . ) effectively , then , the mass matrices in eq . ( 1 ) contain five real dimensionless parameters and one phase : @xmath37 , @xmath27 , @xmath28 , @xmath38 , @xmath39 , and @xmath40 .
we will first describe the kinds of @xmath5 operators and diagrams that give rise to the entries shown in ( 1 ) , and explain the rationale for the those entries .
then we shall proceed to discuss the fit to the data and the predictions of the model . * the origin and rationale of the matrices . * the ( 33 ) entries of the mass matrices contribute to the masses of the heaviest family of quarks and leptons . as is true of most grand unified models of fermion masses
, these entries can come from the minimal kind of yukawa term . since ours is an @xmath5 model ,
these entries arise simply from the operator @xmath41 . here and throughout , the subscript ` @xmath42 ' stands for a higgs multiplet , and a numerical subscript stands for the family number of a quark / lepton multiplet . the parameters @xmath43 and @xmath44 in eq .
( 1 ) come from the @xmath45-breaking vacuum expectation values ( vevs ) in @xmath46 and @xmath47 , respectively .
we use ( @xmath48 to represent a @xmath49 multiplet of @xmath50 contained in a @xmath51 multiplet of @xmath5 .
while the masses of the third family satisfy the minimal @xmath50 ( and @xmath5 ) relation @xmath52 , the middle family does not satisfy the corresponding @xmath53 relation .
as is well known , however , a good fit is given by the georgi - jarlskog relation @xcite , @xmath54 .
perhaps the simplest way to get a factor of 1/3 into the elements of the mass matrices is through an adjoint higgs multiplet , @xmath55 .
such a higgs multiplet is needed in any case for the breaking of @xmath5 down to the standard model group . in the simplest breaking scheme that naturally solves the doublet - triplet splitting problem " , the so - called dimopoulos - wilczek or
missing vev " mechanism @xcite , there is an adjoint higgs multiplet whose vev is proportional to the @xmath5 generator @xmath56 .
this is in fact just what is needed to get a relative factor of @xmath57 between entries in the quark and lepton mass matrices .
the lowest - dimension yukawa operators that contains the adjoint higgs multiplet are products of the form @xmath58 .
there are two ways to contract the @xmath5 indices , but it is easy to show that if @xmath59 , then only the _ flavor - antisymmetric _ piece contributes to the mass matrices .
this accounts for the form of the @xmath28 " entries in eq .
( 1 ) .
note that these appear with a factor of @xmath57 in the quark matrices and @xmath60 in the lepton matrices , reflecting the generator @xmath56 .
one also sees that because the operator being discussed contains the same @xmath61 that appears in the yukawa operator for the ( 33 ) entries , the @xmath28 entries of @xmath18 and @xmath62 are @xmath63 times those of @xmath64 and @xmath65 .
the operator @xmath66 needed for the @xmath28 entries arises in a simple fashion by integrating out a superheavy ( _ i.e. _ gut - scale ) family plus conjugate - family pair : @xmath67 and is naturally of order @xmath68 as required .
this is shown in fig . 1 .
such an operator could also arise from planck - scale physics , but it would then be too small .
the next pieces of the mass matrices to be discussed are the @xmath27 " entries .
these have several rationales , and are in fact the key to the whole model .
we shall first explain how such entries arise , and then why they are important for getting a realistic model .
the feature of the @xmath27 entries that distinguishes them from what is found in most published models is that they are neither flavor - symmetric nor flavor - antisymmetric , but rather lopsided " : @xmath27 appears in @xmath69 but not in @xmath70 .
though it is not common to consider such lopsided forms , there is really no reason from the point of view of grand unification not to do so .
in fact , it may well be the neglect of such possibilities that has handicapped the search for realistic quark and lepton mass matrices .
the @xmath27 entries arise very simply from integrating out an @xmath5 vector @xmath71 of quarks and leptons as shown in fig .
2 . the resulting operator is of the form @xmath72_{10 } [ { \bf 16}_3 { \bf 16}'_h]_{10}$ ] .
the notation @xmath73_{10}$ ] indicates that what is contained in the brackets is contracted into a @xmath71 of @xmath5 as a result of the virtual @xmath71 fermions appearing in fig .
2 .
the higgs multiplet denoted @xmath74 is assumed to develop a gut - scale vev in the @xmath75 direction , while the other spinor higgs , denoted @xmath76 , gets a weak - scale @xmath77-doublet vev in the @xmath78 direction .
thus , in @xmath50 notation , the operator from fig .
2 gives @xmath79 .
one observes that since the left - handed down - quarks are in @xmath71 s of @xmath50 , not @xmath80 s , this operator contributes to @xmath81 and not @xmath82 .
on the other hand , since the left - handed leptons are in @xmath80 s , not @xmath71 s , the operator contributes to @xmath83 and not @xmath84 .
as we shall see , this fact that the @xmath27 entry is transposed in @xmath16 relative to that in @xmath15 is very important for understanding the large @xmath85 mixing that may have been seen in the atmospheric neutrino data .
this transposition is a consequence of the group theory of @xmath50 which also implies that @xmath86 in the minimal @xmath50 model .
the upshot is that integrating out @xmath71 s of fermions in @xmath5 can quite typically give just such lopsided entries as the @xmath27 entries in eq .
( 1 ) .
note that the @xmath27 entries only appear in @xmath15 and @xmath16 , and not in @xmath14 and @xmath17 , following from the fact that @xmath76 contains only @xmath80 and not @xmath87 of @xmath50 .
now , why should such lopsided entries help to get a realistic model ?
it turns out that these @xmath27 entries play at least six vital roles : \(i ) in the first place , it is easy to see from the seesaw - type formulas , that setting @xmath88 in eq .
( 1 ) would give @xmath89 and @xmath90 .
thus one would get a ratio of ratios of 1/9 instead of the desired georgi - jarlskog value of 1/3 . if , however , @xmath27 is present and is large compared to @xmath28 , then one has @xmath91 , and @xmath92 , so that the desired georgi - jarlskog factor emerges .
\(ii ) without the @xmath27 entries , the ( 23 ) blocks of @xmath15 and @xmath14 would be proportional , and consequently there would be no km mixing for those families , that is @xmath10 would vanish .
\(iii ) moreover , the lopsided @xmath27 entries explain a peculiar fact about the magnitude of @xmath10 .
if @xmath81 and @xmath82 were of equal magnitude , as is the case with the purely flavor - symmetric or antisymmetric forms usually assumed , then @xmath10 would get a contribution from @xmath15 that is @xmath93 , which is much too big ( by about a factor of 4 ) . by having the large lopsided contribution with @xmath94 ,
one finds @xmath95 , which leads to @xmath96 , as observed .
this cures one of the main problems with textures of the fritzsch type @xcite as noted previously by branco and silva - marcos @xcite in a phenomenological study of the quark sector .
other authors @xcite have also recently appreciated this fact in the context of @xmath50 models .
\(iv ) because @xmath27 naturally only appears in @xmath15 and @xmath16 but not in @xmath14 , one explains why @xmath97 and @xmath98 , as actually observed . in the absence of the @xmath27 entries ,
one would obtain the minimal @xmath5 prediction @xmath99 , which is wrong by about a factor of 20 .
\(v ) the fact that the large @xmath27 entry appears transposed in @xmath16 relative to @xmath15 means that it makes the ( 32 ) mixing of the left - handed leptons _
large _ compared to the corresponding mixing of the left - handed quarks .
one thus explains why @xmath4 mixing is much larger than @xmath10 , which is one of the great puzzles in understanding the atmospheric neutrino oscillations .
\(vi ) finally , we shall see later that the @xmath27 entries explain why the @xmath100 mixing angle is significantly smaller than @xmath101 , assuming the small - angle msw solution of the solar neutrino problem . what remains to be explained about the mass matrices in eq .
( 1 ) are the entries @xmath38 and @xmath26 .
these arise very simply from the diagrams in fig .
3 . note that because the effective operators are again of the form @xmath102 , as for the @xmath27 entries , these diagrams only contribute to @xmath15 and @xmath16 , and not to @xmath14 and @xmath17 .
note also that because they come from integrating out a @xmath71 of higgs , they are _ flavor - symmetric_. these contributions are assumed to be much smaller than the others . the fact that the above terms appear and no others presupposes that some abelian symmetries exist that distinguish among families .
such symmetries have been used by many authors . for
an interesting recent publication with a number of references cited , see @xcite .
we note that realistic models involving such symmetries can be constructed giving rise to the matrices in eq .
( 2 ) .
much of the structure of such a model has already been described in @xcite and @xcite .
in addition to what was presented there , there have to be @xmath5 @xmath71 s of higgs fields to be integrated out to give the @xmath38 and @xmath26 entries , as in fig .
3 .
such fields and the necessary couplings can be introduced , with appropriate abelian symmetry , without destabilizing the gauge hierarchy or creating other difficulties @xcite . * the predictions of the matrices . * in expressing the predictions of the model
, it proves convenient to use instead of the parameters @xmath38 and @xmath26 , the linear combinations @xmath103 and @xmath104 . essentially , @xmath105 is the cabibbo angle , and @xmath106 is the corresponding mixing angle of the _ right - handed _
@xmath107 and @xmath108 quarks .
that is to say , if one diagonalizes the ( 23 ) block of @xmath64 in eq .
( 1 ) , and calls the resulting matrix @xmath109 , then @xmath110 and @xmath111 . in the phase convention being adopted , @xmath106 is real and @xmath105 has a phase that shall be called @xmath112 . the parameters , then , are @xmath63 , @xmath27 , @xmath28 , @xmath106 , @xmath113 , and @xmath112 . in terms of these one
has the following relations , which are given only to leading order in the small parameters @xmath28 , @xmath106 , and @xmath105 : @xmath114 & \hfill(2k)\\ & & & \\ u_{\mu 3 } & \equiv & \sin \theta_{\mu \tau } \cong - \frac{\sigma}{\sqrt{\sigma^2 + 1 } } + o(\epsilon ) & \hfill(2l)\\ & & & \\ u_{e2 } & \cong & - \frac{1}{3 } t_r \cos \theta_{\mu \tau } & \hfill(2m)\\ & & & \\
u_{e3 } & \cong & \frac{1}{3 } t_r
\sin \theta_{\mu \tau } & \hfill(2n)\\ \end{array}\ ] ] predictions involving only the second and third families were given already in @xcite , where a fit was done using more exact expressions , and taking into account renormalization - group running from the unification scale down to low scales as well as radiative corrections to the heavy quark masses coming from supersymmetry . from @xmath98 and @xmath10 which have small experimental uncertainties , @xmath28 and @xmath27 , which was previously called @xmath0 in @xcite , were extracted ; cf .
, eqs .
( 2e ) and ( 2i ) .
it was found that the best fit was given by @xmath115 from eq .
( 2c ) one can then obtain a prediction for the charm quark mass . in @xcite
it was found that @xmath116 gev , in satisfying agreement with the experimental value @xmath117 gev . from eq .
( 2l ) one sees that the model predicts a large , and in fact a nearly maximal , mixing of @xmath118 and @xmath119 . the element @xmath120 , which describes this mixing , receives a large contribution from diagonalizing the mass matrix of the charged leptons , @xmath16 .
this large contribution is well determined in the model , since @xmath16 is known , and comes out to be approximately @xmath121 .
however , @xmath122 also receives a small contribution from diagonalizing the neutrino mass matrix , which has the usual seesaw form @xmath123 @xcite .
while @xmath17 is known in the model , @xmath124 is not at least not without some further assumptions .
nevertheless , one can say that the contribution to @xmath122 from diagonalizing @xmath125 is of order @xmath28 , following from the form of @xmath17 .
though relatively small , this contribution makes impossible a precise prediction of the @xmath4 mixing angle , though it is possible to say that it is nearly maximal .
this source of uncertainty does not affect the predictions for the other two lepton mixing angles , however , because of the fact that @xmath17 , and therefore @xmath125 , have zeros in the first row and column .
these zeros are doubtless lifted by the higher order effects that generate @xmath20 , but it is easy to see that this should have a negligible effect on @xmath126 and @xmath127 .
the other predictions for the heavy families are that @xmath128 from eq .
( 2b ) and @xmath129 from eqs .
( 2d ) and ( 2e )
. both of these are well - known relations , and work well in the context of supersymmetric grand unification .
the foregoing four predictions were thoroughly discussed in @xcite .
let us turn now to the eight quantities that involve the first family : @xmath22 , @xmath21 , @xmath20 , @xmath126 , @xmath127 , @xmath130 , re@xmath131 , and im@xmath131 .
these depend on the three additional parameters @xmath106 , @xmath132 , and @xmath112
. the approximate magnitudes of @xmath106 and @xmath132 can be estimated in a simple way , as follows . comparing the well - known empirical relation @xmath133 with eqs .
( 2j ) and ( 2 g ) , one sees immediately that @xmath134 .
this enables one to read off an estimate of the wolfenstein parameters @xcite @xmath0 and @xmath1 from eq .
2 . to leading order in the wolfenstein parameter @xmath135
, one has that @xmath136 .
but the value of that invariant is given in this model ( to leading order in @xmath28 , @xmath106 , and @xmath105 ) by @xmath137.\ ] ] see eqs .
( 2i ) , ( 2j ) , and ( 2k ) .
since @xmath138 and @xmath139 one has , roughly , that @xmath140 corresponding to a circle in the @xmath141 plane . to get the predictions more exactly , it is best to use @xmath130 and @xmath142 , which have the smallest experimental uncertainties , to determine @xmath106 and @xmath132 ; cf .
, eqs .
( 2h ) and ( 2j ) . employing more exact expressions than given in eqs .
( 2 ) , one finds that @xmath143 the phase of @xmath105 , denoted @xmath112 , is so far undetermined . substituting these values into eq .
( 4 ) , and taking into account the corrections to that equation that are higher order in @xmath28 , @xmath106 , and @xmath105 , one finds @xmath144 this circle , plotted as the bold dashed arc in fig .
4 , slices nicely through the presently allowed region @xcite . the prediction for the @xmath2 mixing , cf . eq .
2(m ) , is that @xmath145 this is well within the present limits @xmath146 for the small angle msw solution to solar neutrino problem .
the crucial role of the parameter @xmath27 should be noted . in its absence
one would find the familiar type of prediction , @xmath147 , which is much too large .
however , since there is a large intermingling of the second and third families caused by the parameter @xmath27 the intermingling that causes the large @xmath4 mixing the angles @xmath126 and @xmath127 get intermingled as well .
this leads to the suppression of @xmath126 by a factor of @xmath148 , which brings it into the experimentally allowed range @xcite .
this same intermingling gives rise to a distinctive prediction of this model , namely that @xmath149 using more exact expressions , one finds that @xmath150 the remaining two predictions of the model are that @xmath151 , eq .
( 2f ) , which has been discussed already , and that @xmath152 , eqs .
( 2 g ) and ( 2h ) , the other georgi - jarlskog relation , which gives @xmath153 in excellent agreement with the current - algebra result .
a comment is in order about the robustness of these predictions . in obtaining the value of @xmath106
, we used @xmath142 on the grounds that the experimental uncertainty in this quantity is negligible .
however , one might question whether eq .
( 2h ) is trustworthy .
the point is that since some higher order effects are presumed to generate a mass for the @xmath19 quark , the same effects are likely to make a small contribution to the masses of @xmath107 and @xmath154 as well , though that is not necessarily the case .
for example , @xmath20 could arise from @xmath155 , and a comparable entry in @xmath16 of order @xmath156 would affect @xmath22 at the 3% level .
this would change @xmath106 , and thus @xmath126 , by the same amount .
the effect of such uncertainties on the prediction for the wolfenstein parameters @xmath157 is relatively small .
the prediction in eq .
( 9 ) is even more solid , since @xmath106 cancels out in the ratio of @xmath127 to @xmath126 . * conclusions * we have presented a model of quark and lepton masses and mixings that is based on supersymmetric @xmath5 .
the mass matrices arise from five yukawa terms , each of which is of a type that arises in a straightforward way from simple tree - level diagrams .
the model contains very few parameters , so few that there are nine predictions .
four of the predictions follow from the basic structure of the model and simple group - theoretic considerations .
these are @xmath151 ( which corresponds to the observed fact that @xmath158 , much less than @xmath159 and @xmath160 ) ; @xmath128 ( the well - known and successful @xmath50 relation ) ; @xmath161 ( one of the georgi - jarlskog relations ) ; and @xmath162 ( the other georgi - jarlskog relation ) .
four other predictions are non - trivial quantitative successes of the model .
they are a prediction of the charm quark mass that is correct within the present uncertainties ; the prediction of very large mixing between @xmath118 and @xmath119 , consistent with the recent discovery of atmospheric neutrino oscillations ; a prediction for @xmath12 that slices neatly through the middle of the allowed region for the wolfenstein parameters @xmath0 and @xmath1 ; and a prediction for @xmath163 that is well inside the present allowed range for the small - angle msw solution of the solar neutrino problem .
the last prediction is of a presently unmeasured quantity : @xmath164 .
the model is eminently testable .
the most dramatic test would be a measurement of @xmath165 , which could come from improved proton accelerator bounds or perhaps from a muon collider .
the next most important and fairly clean tests would come from a resolution of the solar neutrino problem and the consequent precise determination of @xmath166 ; and from a measurement of the wolfenstein parameters @xmath0 and @xmath1 .
much more difficult , but hopefully achievable someday , would be precision tests of the predictions for @xmath167 and @xmath168 .
this would require considerable knowledge of the spectrum of the sparticles , knowledge of @xmath169 and the higgs masses , and better determinations of @xmath170 , @xmath171 , and @xmath10 @xcite .
+ one of us ( cha ) thanks the fermilab theoretical physics department for its kind hospitality where much of his work was carried out .
the research of smb was supported in part by the department of energy under contract no .
de - fg02 - 91er-40626 .
fermilab is operated by universities research association inc .
under contract no .
de - ac02 - 76ch03000 with the department of energy
. 99 c.h .
albright , k.s .
babu , and s.m .
barr , phys .
rev .
lett .
* 81 * ( 1998 ) 1167 , hep - ph/9802314 ; to be published in proceedings of the neutrino 98 conference , hep - ph/9805266 .
c.h . albright and s.m .
barr , phys .
rev .
d * 58 * ( 1998 ) 013002 , hep - ph/9712488 .
h. georgi and c. jarlskog , phys .
lett . * 86b * ( 1979 ) 297 .
s. dimopoulos and f. wilczek , report no .
nsf - itp-82 - 07 ( 1981 ) , in _ the unity of fundamental interactions _ , proceedings of the 19th course of the international school of subnuclear physics , erice , italy , 1981 , ed .
a. zichichi ( plenum press , new york , 1983 ) ; k.s .
babu and s.m .
barr , phys .
rev .
d * 48 * ( 1993 ) 5354 .
h. fritzsch , phys . lett . * 73b * ( 1978 ) 317 ; nucl . phys .
* 115b * ( 1979 ) 189 .
g.c .
branco and j.i .
silva - marcos , phys .
lett .
b * 331 * ( 1994 ) 390 .
g. altarelli and f. feruglio , phys .
lett .
b * 439 * ( 1998 ) 112 , hep - ph/9807353 ; g. altarelli and f. feruglio , hep - ph/9812475 ; z. berezhiani and a. rossi , hep - ph/9811447 ; k. hagiwara and n. okamura , hep - ph/9811495 .
j. k. elwood , n. irges , and p. ramond , phys .
rev .
lett .
* 81 * ( 1998 ) 5064 , hep - ph/9807228 and references contained therein .
c.h . albright and s.m .
barr , in preparation .
m. gell - mann , p. ramond , and r. slansky , report no .
calt-68 - 709 : _ supergravity _ ( north holland , amsterdam , 1979 ) ; t. yanagida , in proceedings of the workshop on unified theory and baryon number of the universe , kek , japan , 1979 ( unpublished ) .
l. wolfenstein , phys .
rev .
lett .
* 51 * ( 1983 ) 1945 .
, particle data group , 1998 . |
in the present work , we investigate level - set percolation for the gaussian free field on @xmath0 , @xmath1 . this problem has already received much attention in the past , see for instance @xcite , @xcite , and more recently @xcite , @xcite .
the long - range dependence of the model makes this problem particularly interesting , but also harder to analyze . here
, we prove the existence of a non - trivial critical level for all @xmath1 , and the positivity of this critical level when @xmath12 is large enough .
some of our methods are inspired by the recent progress in the study of the percolative properties of random interlacements , where a similar long - range dependence occurs , see for instance @xcite , @xcite , @xcite , @xcite .
we now describe our results and refer to section [ notation ] for details .
we consider the lattice @xmath0 , @xmath1 , endowed with the usual nearest - neighbor graph structure .
our main object of study is the gaussian free field on @xmath0 , with canonical law @xmath13 on @xmath14 such that , @xmath15 = g ( x , y)$ } , \text { for all } x , y \in \mathbb{z}^d , \end{split}\ ] ] where @xmath16 denotes the green function of simple random walk on @xmath0 , see .
note in particular the presence of strong correlations , see .
for any _ level _ @xmath17 , we introduce the ( random ) subset of @xmath0 @xmath18 sometimes called _ excursion set _
( above level @xmath2 ) .
we are interested in the event that the origin lies in an infinite cluster of @xmath19 , which we denote by @xmath20 , and ask for which values of @xmath2 this event occurs with positive probability . since @xmath21\ ] ] is decreasing in @xmath2 , it is sensible to define the critical point for level - set percolation as @xmath22\ ] ] ( with the convention @xmath23 ) . a non - trivial phase transition is then said to occur if @xmath24 is finite .
it is known that @xmath25 for all @xmath1 and that @xmath26 ( see @xcite , corollary 2 and theorem 3 , respectively ; see also the concluding remark 5.1 in @xcite to understand why the proof does not easily generalize to all @xmath1 ) .
it is also known that when @xmath27 , for large @xmath2 , there is no directed percolation inside @xmath28 , see @xcite , p. 281
( note that this reference studies the percolative properties of the excursion sets of @xmath29 in place of @xmath30 ) .
it is not intuitively obvious why @xmath9 should be finite , for it seems a priori conceivable that infinite clusters of @xmath19 could exist for _ all _ @xmath31 due to the strong nature of the correlations .
we show in corollary [ c2.7 ] that this does not occur and that @xmath32 in fact , we prove a stronger result in theorem [ t2.6 ] . we define a second critical parameter @xmath33 = 0 \big\},\ ] ] where the event @xmath34 refers to the existence of a ( nearest - neighbor ) path in @xmath28 connecting @xmath35 , the ball of radius @xmath36 around @xmath37 in the @xmath38-norm , to @xmath39 , the @xmath38-sphere of radius @xmath40 around @xmath37 .
it is an easy matter ( see corollary [ c2.7 ] below ) to show that @xmath41 now , we prove in theorem [ t2.6 ] the stronger statement @xmath42 and then obtain as a by - product that @xmath43 ( see theorem [ 2.6 ] below for a precise statement ) .
this immediately leads to the important question of whether @xmath9 and @xmath10 actually coincide . in case
they differ , our results imply a marked transition in the decay of the connectivity function of @xmath28 at @xmath44 , see remark [ 2.8 ] below . our second result concerns the critical level @xmath9 in high dimension .
we are able to show in theorem [ t3.3 ] that @xmath45 this is in accordance with recent numerical evidence , see @xcite , chapter 4 .
we actually prove a stronger result than .
namely , we show that one can find a positive level @xmath46 , such that for large @xmath12 , the restriction of @xmath47 to a thick two - dimensional slab percolates , see above .
let us however point out that , by a result of @xcite ( see p. 1151
therein ) , the restriction of @xmath48 to @xmath49 ( viewed as a subset of @xmath0 ) , and , a fortiori , the restriction of @xmath28 to @xmath49 , when @xmath2 is positive , do only contain finite connected components : excursion sets above any non - negative level do not percolate in planes .
we refer to remark [ concrems ] 1 ) for more on this .
we now comment on the proofs .
we begin with .
the key ingredient is a certain ( static ) renormalization scheme very similar to the one developed in section 2 of @xcite for the problem of percolation of the vacant set left by random interlacements ( for a precise definition of this model , see @xcite , section 1 ; we merely note that the two `` corresponding '' quantities are @xmath28 and @xmath50 , the vacant set at level @xmath51 )
. we will be interested in the probability of certain crossing events viewed as functions of @xmath52 , @xmath53\end{aligned}\ ] ] ( see for the precise definition ) , where @xmath54 is a geometrically increasing sequence of length scales , see .
note that by , @xmath55 is decreasing in @xmath2 .
we then explicitly construct an increasing but bounded sequence @xmath56 , with ( finite ) limit @xmath57 , such that @xmath58 this readily implies , since @xmath59 for all @xmath60 , hence @xmath61 vanishes . by separating combinatorial complexity estimates from probabilistic bounds in @xmath62 ,
see and lemma [ l2.1 ] , we are led to investigate the quantity @xmath63 \end{split}\ ] ] ( see for the precise definition ) , where the @xmath64 boxes are indexed by the `` leaves '' of a dyadic tree of depth @xmath65 .
the key to proving is to provide a suitable induction step relating @xmath66 to @xmath67 , for all @xmath60 , where the increase in parameter @xmath68 allows to dominate the interactions ( `` sprinkling '' ) .
this appears in proposition [ p2.2 ] .
one then makes sure that @xmath69 is chosen small enough by picking @xmath46 large , see theorem [ t2.6 ] .
the resulting estimates are fine enough to imply not only that @xmath70 , but even the stretched exponential decay of the connectivity function of @xmath71 , thus yielding .
the proof of then only requires a refinement of this argument .
note that the strategy we have just described is precisely the one used in @xcite for the proof of a similar theorem in the context of random interlacements .
we actually also provide a generalization of proposition [ p2.2 ] , which is of independent interest , but goes beyond what is directly needed here , see proposition 2.2. it has a similar spirit to the main renormalization step leading to the decoupling inequalities for random interlacements in @xcite , see remark [ r2.3 ] .
we now comment on the proof of , which has two main ingredients .
the first ingredient is a suitable decomposition of the field @xmath30 restricted to the subspace @xmath72 into the sum of two independent gaussian fields .
the first field has independent components and the second field only acts as a `` perturbation '' when @xmath12 becomes large , see lemmas [ l3.1 ] and [ l3.2 ] below .
the second ingredient combines the fact that the critical value of bernoulli site percolation on @xmath72 is smaller than @xmath73 ( see @xcite ) , with static renormalization ( see chapter @xmath74 of @xcite , @xcite , @xcite ) , and a peierls - type argument to control the perturbation created by the second field .
this actually enables us to deduce a stronger result than .
namely , we show in theorem [ t3.3 ] that one can find a level @xmath75 and a positive integer @xmath76 , such that for large @xmath12 and all @xmath77 , the excursion set @xmath19 already percolates in the two - dimensional slab @xmath78 as already pointed out , some of our proofs employ strategies similar to those developed in the study of the percolative properties of the vacant set left by random interlacements , see @xcite , @xcite .
this is not a mere coincidence , as we now explain .
continuous - time random interlacements on @xmath0 , @xmath1 , correspond to a certain poisson point process of doubly infinite trajectories modulo time - shift , governed by a probability @xmath79 , with a non - negative parameter @xmath80 playing the role of a multiplicative factor of the intensity measure pertaining to this poisson point process ( the bigger @xmath80 , the more trajectories `` fall '' on @xmath0 ) , see @xcite , @xcite .
this poisson gas of doubly infinite trajectories ( modulo time - shift ) induces a random field of occupation times @xmath81 ( so that the interlacement at level @xmath80 coincides with @xmath82 , whereas the vacant set at level @xmath80 equals @xmath83 ) . this field
is closely linked to the gaussian free field , as the following isomorphism theorem from @xcite shows : @xmath84 it is tempting to use this identity as a transfer mechanism , and we hope to return to this point elsewhere .
we conclude this introduction by describing the organization of this article . in section
[ notation ] , we introduce some notation and review some known results concerning simple random walk on @xmath0 and the gaussian free field . section [ nontrivial pt ] is devoted to proving that excursion sets at a high level do not percolate .
the main results are theorem [ t2.6 ] and corollary [ c2.7 ] .
the positivity of the critical level in high dimension and the percolation of excursion sets at low positive level in large enough two - dimensional slabs is established in theorem [ t3.3 ] of section [ poslarged ] . one final remark concerning our convention regarding constants : we denote by @xmath85 positive constants with values changing from place to place .
numbered constants @xmath86 are defined at the place they first occur within the text and remain fixed from then on until the end of the article . in sections
[ notation ] and [ nontrivial pt ] , constants will implicitly depend on the dimension @xmath12 .
in section [ poslarged ] however , constants will be purely numerical ( and _ independent _ of @xmath12 ) . throughout the entire article , dependence of constants on additional parameters
will appear in the notation .
in this section , we introduce some notation to be used in the sequel , and review some known results concerning both simple random walk and the gaussian free field .
we denote by @xmath87 the set of natural numbers , and by @xmath88 the set of integers .
we write @xmath89 for the set of real numbers , abbreviate @xmath90 and @xmath91 for any two numbers @xmath92 , and denote by @xmath93 $ ] the integer part of @xmath94 , for any @xmath95 .
we consider the lattice @xmath0 , and tacitly assume throughout that @xmath1 . on @xmath0
, we respectively denote by @xmath96 and @xmath97 the euclidean and @xmath38-norms .
moreover , for any @xmath98 and @xmath99 , we let @xmath100 and @xmath101 stand for the the @xmath38-ball and @xmath38-sphere of radius @xmath102 centered at @xmath94 . given @xmath103 and @xmath104 subsets of @xmath0 , @xmath105 stands for the complement of @xmath103 in @xmath0 , @xmath106 for the cardinality of @xmath103 , @xmath107 means that @xmath108 , and @xmath109 denotes the @xmath38-distance between @xmath103 and @xmath104 . if @xmath110 , we simply write @xmath111 . finally , we define the inner boundary of @xmath103 to be the set @xmath112 , and the outer boundary of @xmath103 as @xmath113 .
we also introduce the diameter of any subset @xmath114 , @xmath115 , as its @xmath38-diameter , i.e. @xmath116 .
we endow @xmath0 with the nearest - neighbor graph structure , the edge - set consisting of all pairs of sites @xmath117 , @xmath118 , such that @xmath119 .
a ( nearest - neighbor ) path is any sequence of vertices @xmath120 , where @xmath121 and @xmath122 for all @xmath123 , satisfying @xmath124 for all @xmath125 .
moreover , two lattice sites @xmath126 will be called @xmath127-nearest neighbors if @xmath128 .
a @xmath127-path is defined accordingly .
thus , any site @xmath98 has @xmath129 nearest neighbors and @xmath130 @xmath127-nearest neighbors .
we now introduce the ( discrete - time ) simple random walk on @xmath0 . to this end
, we let @xmath131 be the space of nearest - neighbor @xmath0-valued trajectories defined for non - negative times , and let @xmath132 , @xmath133 , stand for the canonical @xmath134-algebra and canonical process on @xmath131 , respectively . since @xmath1 , the random walk is transient .
furthermore , we write @xmath135 for the canonical law of the walk starting at @xmath98 and @xmath136 for the corresponding expectation . we denote by @xmath16 the green function of the walk , i.e. @xmath137 , \qquad \text{for } x , y \in \mathbb{z}^d,\ ] ] which is finite ( since @xmath138 ) and symmetric .
moreover , @xmath139 due to translation invariance . given @xmath140 , we further denote the entrance time in @xmath141 by @xmath142 , the hitting time of @xmath141 by @xmath143 , and the exit time from @xmath141 by @xmath144 .
this allows us to define the green function @xmath145 killed outside @xmath104 as @xmath146 , \qquad \text{for } x , y \in \mathbb{z}^d.\ ] ] it vanishes if @xmath147 or @xmath148 .
the relation between @xmath149 and @xmath150 for any @xmath140 is the following ( we let @xmath151 ) : @xmath152 , \qquad \text{for } x , y \in \mathbb{z}^d.\ ] ] the proof of is a mere application of the strong markov property ( at time @xmath153 ) .
we now turn to a few aspects of potential theory associated to simple random walk . for any @xmath154 , we write @xmath155 , \qquad x \in k,\ ] ] for the equilibrium measure ( or escape probability ) of @xmath103 , and @xmath156 for its capacity
. it immediately follows from and that the capacity is subadditive , i.e. @xmath157 moreover , the entrance probability in @xmath103 may be expressed in terms of @xmath158 ( see for example @xcite , theorem 25.1 , p. 300 ) as @xmath159 = \sum_{y \in k } g(x , y ) \cdot e_{k}(y),\ ] ] from which , together with classical bounds on the green function ( c.f .
below ) , one easily obtains ( see @xcite , section 1 for a derivation ) the following useful bound for the capacity of a box : @xmath160 we next review some useful asymptotics of @xmath161 . given two functions @xmath162 , we write @xmath163 , as @xmath164 , if they are asymptotic , i.e. if @xmath165 .
[ l1.3 ] @xmath166 @xmath167 = 1 - \frac{7}{2d } + o(d^{-1 } ) , \qquad \text{as } d \to \infty , \label{1.17}\end{aligned}\ ] ] where @xmath168 is viewed as @xmath169 in . for ,
see @xcite , theorem 1.5.4 , for , see @xcite , pp .
246 - 247 . in order to prove , we assume that @xmath170 and define @xmath171 @xmath172 then , under @xmath173 , @xmath174 is a `` lazy '' walk on @xmath175 starting at the origin .
clearly , @xmath176 , where @xmath177 refers to the first return to @xmath37 for the walk @xmath178 . hence , @xmath179 = \big[g^{(y)}(0 ) \big]^{-1 } = \big [ \frac{d}{d-3 } \cdot g^{(d-3)}(0 ) \big]^{-1 } \stackrel{\eqref{1.16}}{= } 1 - \frac{7}{2d } + o(d^{-1}),\end{aligned}\ ] ] as @xmath180 , where @xmath181 denotes the green function of @xmath178 and @xmath182 that of simple random walk on @xmath175 .
we now turn to the gaussian free field on @xmath0 , as defined in . given any subset @xmath114
, we frequently write @xmath183 to denote the family @xmath184 . for arbitrary @xmath185 and @xmath154
, we also use the shorthand @xmath186 for the event @xmath187 and similarly @xmath188 instead of @xmath189 .
next , we introduce certain crossing events for the gaussian free field . to this end , we first consider the space @xmath190 endowed with its canonical @xmath134-algebra and define , for arbitrary disjoint subsets @xmath191 , the event ( subset of @xmath192 ) @xmath193 for any level @xmath52 , we write @xmath194 for the measurable map from @xmath14 into @xmath192 which sends @xmath195 to @xmath196 , and define @xmath197 ( a measurable subset of @xmath14 endowed with its canonical @xmath134-algebra @xmath198 ) , which is the event that @xmath103 and @xmath199 are connected by a ( nearest - neighbor ) path in @xmath28 , c.f . . denoting by @xmath200 the image of @xmath13 under @xmath194 ,
i.e. the law of @xmath201 on @xmath192 , we have that @xmath202 = q^h [ k \longleftrightarrow k'].$ ] note that @xmath203 is an increasing event upon introducing on @xmath14 the natural partial order ( i.e. @xmath204 when @xmath205 for all @xmath98 ) .
we proceed with a classical fact concerning conditional distributions for the gaussian free field on @xmath0 .
we could not find a precise reference in the literature , and include a proof for the reader s convenience .
we first define , for @xmath140 , the law @xmath206 on @xmath14 of the centered gaussian field with covariance @xmath207 = g_{u } ( x , y ) , \qquad \text { for all } x , y \in \mathbb{z}^d,\ ] ] with @xmath145 given by .
in particular , @xmath208 , @xmath206-almost - surely , whenever @xmath209 .
we then have [ l1.2 ] @xmath210 let @xmath211 , @xmath212 and define @xmath213 by @xmath214 where @xmath215 is the @xmath216-measurable map defined as @xmath217 = \sum_{y \in k } p_x [ h_{k } < \infty , x_{h_{k } } = y ] \cdot \varphi_y , \qquad \text{for } x \in \mathbb{z}^d.\ ] ] then , under @xmath13 , @xmath218 note that for all @xmath219 , @xmath220 ( since @xmath221 for @xmath219 , by ) and that for all @xmath219 , @xmath208 , @xmath222-almost surely . hence , it suffices to consider @xmath223 .
we first show independence . by , @xmath224 , @xmath225 , are centered and jointly gaussian .
moreover , they are uncorrelated , since for @xmath226 , @xmath227 , @xmath228 = \operatorname{\mathbb{e}}[\varphi_x \varphi_y ] - \operatorname{\mathbb{e}}[\mu_x \varphi_y ] \stackrel{\eqref{phi},\eqref{mu}}{= } g(x , y ) - \sum_{z \in k}p_x [ h_{k } < \infty , x_{h_{k } } = z ] g(z , y ) \stackrel{\eqref{g - gsubk}}{= } 0.\ ] ] thus , @xmath224 , @xmath225 , are independent . to conclude the proof of lemma [ l1.2 ] , it suffices to show that @xmath229 = \operatorname{\mathbb{e}}^{u } \big [ 1_a \big ( ( \varphi_x)_{x \in u}\big ) \big ] , \qquad \text{for all } a \in \operatorname{\mathcal{f}}_u,\ ] ] where @xmath230 stands for the canonical @xmath134-algebra on @xmath231 .
furthermore , choosing some ordering @xmath232 of @xmath104 , by dynkin s lemma , it suffices to assume that @xmath233 has the form @xmath234 we fix some @xmath233 of the form , and consider a subset @xmath235 such that @xmath236 ( we will soon let @xmath237 increase to @xmath238 ) .
we let @xmath239 , @xmath240 , denote the law of simple random walk on @xmath235 starting at @xmath94 killed when exiting @xmath235 ( its green function corresponds to @xmath241 ) , and define @xmath242 for @xmath240 as in but with @xmath243 replacing @xmath135 in the definition of @xmath215 .
it then follows from proposition 2.3 in @xcite ( an analogue of the present lemma for _ finite _ graphs ) that @xmath244 = \operatorname{\mathbb{e}}^{v\setminus k } \big [ 1_{a^v } \big ( ( \varphi_x)_{x \in v\setminus k}\big ) \big],\ ] ] with @xmath245 , @xmath246 as defined in and @xmath247 . letting @xmath248
, it follows that @xmath249 , @xmath250 , hence by dominated convergence that both sides of converge towards the respective sides of , thus completing the proof .
[ r1.3]@xmath210 lemma [ l1.2 ] yields a choice of regular conditional distributions for @xmath251 conditioned on the variables @xmath184 , which is tailored to our future purposes .
namely , @xmath13-almost surely , @xmath252 = \widetilde{{\mathbb{p } } } \big [ ( \widetilde{\varphi}_x + \mu_x)_{x \in \mathbb{z}^d } \in \cdot \ \big],\ ] ] where @xmath215 , @xmath98 is given by , @xmath253 does not act on @xmath254 , and @xmath255 is a centered gaussian field under @xmath253 , with @xmath220 , @xmath253-almost surely for @xmath219 .
lemma [ l1.2 ] also provides the covariance structure of this field ( namely @xmath256 , with @xmath212 ) , but its precise form will be of no importance in what follows .
note that conditioning on @xmath184 produces the ( random ) _ shift _ @xmath215 , which is _ linear _ in the variables @xmath257 , @xmath227 .
@xmath258 the explicit form of the conditional distributions in readily yields the following result , which can be viewed as a consequence of the fkg - inequality for the free field ( see for example @xcite , chapter 4 ) .
[ fkg]@xmath210 let @xmath259 , @xmath260 , and assume @xmath261 ( the canonical @xmath134-algebra on @xmath14 ) is an increasing event . then @xmath262 \ \leq \ { \mathbb{p}}\big[a \big| \
\varphi_{|_{k } } \geq \alpha \big],\ ] ] where the left - hand side is defined by the version of the conditional expectation in .
intuitively , augmenting the field can only favor the occurrence of @xmath233 , an increasing event . on the event @xmath263
, we have , for @xmath215 , @xmath98 , as defined in , @xmath264 \geq \alpha p_x[h_{k } < \infty ] \stackrel{\text{def.}}{= } m_x(\alpha ) , \qquad \text{for all } x\in \mathbb{z}^d,\ ] ] with equality instead on the event @xmath265 .
since @xmath233 is increasing , this yields , with a slight abuse of notation , @xmath266 \cdot 1_{\ { \varphi_{\vert_{k } } \geq \alpha \ } } \hspace{-1ex } & \stackrel{\eqref{phi_cond_exps}}{= } & \hspace{-1ex } \widetilde { \mathbb{p } } \big [ a \big ( ( \widetilde{\varphi}_x + m_x(\alpha))_{x \in \mathbb{z}^d } \big)\ \big ] \cdot 1_{\ { \varphi_{\vert_{k } } \geq \alpha \ } } \\ & \leq & \hspace{-1ex } \widetilde { \mathbb{p } } \big [ a \big ( ( \widetilde{\varphi}_x + \mu_x)_{x \in \mathbb{z}^d } \big)\ \big ] \cdot 1_{\ { \varphi_{\vert_{k } } \geq \alpha \ } } \\ & \stackrel{\eqref{phi_cond_exps}}{= } & \hspace{-1ex } \mathbb{p } \big[a \big| \varphi_{\vert_{k}}\big ] \cdot 1_{\ { \varphi_{\vert_{k } } \geq \alpha \}}. \end{array}\ ] ] integrating both sides with respect to the probability measure @xmath267 $ ] , we obtain @xmath268 \leq e_\nu \big[\mathbb{p } \big[a | \varphi_{\vert_{k}}\big ] \cdot 1_{\ { \varphi_{\vert_{k } } \geq \alpha \ } } \big ] = { \mathbb{p}}\big[a \big| \
\varphi_{|_{k } } \geq \alpha \big].\ ] ] this completes the proof of lemma [ fkg ] .
we now introduce the canonical shift @xmath269 on @xmath14 , such that @xmath270 , for arbitrary @xmath271 and @xmath272 .
the measure @xmath13 is invariant under @xmath269 , i.e. @xmath273= { \mathbb{p}}[a]$ ] , for all @xmath274 ( the canonical @xmath134-algebra on @xmath14 ) , by translation invariance of @xmath275 ( see below ) , and has the following mixing property : @xmath276 = { \mathbb{p}}[a ] \ , { \mathbb{p}}[b ] , \qquad \text{for all } a , b \in \operatorname{\mathcal{f}}\ ] ] ( one first verifies for @xmath277 depending on finitely many coordinates with the help of and the general case follows by approximation , see @xcite , pp.157 - 158 ) .
the following lemma gives a @xmath37-@xmath278 law for the probability of existence of an infinite cluster in @xmath28 , the excursion set above level @xmath52 , c.f . .
[ perc_prop]@xmath210 let @xmath279 $ ] , for arbitrary @xmath52 .
one then has the following dichotomy : @xmath280 where @xmath281 $ ] . in particular , recalling the definition of @xmath282 , implies that @xmath283 for all @xmath284 , and @xmath285 for all @xmath286 .
this follows by ergodicity , which is itself a consequence of the mixing property .
[ uniqueness]@xmath210 when @xmath283 , in particular in the supercritical regime @xmath284 , the infinite cluster in @xmath19 is @xmath13-almost surely unique .
this follows by theorem 12.2 in @xcite ( burton - keane theorem ) , because the field @xmath201 is translation invariant and has the finite energy property ( see @xcite , definition 12.1 ) .
the main goal of this section is the proof of theorem [ t2.6 ] below , which roughly states that @xmath7 ( and hence @xmath287 , c.f .
corollary [ c2.7 ] ) is finite for all @xmath1 , and that the connectivity function of @xmath28 , c.f . ,
has stretched exponential decay for arbitrary @xmath288 .
the proof involves a certain renormalization scheme akin to the one developed in @xcite and @xcite in the context of random interlacements .
this scheme will be used to derive recursive estimates for the probabilities of certain crossing events , c.f .
proposition [ p2.2 ] , which can subsequently be propagated inductively , c.f .
proposition [ p2.5 ] .
the proper initialization of this induction requires a careful choice of the parameters occurring in the renormalization scheme .
the resulting bounds constitute the main tool for the proof of the central theorem [ t2.6 ] .
in addition , an extension of proposition [ p2.2 ] can be found in remark [ r2.3 ] 2 ) .
we begin by defining on the lattice @xmath0 a sequence of length scales @xmath289 where @xmath290 and @xmath291 are both assumed to be integers and will be specified below .
hence , @xmath76 represents the finest scale and @xmath292 correspond to increasingly coarse scales .
we further introduce renormalized lattices @xmath293 and note that @xmath294 for all @xmath295 . to each @xmath296
, we attach the boxes @xmath297 where we define @xmath298 , the box of side length @xmath36 _ attached to _
@xmath94 , for any @xmath98 and @xmath299 ( not to be confused with @xmath300 ) .
moreover , we let @xmath301 so that @xmath302 defines a partition of @xmath0 into boxes of side length @xmath303 for all @xmath60 , and @xmath304 , @xmath296 , is simply the union of @xmath305 and its @xmath127-neighboring boxes at level @xmath65 . moreover , for @xmath306 and @xmath296 , @xmath305 is the disjoint union of the @xmath307 boxes @xmath308 at level @xmath309 it contains .
we also introduce the indexing sets @xmath310 and given @xmath311 , @xmath306 , we consider the sets of labels @xmath312
\big\ } \neq \emptyset \big\}. \end{split}\ ] ] note that for any two indices @xmath313 , @xmath314 , we have @xmath315 and @xmath316 . finally , given @xmath296 , @xmath60 , we introduce @xmath317 , a family of subsets @xmath318 of @xmath319 ( soon to be thought of as binary trees ) defined as @xmath320 hence , any @xmath321 can naturally be identified as a binary tree having root @xmath311 and depth @xmath65 .
moreover , the following bound on the cardinality of @xmath317 is easily obtained , @xmath322 is a suitable constant .
we now consider the gaussian free field @xmath323 on @xmath0 defined in and introduce the crossing events ( c.f .
for the notation ) @xmath324 three properties of the events @xmath325 will play a crucial role in what follows . denoting by @xmath326 the @xmath134-algebra on @xmath14 generated by the variables @xmath257 ,
@xmath327 , we have @xmath328 indeed , the property that @xmath325 decreases with @xmath2 follows since @xmath329 for all @xmath330 by definition , c.f . .
next , we provide a lemma which separates the combinatorial complexity of the number of crossings in @xmath325 from probabilistic estimates , using @xmath317 as introduced in
. this separation will be key in obtaining estimates fine enough to yield the desired stretched exponential decay .
albeit being completely analogous to lemma 2.1 in @xcite , we repeat its proof , for it comprises an essential geometric observation concerning the events @xmath325 .
[ l2.1 ] @xmath331 , @xmath311 , @xmath332 @xmath333 \leq |\lambda_{n , x } | \sup_{\operatorname{\mathcal{t}}\in \lambda_{n , x } } { \mathbb{p}}[a_{\operatorname{\mathcal{t}}}^h ] , \qquad \text{where } \qquad a_{\operatorname{\mathcal{t}}}^h = \bigcap_{(0,y ) \in \operatorname{\mathcal{t}}\cap \operatorname{\mathcal{i}}_0 } a_{0,y}^h.\ ] ] we use induction on @xmath65 to show that @xmath334 for all @xmath311 , from which immediately follows . when @xmath335 , is trivial .
assume it holds for all @xmath336 .
for any @xmath311 , a path in @xmath337 starting in @xmath305 and ending in @xmath338 must first cross the box @xmath339 for some @xmath340 , and subsequently @xmath341 for some @xmath342 before reaching @xmath338 , c.f .
figure [ fig1 ] below .
thus , @xmath343 upon applying the induction hypothesis to @xmath344 and @xmath345 separately , the claim follows .
[ h ! ] .,title="fig:",width=377 ] before proceeding , we remark that the events @xmath346 , with @xmath52 and @xmath321 for some @xmath311 , @xmath347 , defined in inherit certain properties from the events @xmath348 , @xmath349 .
namely , it follows from and that @xmath350 and that , for any two levels @xmath351 , @xmath352 further , given any @xmath347 , @xmath311 , and @xmath321 , we define the set @xmath353 hence , @xmath354 is the disjoint union of @xmath64 boxes of side length @xmath355 each , and @xmath356 .
it then immediately follows from the definition of @xmath346 in that ( see above for the notation @xmath357 ) @xmath358 finally , upon introducing @xmath359 , \qquad \text{for } ( n , x ) \in \operatorname{\mathcal{i}}_n , \ n\geq 0,\ ] ] which is well - defined ( i.e. independent of @xmath296 ) by translation invariance , we obtain @xmath360 whenever @xmath330 , by . note
also that @xmath361.\ ] ] we now derive the aforementioned `` recursive bounds '' for the probabilities @xmath67 , c.f .
below , along a suitable increasing sequence @xmath362 ( one - step renormalization ) .
these estimates will be key in proving theorem [ t2.6 ] below .
[ p2.2 ] @xmath363 , @xmath364 there exist positive constants @xmath365 and @xmath366 such that , defining @xmath367 then , given any positive sequence @xmath368 satisfying @xmath369 and any increasing , real - valued sequence @xmath362 satisfying @xmath370 one has @xmath371 the main idea of the proof is to `` decouple '' the event @xmath372 , where @xmath373 and @xmath374 are the ( binary ) subtrees at level @xmath65 of some given tree @xmath375 , @xmath376 , using the increase in parameter @xmath68 to dominate the interactions ( `` sprinkling '' ) .
we let @xmath60 , consider some @xmath377 and some tree @xmath378 .
we decompose @xmath379 where @xmath380 , @xmath314 are the two descendants of @xmath381 in @xmath318 and @xmath382 that is @xmath383 is the ( sub-)tree consisting of all descendants of @xmath380 in @xmath318 .
thus , the union in is over disjoint sets .
note in particular that @xmath384 . by construction ,
the subsets @xmath385 @xmath386 , for @xmath314 , see , satisfy @xmath387 . for sake of clarity , and since @xmath381 and @xmath318 will be fixed throughout the proof , we abbreviate @xmath388 in order to estimate the probability of the event @xmath389 , @xmath52 , defined in , we introduce a parameter @xmath390 and write @xmath391 & \leq { \mathbb{p}}\big [ a_{\operatorname{\mathcal{t}}_1}^h \cap a_{\operatorname{\mathcal{t}}_2}^h \cap \big\ { \text{max}_{_{k_1 } } \varphi \leq \alpha \big\ } \big ] \ + \ { \mathbb{p}}\big [ \text{max}_{_{k_1 } } \varphi > \alpha \big ] \nonumber \\ & = \mathbb{e } \big [ 1_{a_{\operatorname{\mathcal{t}}_1}^h } \cdot 1 _ { \ { \text{max}_{_{k_1 } } \varphi \ \leq \alpha \ } } \cdot \mathbb{p } [ a_{\operatorname{\mathcal{t}}_2}^h \ | \varphi_{_{k_1 } } ] \big ] \ +
\ { \mathbb{p}}\big [ \text{max}_{_{k_1 } } \varphi > \alpha \big ] , \end{aligned}\ ] ] where @xmath392 and the second line follows because @xmath393 is measurable with respect to @xmath394 , c.f . .
we begin by focusing on the conditional probability @xmath395 $ ] in . using and applied to @xmath396 , and with a slight abuse of notation , we find @xmath397 = \widetilde { \mathbb{p } } \big [ a_{\operatorname{\mathcal{t}}_2}^h \big ( ( \widetilde{\varphi}_x + \mu_x)_{x \in k_2 } \big)\ \big ] , \qquad { \mathbb{p}}\text{-almost surely},\ ] ] where @xmath398 $ ] . on the event @xmath399
, we have , for all @xmath400 , @xmath401 \leq \alpha \cdot p_x [ h_{k_1 } < \infty ] \stackrel{\text{def.}}{= } m_x ( \alpha),\ ] ] which is deterministic and linear in @xmath402 .
moreover , we can bound @xmath403 as follows . by virtue of , @xmath404 \leq \text{cap}(k_1 ) \cdot \sup_{y \in k_1 } g(x , y)$ ] for all @xmath400 .
since @xmath405 consists of @xmath64 disjoint boxes of side length @xmath355 , c.f . and , its capacity can be bounded , using and , as @xmath406 . by , and the observation that @xmath407 whenever @xmath408 and @xmath409 , it follows that @xmath410 which defines the constant @xmath365 from , and the factor @xmath411 is kept for later convenience .
returning to the conditional probability @xmath395 $ ] , we first observe that , on the event @xmath399 and for any @xmath400 , the inequality @xmath412 implies @xmath413 hence , on the event @xmath399 , @xmath414 & \stackrel{\eqref{p2.2.2}}{= } & \hspace{-1ex } \widetilde { \mathbb{p } } \big [ a_{\operatorname{\mathcal{t}}_2}^h \big ( ( \widetilde{\varphi}_x + \mu_x)_{x \in k_2 } \big)\ \big ] \\[1.2ex ] & \leq & \hspace{-1ex } \widetilde { \mathbb{p } } \big [ a_{\operatorname{\mathcal{t}}_2}^{h - \gamma } \big ( ( \widetilde{\varphi}_x -m_x(\alpha))_{x \in k_2 } \big)\ \big ] = \mathbb{p } [ a_{\operatorname{\mathcal{t}}_2}^{h - \gamma } \
| \varphi_{\vert_{k_1 } } = - \alpha ] , \end{array } \end{split}\ ] ] where the last equality follows by , noting that , on the event @xmath415 , we have @xmath416 for all @xmath400 , c.f . .
applying lemma [ fkg ] to the right - hand side of , we immediately obtain that , on the event @xmath399 , @xmath417 \leq \mathbb{p } [ a_{\operatorname{\mathcal{t}}_2}^{h - \gamma } \ | \varphi_{\vert_{k_1 } } \geq - \alpha ] \leq \mathbb{p } [ a_{\operatorname{\mathcal{t}}_2}^{h - \gamma } ] \cdot \big ( \mathbb{p } [ \varphi_{\vert_{k_1 } } \geq - \alpha ] \big)^{-1}.\ ] ] at last , we insert into , noting that , since @xmath30 has the same law as @xmath418 , we have @xmath419 = 1- { \mathbb{p}}[\text{min}_{_{k_1 } } \varphi < - \alpha ] = 1- { \mathbb{p}}[\text{max}_{_{k_1 } } \varphi > \alpha ] $ ] , to get @xmath420 \leq { \mathbb{p}}\big [ a_{\operatorname{\mathcal{t}}_1}^h \big ] \cdot { \mathbb{p}}\big [ a_{\operatorname{\mathcal{t}}_2}^{h - \gamma } \big ] \cdot \big ( 1- { \mathbb{p}}[\text{max}_{_{k_1 } } \varphi > \alpha ] \big)^{-1 } + { \mathbb{p}}\big [ \text{max}_{_{k_1 } } \varphi > \alpha \big].\ ] ] next , we turn our attention to the term @xmath421 $ ] . by virtue of the btis - inequality ( c.f @xcite , theorem 2.1.1 ) , for arbitrary @xmath211 , we have @xmath422 \leq \exp \bigg\{-\frac{\big ( \alpha - \operatorname{\mathbb{e}}\big [ \text{max}_{_{k } } \varphi \big ] \big)^2}{2 g(0 ) } \bigg\ } , \qquad \text{if } \
\alpha > \operatorname{\mathbb{e}}\big [ \text{max}_{_{k } } \varphi \big].\ ] ] in order to bound @xmath423 $ ] , we write , using fubini s theorem , @xmath424 \leq \operatorname{\mathbb{e}}\big [ \text{max}_{_{k } } \varphi^+ \big ] = \int_0^\infty \text{d}u
\ { \mathbb{p}}\big [ \text{max}_{_{k } } \varphi^+ > u \big ] \leq a + \int_a^\infty \text{d}u \ { \mathbb{p}}\big [ \text{max}_{_{k } } \varphi^+ > u \big],\ ] ] for arbitrary @xmath425 .
recalling that @xmath426= g(0)$ ] for all @xmath98 , c.f . , and introducing an auxiliary variable @xmath427 , we can bound the integrand as @xmath428 \leq |k| \cdot { \mathbb{p}}[\varphi_0 > u ] = |k| \cdot { \mathbb{p}}\big[\psi > g(0)^{-1/2}u \big ] \leq |k| \cdot e^{-u^2/2g(0)},\ ] ] where we have used in the last step that @xmath429 \leq e^{-a^2/2}$ ] , for @xmath430 , which follows readily from markov s inequality , since @xmath429 \leq \min_{\lambda > 0}e^{-\lambda a}\operatorname{\mathbb{e}}[e^{\lambda \psi } ] = \min_{\lambda > 0}e^{-\lambda a + \lambda^2/2}$ ] , and the minimum is attained at @xmath431 . inserting the bound for @xmath432 $ ] into yields , for arbitrary @xmath433 , @xmath434 \leq a + |k| \int_a^\infty \text{d}u
\ e^{-\frac{u^2}{2g(0 ) } } \leq a + c |k| \cdot e^{-a^2/2g(0)}.\ ] ] we select @xmath435 ( so that @xmath436 ) , by which means readily implies that @xmath437 \leq c \sqrt{\log|k| } , \qquad \text{for all } \quad \emptyset \neq k
\subset \subset \mathbb{z}^d.\ ] ] in the relevant case @xmath438 with @xmath439 , we thus obtain @xmath440 \leq c_2 \big ( 2g(0 ) \log(2^n ( 3l_0)^d)\big)^{1/2 } \stackrel{\eqref{2.8.1}}{= } \sqrt{2 g(0 ) } \cdot m(n , l_0),\ ] ] where the first inequality defines the constant @xmath366 from .
we now require @xmath441 thus applies and yields @xmath442 \leq \min \big\ { 1/2 , \ e^ { - \big(\frac{\alpha}{\sqrt{2g(0 ) } } -m(n , l_0)\big)^2 } \big\}.\ ] ] returning to , and using that @xmath443 for all @xmath444 ( with @xmath445 $ ] ) , we finally obtain , for all @xmath402 satisfying and @xmath446 , @xmath447 \ \leq \ { \mathbb{p}}[a_{\operatorname{\mathcal{t}}}^h ] \hspace{-1ex } & \leq & \hspace{-1ex } { \mathbb{p}}\big [ a_{\operatorname{\mathcal{t}}_1}^h \big ] \cdot { \mathbb{p}}\big [ a_{\operatorname{\mathcal{t}}_2}^{h - \gamma } \big ] + 3 \cdot { \mathbb{p}}\big [ \text{max}_{_{k_1 } } \varphi > \alpha \big ] \\ & \stackrel{\eqref{p2.2.12}}{\leq } & \hspace{-1ex } { \mathbb{p}}\big [ a_{\operatorname{\mathcal{t}}_1}^{h- \gamma } \big ] \cdot { \mathbb{p}}\big [ a_{\operatorname{\mathcal{t}}_2}^{h - \gamma } \big ] + 3 e^{-(\beta - m(n , l_0))^2 } , \end{array } \end{split}\ ] ] where we have set @xmath448 . the claim
now readily follows upon taking suprema over all @xmath375 on both sides of , letting @xmath449 , @xmath450 ( @xmath2 was arbitrary ) , @xmath451 , so that requiring @xmath452 , by virtue of , is nothing but . noting condition for @xmath453
, we precisely recover .
this concludes the proof of proposition [ p2.2 ] .
[ r2.3 ] @xmath210 \1 ) the bound , which we have derived using an elementary argument , also follows from a more general ( and stronger ) estimate .
one knows that ( in fact , this holds for a large class of gaussian fields , c.f .
@xcite , theorem 1.3.3 ) , @xmath454 \leq c \int_0^{\frac{1}{2 } \sup_{x , y , \in k } d(x , y ) } \sqrt{\log\big ( n(\varepsilon)\big ) } \ \text{d } \varepsilon,\ ] ] where @xmath455 \big)^{1/2},$ ] @xmath118 , @xmath154 , @xmath456 denotes the smallest number of closed balls of radius @xmath457 in this metric covering @xmath103 , and @xmath458 is a universal constant . clearly , @xmath459 for all @xmath460
. moreover , @xmath461 by virtue of .
inserting this into immediately yields the bound .
\2 ) we mention a generalization of proposition [ p2.2 ] , which is of independent interest , but will not be needed in what follows .
consider integers @xmath462 , @xmath291 , and a collection @xmath463 , @xmath464 , of events in @xmath192 ( @xmath465 , see above ) , such that @xmath466 where @xmath467 , @xmath272 , stand for the canonical coordinates on @xmath192 . given @xmath52 , @xmath60 , @xmath296 , and @xmath321 , we replace @xmath346 in by ( see below for the notation @xmath194 ) @xmath468 and @xmath469 in by @xmath470.\ ] ] one then has the following generalization of proposition [ p2.2 ] : @xmath363 , @xmath291 , @xmath471 assume that @xmath472 that @xmath368 is a positive sequence , @xmath362 a real increasing sequence , such that , hold .
then , @xmath473 if instead , @xmath474 @xmath368 is a positive sequence , @xmath362 a real decreasing sequence , so that holds and holds for @xmath475 , then holds as well .
the arguments employed in the proof of proposition [ p2.2 ] yield the first statement ( with @xmath463 , @xmath464 , increasing events ) . to derive the second statement ( when @xmath463 , @xmath464 , are decreasing events ) , one argues as follows .
one introduces the inversion @xmath476 such that @xmath477 , for all @xmath272 , and the collection of `` flipped '' events @xmath478 , @xmath464 .
one defines @xmath479 as in with @xmath480 , @xmath481 , in place of @xmath482 , @xmath481 .
now observe that @xmath483 has the same law as @xmath251 under @xmath13 , and that for any @xmath52 , @xmath484 has the same law @xmath200 as @xmath485 under @xmath13 . from this , we infer that for all @xmath52 , @xmath296 , and @xmath321 , @xmath486 & = q^h \big [ \bigcap_{(0,y ) \in
\operatorname{\mathcal{t}}\cap \operatorname{\mathcal{i}}_0 } d_y \big ] = { \mathbb{p}}\big [ \big ( 1\ { \varphi_x < -h \ } \big)_{x \in \mathbb{z}^d } \in \bigcap_{(0,y ) \in \operatorname{\mathcal{t}}\cap \operatorname{\mathcal{i}}_0 } d_y \big ] \\ & = q^{-h } \big [ \bigcap_{(0,y ) \in
\operatorname{\mathcal{t}}\cap \operatorname{\mathcal{i}}_0 } \overline{d}_y \big ] = { \mathbb{p}}\big[\ \overline{d}_{\operatorname{\mathcal{t}}}^{-h } \big ] .
\end{split}\ ] ] when holds , the events @xmath487 , @xmath464 , satisfy , and thanks to the identity , the second statement of proposition 2.2 is reduced to the first statement .
this concludes the proof of proposition 2.2. \3 ) there is an analogy between proposition 2.2 and the main renormalization step theorem 2.1 of @xcite , for the decoupling inequalities of random interlacements ( see theorem 2.6 of @xcite ) .
note however , that unlike condition ( 2.7 ) of @xcite ( see also ( 2.70 ) in @xcite ) , and tie in the finest scale @xmath76 to the sequence @xmath362 .
this feature has to do with the role of the cut - off level @xmath402 we introduce in and the remainder term it produces .
@xmath258 we now return to proposition [ p2.2 ] and aim at propagating the estimate inductively . to this end , we first define , for all @xmath60 , @xmath488 where @xmath489 is a certain parameter to be specified below in proposition [ p2.5 ] and later in .
note in particular that condition holds for this choice of @xmath490 . in the next proposition ,
we inductively derive bounds for @xmath67 , @xmath60 , given any sequence @xmath491 satisfying the assumptions of proposition [ p2.2 ] , provided the induction can be initiated , see below .
[ p2.5 ] @xmath210 assume @xmath492 and @xmath493 are such that @xmath494 and let the sequence @xmath362 satisfy with @xmath490 as defined in .
then , @xmath495 we define a sequence @xmath496 inductively by @xmath497 with @xmath498 given by ( the factor following @xmath499 in should be viewed as the @xmath500-th root of the remainder term on the right - hand side of ) .
then , implies that @xmath501 for all @xmath60 .
moreover , as we now see , @xmath502 this is clear for @xmath503 . when @xmath306 , first note that by virtue of , @xmath504 moreover , implies @xmath505 for all @xmath506 , which , inserted into , yields @xmath507 where we have used @xmath501 and @xmath508 for all @xmath95 in the second inequality
hence , holds .
we will now show by induction on @xmath65 that @xmath509 which , together with , implies .
the inequality holds for @xmath335 by assumption , c.f . .
assume now it holds for some @xmath65 .
by proposition [ 2.2 ] , we find @xmath510^{2^{n+1 } } \stackrel{\eqref{p2.5.1}}{= } e^{-k_{n+1}2^{n+1}}.\end{aligned}\ ] ] this concludes the proof of and thus of proposition [ p2.5 ] .
@xmath210 although we will not need this fact in what follows , let us point out that a straightforward adaptation of proposition [ p2.5 ] holds in the context of proposition 2.2. @xmath258 we will now state the main theorem of this section and prove it using proposition [ p2.5 ] . to this end , we select @xmath511 appearing in proposition [ p2.5 ] as follows : @xmath512 moreover , we will solely consider sequences @xmath56 with @xmath513 so that condition is satisfied .
we recall that @xmath498 is given by , which now reads @xmath514 where we have substituted @xmath515 from and @xmath511 from .
note that @xmath516 @xmath517 and @xmath46 are the only parameters which remain to be selected .
we finally proceed to the main [ t2.6]@xmath210 the critical point @xmath7 defined in satisfies @xmath518 moreover , for all @xmath1 and @xmath519 , there exist positive constants @xmath520 , @xmath521 and @xmath522 @xmath523 depending on @xmath12 and @xmath524 such that @xmath525 \leq c(h ) \cdot e^{-c'(h)l^{\rho } } , \qquad \text{for all } l \geq 1.\ ] ] in particular , the connectivity function @xmath526 $ ] of the excursion set above level @xmath2 has stretched exponential decay , i.e. there exists @xmath527 such that @xmath528 \leq c(h ) \cdot e^{-c''(h)|x|^{\rho } } , \qquad \text{for all } x \in \mathbb{z}^d , \ h > h_{**}(d ) , \text { and } d \geq 3.\ ] ] [ c2.7 ] @xmath210 the excursion set @xmath28 above level @xmath2 defined in undergoes a non - trivial percolation phase transition for all @xmath1 , i.e. @xmath529 and @xmath530 = \left \ { \begin{array}{rl } 1 , & \text { if } h < h _ * \\ 0 , & \text { if } h > h_*. \end{array } \right.\ ] ] the lower bound @xmath4 in follows from corollary 2 of @xcite . in order to establish the finiteness in , it suffices to show @xmath531 and to invoke the above theorem [ t2.6 ] . to this end
, we note that by definition ( c.f . ) , @xmath532 , \qquad \text{for all } l \geq 1,\ ] ] and @xmath531 readily follows . as for
, it is an immediate consequence of lemma [ perc_prop ] .
@xmath258 to prove , it suffices to construct an explicit level @xmath533 with @xmath534 such that @xmath535 $ ] decays polynomially in @xmath36 , as @xmath536 .
in fact , we will even show that @xmath535 $ ] has stretched exponential decay .
we begin by observing that the sequence @xmath362 defined in has a finite limit @xmath537 for every choice of @xmath516 @xmath517 and @xmath46 .
indeed , @xmath498 as given by satisfies @xmath538 for all @xmath60 , hence @xmath539 since we assumed @xmath291 .
we set @xmath540 and now show with proposition [ p2.5 ] that there exists @xmath541 sufficiently large such that , defining @xmath542 we have @xmath543 \leq c \cdot e^{-c'l^\rho } , \qquad \text { for all } l \geq 1,\ ] ] for suitable @xmath544 and @xmath545 . to this end , we note that @xmath546 defined in is bounded by @xmath547 \stackrel{\eqref{p2.2.8}}{\leq } \exp \bigg\{-\frac{\big ( h_0 - \operatorname{\mathbb{e}}\big [ \text{max}_{_{\widetilde{b}_{0,x=0 } } } \varphi \big ] \big)^2}{2 g(0 ) } \bigg\ } , \end{aligned}\ ] ] when @xmath548 ( e.g. using to bound @xmath549 $ ] ) .
in particular , since @xmath511 in is completely determined by the choices , we see that @xmath550 for all @xmath548 , i.e. condition holds for sufficiently large @xmath46 . by proposition [ p2.5 ] , setting @xmath551 and @xmath533 as in , we obtain @xmath552 therefore , we find that for all @xmath60 and @xmath296 , @xmath553 \stackrel{\eqref{2.7}}{\leq } |\lambda_{n , x}| \cdot p_n ( \bar{h } ) \stackrel{\eqref{2.6.1 } , \eqref{t2.6.3}}{\leq } \big ( c_0 l_0^{2(d-1 ) } \big)^{2^n } \big ( 2c_0 l_0^{2(d-1 ) } \big)^{-2^n } = 2^{-2^n}.\ ] ] we now set @xmath554 , whence @xmath555 . given @xmath299 , we first assume there exists @xmath60 such that @xmath556
. then , since @xmath557 \leq { \mathbb{p}}\bigg [ \bigcup_{x \in \operatorname{\mathbb{l}}_n : b_{n , x } \cap s(0,l ) \neq \emptyset } \big\ { b_{n , x } \stackrel{\geq \bar{h}}{\longleftrightarrow } \partial^i \widetilde{b}_{n , x } \big\ } \bigg],\ ] ] and the number of sets contributing to the union on the right - hand side is bounded by @xmath558 , readily implies , and by adjusting @xmath559 , will hold for @xmath560 as well .
it follows that @xmath561 , which completes the proof of .
we now turn to the proof of .
let @xmath2 be some level with @xmath562 , and define @xmath563 .
since @xmath564 , we may choose @xmath565 such that @xmath566 = 0 $ ] , which readily implies @xmath567 ( see for the definition of @xmath568 ) . moreover , we let @xmath569 + 1 \big),\ ] ] so that @xmath291 is an integer , as required . from , it is then easy to see , using , that @xmath570 , for all @xmath60 .
hence , the limit @xmath571 of the increasing sequence @xmath362 defined in satisfies @xmath572 thus , and imply that @xmath573 whenever @xmath574 .
moreover , @xmath575 where the last inequality follows by .
we thus select @xmath576 so that both and @xmath573 hold .
since condition is satisfied , proposition [ p2.5 ] yields @xmath577 from which point on one may argue in the same manner as for the proof of to infer ( with @xmath2 in place of @xmath533 ) and subsequently deduce .
in particular , this involves defining @xmath554 , which depends on @xmath2 ( and @xmath12 ) through @xmath517 .
the stretched exponential bound for the connectivity function of @xmath28 is an immediate corollary of , since @xmath526 \leq { \mathbb{p}}\big [ b(0,l ) \stackrel{\geq h}{\longleftrightarrow } s(0,2l ) \big ] \leq c(h ) e^{- c''(h)|x|^\rho}$
] whenever @xmath578 .
@xmath258 [ 2.8 ] @xmath210 \1 ) an important open question is whether @xmath9 equals @xmath10 or not . in case
the two differ , the decay of @xmath579 $ ] as @xmath536 , for @xmath580 , exhibits a sharp transition . indeed
, first note that by , for all @xmath580 , @xmath581 \longrightarrow 0 $ ] , as @xmath582 .
if @xmath583 , then by definition of @xmath10 , @xmath584 = \infty.\ ] ] hence @xmath581 $ ] decays to zero with @xmath36 , but with an at most polynomial decay for @xmath585 . however , for @xmath288 , @xmath581 $ ] has a stretched exponential decay in @xmath36 , by .
\2 ) the proof of theorem [ t2.6 ] works just as well if we replace the assumption @xmath8 by @xmath586 , where @xmath587 is defined similarly as @xmath10 in , simply replacing the `` lim '' by a `` liminf '' in , i.e. @xmath588 = 0 \big\}.\ ] ] hence @xmath589 $ ] has stretched exponential decay in @xmath36 when @xmath590 , and one has in fact the equality @xmath591 @xmath258
the main goal of this section is the proof of theorem [ t3.3 ] below , which roughly states that in high dimension , for small but _ positive _ @xmath2 , the excursion set @xmath28 contains an infinite cluster with probability @xmath278 .
we will prove the stronger statement that percolation already occurs in a two - dimensional slab @xmath592 for sufficiently large @xmath76 , see below .
the proof essentially relies on two main ingredients .
the first ingredient is a suitable decomposition , for large @xmath12 , of the free field @xmath30 restricted to @xmath593 ( viewed as a subset of @xmath0 ) , into the sum of two independent gaussian fields @xmath594 and @xmath595 ( c.f . and below for their precise definition ) .
the field @xmath594 is i.i.d . and the dominant part , while @xmath595 only acts as a `` perturbation . ''
the key step towards this decomposition appears in lemma [ l3.1 ] .
the second ingredient is a peierls - type argument , which comprises several steps : first , the sublattice @xmath168 is partitioned into blocks of side length @xmath76 , which are declared `` good '' if certain events defined separately for @xmath595 and @xmath594 occur simultaneously . roughly speaking ,
these events are chosen in a way that suitable excursion sets of the dominant field @xmath594 percolate well and the perturbative part @xmath595 does nt spoil this percolation ( see , and for precise definitions ) .
moreover , @xmath127-connected components of bad blocks are shown to have small probability , see lemma [ peierlsbound ] .
this ensures that the usual method of peierls contours is applicable , which in turn allows the conclusion that an infinite cluster of @xmath28 exists within the above - mentioned slab with positive probability ( and with probability @xmath278 by ergodicity ) .
we note that @xmath595 does nt have finite - range dependence , which renders impractical the use of certain well - known stochastic domination theorems ( see for example @xcite , @xcite ) . one word on notation : in what follows , we identify @xmath596 , @xmath597 , with the set of points @xmath598 satisfying @xmath599 .
we recall that in this section , constants are numerical unless dependence on additional parameters is explicitly indicated .
moreover , we shall assume throughout this section that @xmath600 we also recall that @xmath275 , c.f . , stands for the green function on @xmath0 . without further ado
, we begin with [ l3.1 ] @xmath601covariance decomposition@xmath471 let @xmath602 .
there exists a function @xmath603 on @xmath604 such that @xmath605 where @xmath606 , @xmath607 as @xmath180 , @xmath608 denotes the kronecker symbol , and @xmath603 is the kernel of a translation invariant , bounded , positive operator @xmath609 on @xmath610 , which is the operator of convolution with @xmath611 .
its spectral radius @xmath612 satisfies @xmath613 the operator @xmath614 , for @xmath219 and @xmath615 , is a convolution operator , which is bounded and self - adjoint on @xmath616 , by , , as well as the translation invariance and the symmetry of @xmath16 ( letting @xmath617 , we also use that @xmath618 , a special case of young s inequality , see @xcite , pp .
28 - 29 ) . moreover , by @xcite , p25.2 ( b ) , p. 292
, it has an inverse @xmath619 where @xmath620 is the bounded self - adjoint operator on @xmath621 , @xmath622 , for @xmath219 and @xmath615 , with kernel @xmath623 = \pi ( 0,y - x ) , \qquad \text{for } x , y \in k.\ ] ] introducing @xmath624 \in ( 0,1 ) \qquad \text{(recall \eqref{3.0})},\ ] ] we can write @xmath625 where @xmath626 is the bounded self - adjoint operator on @xmath621 defined by @xmath627 , for @xmath219 , @xmath615 , and @xmath628 = \sum_{y \neq x } \pi(x , y ) , & \text{if } y = x . \end{array } \right.\ ] ] note that by , , @xmath629 and @xmath626 is a convolution operator on @xmath621 . by young s inequality ( see above ) , its operator norm @xmath630 thus satisfies @xmath631 \stackrel{\eqref{kappa}}{= } 2(1-\kappa).\ ] ] observe also that @xmath626 is a positive operator .
indeed , @xmath632 , for all @xmath615 , where @xmath633 denotes the inner product in @xmath621 . by and , we can write , for arbitrary , @xmath634 , @xmath635 \\ & = \kappa^{-1 } \big[(1-a)i + t_a \big ] , \end{split}\ ] ] where @xmath636 is the bounded operator on @xmath621 , @xmath637 which is self - adjoint by , since @xmath233 is .
we will now select @xmath634 in such a way that the operator @xmath636 is positive . by self - adjointness , we know that the spectrum @xmath638 of @xmath639 satisfies @xmath640 \subseteq \mathbb{r},\ ] ] where @xmath641 and @xmath642 . by the spectral theorem , and using that the function @xmath643 is increasing in @xmath95 , it follows that @xmath644 selecting @xmath645 , we see that @xmath646 is a positive operator .
moreover , the application of the spectral theorem and the positivity of @xmath626 also show that @xmath647 .
if we now define @xmath648 ( by ) , and @xmath649 , so that @xmath650 for @xmath219 and @xmath615 , then readily yields .
in addition , @xmath609 is translation invariant , and by , we see that @xmath651 tends to @xmath278 as @xmath180 , and the spectral radius of @xmath609 satisfies @xmath652 for a suitable constant @xmath653 .
this concludes the proof of lemma [ l3.1 ] .
we now decompose the gaussian free field according to lemma [ l3.1 ] . to this end , we let @xmath654 , @xmath655 , be probabilities on auxiliary probability spaces @xmath656 , @xmath657 , respectively endowed with random fields @xmath658 , @xmath659 , such that @xmath660 = \sigma^2 ( d ) \cdot \delta(x , y)$ , for all $ x , y \in \mathbb{z}^3 $ } , \end{split}\ ] ] and @xmath661 = g'(x , y)$ , for all $ x , y \in \mathbb{z}^3$}. \end{split}\ ] ] then , by and usual gaussian field arguments ( see @xcite , p.11 ) , @xmath662 moreover , given any level @xmath52 , we define the ( random ) sets @xmath663 and @xmath664 , @xmath665 in analogous manner .
a crucial point is that the field @xmath595 acts only as a small perturbation when @xmath12 is large , which is entailed in and more quantitatively in the following lemma , the proof of which uses ideas developed in @xcite ( see in particular theorem 2.4 therein ) .
[ l3.2 ] @xmath210 there exists a decreasing function @xmath666 \longrightarrow ( 0 , 1)$ ] , with @xmath667 , such that for all @xmath31 and @xmath170 satisfying @xmath668 , and all @xmath669 , @xmath670 \leq \big [ v ( h^2 d)\big]^{|a|}.\ ] ] first note that @xmath671 \leq { \mathbb{p}}_\xi \big [ \sum_{x \in a } \xi_x^2 > h^2 |a| \big ] , \qquad \text{for all } h > 0.\ ] ] now , assume some ordering of @xmath669 has been specified , and let @xmath672 denote the covariance matrix of the gaussian vector @xmath673 , with decreasing eigenvalues @xmath674 , @xmath675 , and write @xmath676 for its spectral radius .
finally , define the diagonal matrix @xmath677 . by spectral decomposition , @xmath678 for some orthogonal matrix @xmath679 .
let @xmath680 be a gaussian vector whose components are i.i.d .
standard gaussian variables , and @xmath681 be its law
. then @xmath682 , i.e. @xmath683 , and thus @xmath684 inserting this into yields @xmath685 \leq { \mathbb{p}}_{\tilde{\xi } } \big [ \sum_{1 \leq i \leq |a| } \tilde{\xi}_i^2 > \rho_a^{-1 } h^2 |a| \big ] \leq \min_{0 < a < 1 } e^{-\frac{a h^2 |a|}{2 \rho_a } } \mathbb{e}_{\tilde{\xi } } \big [ \prod_{1 \leq i \leq |a| } e^ { \frac{a}{2 } { \tilde{\xi}_i}^2 } \big],\ ] ] where we have used markov s inequality in the last step . but @xmath686 = ( 1-a)^{-1/2}$ ] for all @xmath687 and @xmath675 , thus yielding @xmath688
\leq \min_{0 < a < 1 } \big [ \sqrt{1-a } \cdot e^{\frac{a h^2 } { 2 \rho_a } } \big]^{-|a|}.\ ] ] one easily verifies that the function @xmath689 attains a maximum in the interval @xmath690 only if @xmath691 , which certainly holds if @xmath668 by lemma [ l3.1 ] . in this regime , the maximum is reached for @xmath692 . inserting this into , we obtain @xmath685 \leq \big [ \big(e h^2\rho_a^{-1}\big)^{1/2 } \cdot e^{- \frac { h^2}{2 \rho_a } } \big]^{|a|}.\ ] ] the function @xmath693 is @xmath690-valued and monotonically decreasing on @xmath694 .
thus , @xmath668 ensures that @xmath695 and @xmath696 .
this completes the proof of lemma [ l3.2 ] .
we are now ready to introduce a central quantity before proceeding to the main theorem of this section .
namely , we define , for all @xmath170 , @xmath52 and positive integers @xmath76 , @xmath697.\ ] ] [ t3.3 ] @xmath210 there exists @xmath698 , a level @xmath541 and an integer @xmath699 such that @xmath700 in particular , the critical level @xmath3 defined in satisfies @xmath701 to begin with , we note that immediately follows from . in order to prove , we will use the decomposition of the gaussian free field restricted to @xmath168 and the bounds obtained in lemma [ l3.2 ] to perform a peierls - type argument .
we let @xmath702 , for @xmath703 , and observe that the law of @xmath704 on @xmath705 ( endowed with its canonical @xmath134-algebra ) is @xmath706 , where @xmath707 so that @xmath708 $ ] ( see ) . for arbitrary @xmath709 , @xmath462 , @xmath170 and @xmath710
, we define the increasing event ( part of @xmath705 ) , @xmath711 and obtain , for all @xmath712 , using the decomposition , @xmath713 \\ & = \int_{\omega_\xi } \text{d } { \mathbb{p}}_\xi(\omega_\xi ) \operatorname{\mathbf{p}}_{p(2h , \sigma(d ) ) } \big[a_\infty^h ( \omega_\xi ) \big ] \\ & \geq \int_{\omega_\xi } \text{d } { \mathbb{p}}_\xi(\omega_\xi ) \operatorname{\mathbf{p}}_{p ' } \big[a_\infty^h ( \omega_\xi ) \big ] , \end{split}\ ] ] where we have used that @xmath714 is increasing in @xmath715 for every fixed @xmath716 in the last line . since @xmath717 for @xmath170
, we have @xmath718 for all @xmath719 and @xmath170 .
thus , yields @xmath720,\ ] ] for all @xmath170 , @xmath709 and @xmath462 , where @xmath721 is a field of independent centered gaussian variables with variance @xmath722 , i.e. as in but with @xmath722 in place of @xmath651 , and @xmath723 denotes the probability on @xmath724 governing @xmath725 .
we will prove that the probability on the right - hand - side of is equal to one for all @xmath726 , @xmath727 , with suitable @xmath728 , @xmath46 and @xmath76 .
the claim will immediately follow from this .
we first construct certain families of `` good '' events for the gaussian fields @xmath725 and @xmath595 . to this end , we define boxes @xmath729 for any @xmath730 and positive integer @xmath36 , and introduce a renormalized lattice @xmath731 we begin with @xmath725 and define @xmath732 , for any @xmath733 and @xmath734 , to be the ( possibly empty ) open ( i.e. value @xmath278 ) cluster in @xmath735 containing the most vertices .
if several such clusters exist , we choose one according to some given prescription ( say using lexicographic order ) . for arbitrary @xmath733
, we introduce the event @xmath736 ( in the canonical @xmath134-algebra of @xmath737 ) as follows : given some @xmath738 , we let @xmath739 if and only if 1 .
@xmath732 is a _ crossing _
cluster for @xmath735 in the first two axes - directions , i.e. for @xmath740 there exists an open path @xmath741 in @xmath732 with endvertices @xmath742 satisfying @xmath743 and @xmath744 .
@xmath732 is the _ only _ open cluster @xmath745 of @xmath735 having the property @xmath746 .
having introduced @xmath736 for all @xmath60 and @xmath733 , we define , for @xmath747 , the measurable map @xmath748 @xmath749 , and the events ( in @xmath750 ) @xmath751 we now turn to the ( `` good '' ) events for @xmath595 , set @xmath752 and note that @xmath753 = { \mathbb{p}}_\xi [ g_{0}^{-h } ] $ ] , for all @xmath733 , by translation invariance of @xmath754 , see . with `` good '' events for @xmath725 and @xmath595 at hand ( see and ) , for any @xmath31 , we define a vertex @xmath733 to be @xmath2-_good _ if the event @xmath755 occurs ( under @xmath756 ) , and @xmath2-_bad _ otherwise .
a sequence @xmath757 , @xmath758 , in @xmath759 such that @xmath760 for all @xmath761 will be called a _
nearest - neighbor path in @xmath759 _ ( we will refer to @xmath381 as the length of the path ) . similarly , a _ @xmath127-nearest - neighbor path in @xmath759 _ is any sequence in @xmath759 subject to the weaker condition @xmath762 for all @xmath763 .
a crucial property is that percolation of @xmath2-good sites in @xmath759 implies percolation of @xmath764 ( in the slab @xmath765 ) , which we state as a [ l3.6 ] @xmath210 let @xmath31 , @xmath757 , @xmath758 , be a nearest - neighbor path in @xmath759 and assume that all vertices @xmath766 , @xmath767 are @xmath2-good . then , the corresponding clusters @xmath768 @xmath601pertaining to the events @xmath769 are subsets of @xmath770 , which are all connected within the set @xmath771 .
in particular , lemma [ l3.6 ] implies that if @xmath232 is an infinite nearest - neighbor path of @xmath2-good vertices in @xmath759 which is unbounded , then the set @xmath772 ( a subset of @xmath765 ) contains an infinite cluster .
it suffices to consider the case @xmath773 .
the general case then follows by induction on @xmath381 .
thus , let @xmath774 , @xmath775 , be both @xmath2-good .
the following holds for @xmath314 : by definition of @xmath776 , the set @xmath777 contains a cluster @xmath778 which is crossing in the first two axes - directions .
moreover , since @xmath779 occurs , @xmath780 for all @xmath781 , i.e. @xmath782 .
it remains to show the clusters @xmath783 and @xmath784 are connected within @xmath785 .
let @xmath786 be such that @xmath787 . since @xmath783 is crossing for @xmath788 in the @xmath789-th direction , @xmath790 .
hence @xmath783 and @xmath784 are connected within @xmath791 by condition ii ) in the above definition of the events @xmath736 .
this concludes the proof of lemma [ l3.6 ] .
@xmath258 we now carry on with the proof of theorem [ t3.3 ] , and select the parameters @xmath46 , @xmath76 and @xmath728 . first note that @xmath792 , the critical level for bernoulli site percolation on @xmath168 , satisfies @xmath793 ( see @xcite , theorem 4.1 ) .
we may thus choose @xmath75 such that @xmath794 \stackrel{\eqref{p_h_sigma}}{=}p\big(2 h_0 , \frac{1}{\sqrt{2}}\big ) = \frac{1}{2 } \big(\frac{1}{2 } + p_c^{\text{site}}(\operatorname{\mathbb{z}^3 } ) \big),\ ] ] which means that the bernoulli site percolation model on @xmath168 associated to choosing sites @xmath730 where @xmath795 , is supercritical .
by the site - percolation version of theorem 7.61 in @xcite , we thus obtain that @xmath796 = 1 , \qquad \text{for all } x \in \operatorname{\mathscr{l}}.\ ] ] moreover , the collection @xmath797 is @xmath798-dependent and , ( see @xcite , theorem 7.65 , or @xcite , theorem @xmath799 ) , there exists a non - decreasing function @xmath800 \longrightarrow [ 0,1]$ ] with @xmath801 such that if @xmath802 \geq \delta$ ] , then @xmath803 using , we then choose @xmath76 the smallest positive integer such that @xmath804 \big ) \geq 1 -1/40.\ ] ] having fixed @xmath46 and @xmath76 , we choose , in the notation of lemma [ l3.2 ] , a constant @xmath805 such that @xmath806 ( recall that @xmath807 is monotonically decreasing ) and define @xmath808 + 1,\ ] ] so that and yield @xmath809 we now proceed to the last step of the proof , which mainly encompasses a peierls argument . to this end , we introduce , for @xmath733 and @xmath810 a multiple of @xmath76 , the event @xmath811 that @xmath94 is connected to @xmath812 , the restriction to @xmath759 of the @xmath38-sphere of radius @xmath810 centered around @xmath94 , by a @xmath127-path of @xmath46-bad vertices in @xmath759 . in the following lemma , we show that this event has small probability .
[ peierlsbound ] @xmath210 for all @xmath726 , @xmath306 , and @xmath733 , @xmath813 \leq 2^{-n}.\ ] ] for @xmath733 and @xmath306 , we denote by @xmath814 the set of self - avoiding @xmath127-paths in @xmath759 starting at @xmath94 of length @xmath65 . for @xmath815 to occur , there must be a self - avoiding @xmath127-path @xmath816 of @xmath46-bad vertices in @xmath759 starting at @xmath94 , hence @xmath817 & \leq{\mathbb{p}}_{\psi^0 } \otimes { \mathbb{p}}_{\xi } \big [ \bigcup_{\gamma \in\gamma_{x , n}^ * } \ { \gamma \text { is $ h_0$-bad}\}\big ] \\ & \leq |\gamma_{x , n}^*| \sup_{\gamma = ( x_i)_{0\leq i \leq n } \in \gamma_{x , n}^ * } { \mathbb{p}}_{\psi^0 } \otimes { \mathbb{p}}_{\xi } \big [ \bigcap_{i=1}^n \big ( ( f_{x_i}^{2h_0})^c \cup ( g_{x_i}^{-h_0})^c \big ) \big ] , \end{split}\ ] ] where we have identified @xmath818 with @xmath819 in the last line ( and similarly for @xmath820 ) . for arbitrary @xmath821 ,
the probability on the right - hand side of is equal to ( setting @xmath822 = \{1,\dots , n \}$ ] ) @xmath823 } \ \bigcap_{j \in [ [ n ] ] \setminus \{i_1 , \dots , i_k \ } } ( f_{x_j}^{2h_0})^c \ \bigcap_{i\in \{i_1 , \dots , i_k \ } } ( g_{x_i}^{-h_0})^c \big ] \\ & \leq \sum_{k=0}^n \binom{n}{k } \sup_{\{i_1 , \dots , i_k \ } \subset [ [ n ] ] } { \mathbb{p}}_{\psi^0 } \big [ \bigcap_{j \in [ [ n ] ] \setminus \{i_1 , \dots , i_k \ } } ( f_{x_j}^{2h_0})^c \big ] \cdot { \mathbb{p}}_{\xi } \big [ \bigcap_{i\in \{i_1 , \dots , i_k \ } } ( g_{x_i}^{-h_0})^c \big ] .
\end{split}\ ] ] by stochastic domination and our choice of @xmath76 , c.f . and
, we have @xmath824 \setminus \{i_1 , \dots , i_k \ } } ( f_{x_j}^{2h_0})^c \big ] \leq 40^{-(n - k)}.\ ] ] when @xmath825 , @xmath826 , simultaneously occur , we can choose @xmath789 sites @xmath827 , @xmath828 , in the respective boxes @xmath829 , @xmath828 , such that @xmath830 ( c.f .
definition ) .
any such @xmath827 belongs to exactly four boxes @xmath831 , @xmath733 . since @xmath832 is self - avoiding in @xmath759 , we thus have @xmath833 . as a result ,
lemma [ l3.2 ] yields @xmath834 \leq ( 2l_0)^{3k } \cdot \big ( v(h_0 ^ 2d_0 ) \big)^{\frac{k}{4 } } \stackrel{\eqref{3.probad0}}{\leq } 40^{-k}.\ ] ] note that @xmath835 .
putting , and together , and substituting the resulting bound into , we finally obtain , @xmath836 \leq 8^n \sum_{k=0}^n \binom{n}{k } 40^{-k}40^{-(n - k)}= ( 2 / 5 ) ^n < 2^{-n}.\ ] ] this completes the proof of lemma [ peierlsbound ] .
@xmath258 we can now conclude the proof of theorem [ t3.3 ] . for arbitrary @xmath726 , we consider the event that the set @xmath837 is surrounded by a @xmath127-circuit ( i.e. a self - avoiding @xmath127-path except for the end point which coincides with the starting point ) of @xmath46-bad vertices in @xmath759 . considering a point of this circuit on the first axis with largest
coordinate , we see that the probability of this event is bounded by @xmath838 \stackrel{\eqref{3.peierls}}{\leq } \sum_{n=2}^\infty 2^{-n } < 1.\ ] ] if this event does not occur , then by planar duality ( c.f .
@xcite , section 11.2 ) , there exists an infinite self - avoiding nearest - neighbor path @xmath839 of @xmath46-good vertices in @xmath759 .
lemma [ l3.6 ] ( see in particular the remark following it ) then implies that the set @xmath840 contains an infinite cluster . by
, this event happens with positive probability , so that @xmath841 > 0 , \
\text{for all } d \geq d_0.\ ] ] it then follows by and ergodicity ( see lemma [ perc_prop ] ) that the value of @xmath842 is in fact one .
this proves and thus concludes the proof of theorem [ t3.3 ] .
@xmath210 [ concrems ] \1 ) we show in theorem [ t3.3 ] that @xmath843 percolates in a two - dimensional slab for small but positive @xmath2 when @xmath12 is sufficiently large .
however , it should be underlined that @xmath844 does not percolate for any @xmath845 , as we now briefly explain .
indeed , the conditions of theorem 14.3 in @xcite ( which is itself a variant of @xcite when the finite energy condition holds ) are met for the law of @xmath846 on @xmath847 under @xmath13 ( in particular , positive correlations , see definition 14.1 in @xcite , follow from the fkg - inequality ) . hence @xmath848 and its complement in @xmath49 can not _ both _ have infinite clusters . observe that @xmath846 and @xmath849 have the same law under @xmath13 , by symmetry . if @xmath848 percolated ( with probability one , by ergodicity )
, the same would hold true as well for @xmath850 , leading to a contradiction .
therefore , @xmath848 does not percolate .
\2 ) with lemma [ l3.2 ] at hand , one may immediately apply the criterion of molchanov and stepanov ( c.f .
@xcite , theorem 2.1 ) to infer that @xmath851 ( c.f . for notation ) percolates _ strongly _ in @xmath168 when @xmath852 is sufficiently large , i.e. not only does @xmath851 percolate , but in addition @xmath853 does nt , @xmath854-almost surely . however , we note that @xmath855 does not percolate strongly , since for all @xmath31 , we have @xmath856 < 1/2 $ ] , hence in particular @xmath857 , where both @xmath855 and its complement possess an infinite cluster with positive probability ( in fact with probability one ) .
\3 ) it remains open whether @xmath3 is actually strictly positive for _ all _ @xmath1 .
recent simulations suggest this is the case when @xmath858 , with an approximate value @xmath859 \simeq 0.16 $ ] , see @xcite , section 4.1.2 , and figure 4.1 in appendix 4.4 .
pierre - franois rodriguez would like to thank balzs rth and artm sapozhnikov for presenting their recent work @xcite , as well as david belius and alexander drewitz for stimulating discussions .
r. j. adler and j. e. taylor .
random fields and geometry . _ springer _ , new york ( 2007 ) .
j. bricmont , j. l. lebowitz and c. maes .
percolation in strongly correlated systems : the massless gaussian field .
_ j. stat .
_ , * 48*(5/6 ) : pp .
1249 - 1268 ( 1987 ) . h. cramr and m. r. leadbetter .
stationary and related stochastic processes . _ wiley _ , new york ( 1967 ) .
m. campanino and l. russo .
an upper bound on the critical percolation probability for the three - dimensional cubic lattice .
_ , * 13*(2 ) : pp .
478 - 491 ( 1985 ) .
percolation transition for some excursion sets .
_ , * 9*(paper no .
10 ) : pp .
255 - 292 ( 2004 ) .
georgii , o. hggstrm and c. maes . the random geometry of equilibrium phases .
_ phase transitions and critical phenomena _
( edited by c. domb and j. lebowitz ) , * 14 * : pp.1 - 142 . academic press ( 2001 ) .
a. gandolfi , m. keane and l. russo . on the uniqueness of the infinite occupied cluster in dependent two - dimensional site percolation .
prob . _ * 16*(3 ) : pp .
1147 - 1157 ( 1988 )
. g. r. grimmett .
_ springer _ , berlin , 2nd ed .
g. r. grimmett and j. m. marstrand .
the supercritical phase of percolation is well - behaved .
( london ) _ , series a , * 430 * : pp .
439 - 457 ( 1990 ) .
o. hggstrm and j. jonasson .
uniqueness and non - uniqueness in percolation theory .
surv . _ * 3 * : pp .
289 - 344 ( 2006 ) .
intersections of random walks .
_ birkhuser _ , basel ( 1991 ) . j. l. lebowitz and h. saleur .
percolation in strongly correlated systems .
a _ , * 138 * : pp .
194 - 205 ( 1986 ) .
t. m. liggett . interacting particle systems .
_ springer _ , new york ( 1985 ) .
t. m. liggett , r. h. schonmann and a. m. stacey .
domination by product measures .
prob . _ * 25*(1 ) : pp .
71 - 95 ( 1997 ) .
v. marinov .
percolation in correlated systems .
_ phd thesis , rutgers university .
_ available online at : http://www.books.google.com/books?isbn=0549701338 ( 2007 ) .
e. w. montroll .
random walks in multidimensional spaces , especially on periodic lattices .
_ , * 4*(4 ) : pp .
241 - 260 ( 1956 ) .
s. a. molchanov and a. k. stepanov .
percolation in random fields i. _ teoret . mat .
_ , * 55*(2 ) : pp .
246 - 256 ( 1983 ) .
a. pisztora .
surface order large deviations for ising , potts , and percolation models .
_ probab . theory related fields _ , * 104 * , pp . 427 - 466 ( 1996 ) .
b. rth and a. sapozhnikov .
the effect of small quenched noise on connectivity properties of random interlacements .
available at arxiv:1109.5086 .
m. reed and b. simon .
methods of modern mathematical physics , vol .
ii : fourier - analysis , self - adjointness .
_ academic press _
, new york ( 1975 ) .
v. sidoravicius and a.s .
connectivity bounds for the vacant set of random interlacements .
henri poincar , probabilits et statistiques _ , * 46*(4 ) : pp .
976 - 990 ( 2010 ) .
f. spitzer .
principles of random walk . _
, new york , 2nd ed .
a.s . sznitman .
vacant set of random interlacements and percolation .
_ * 171 * : pp .
2039 - 2087 ( 2010 ) .
a.s . sznitman .
decoupling inequalities and interlacement percolation on @xmath860 .
_ inventiones mathematicae _ , * 187 * : pp .
645 - 706 ( 2012 ) .
a.s . sznitman .
an isomorphism theorem for random interlacements . _ electron .
_ * 17*(9 ) : pp.1 - 9 ( 2012 ) .
a.s . sznitman .
topics in occupation times and gaussian free fields .
_ zurich lecture notes in advanced mathematics _ , ems , zurich ( 2012 ) .
a. teixeira . on the size of a finite vacant cluster of random interlacements with small intensity .
theory relat .
fields _ , * 150*(3 - 4 ) : pp . 529 - 574 ( 2011 ) . | we consider level - set percolation for the gaussian free field on @xmath0 , @xmath1 , and prove that , as @xmath2 varies , there is a non - trivial percolation phase transition of the excursion set above level @xmath2 for all dimensions @xmath1 .
so far , it was known that the corresponding critical level @xmath3 satisfies @xmath4 for all @xmath1 and that @xmath5 is finite , see @xcite .
we prove here that @xmath3 is finite for all @xmath1 .
in fact , we introduce a second critical parameter @xmath6 , show that @xmath7 is finite for all @xmath1 , and that the connectivity function of the excursion set above level @xmath2 has stretched exponential decay for all @xmath8 . finally , we prove that @xmath9 is strictly positive in high dimension .
it remains open whether @xmath9 and @xmath10 actually coincide and whether @xmath11 for all @xmath1 .
pierre - franois rodriguez and alain - sol sznitman |
the hot and dense matter expected to be formed in the partonic phase after ultra - relativistic heavy ion collisions ( urhic ) dynamically evolve in space and time due to high internal pressure .
the system cools due to expansion and reverts back to hadronic matter from the partonic phase .
it is well known that the average magnitude of radial flow at the freeze - out surface can be extracted from the transverse momentum ( @xmath2 ) spectra of the hadrons .
however , hadrons being strongly interacting objects can bring the information of the state of the system when it is too dilute to support collectivity .
the electromagnetic ( em ) probes , _
i.e. _ photons and dileptons on the other hand are produced and emitted @xcite ( see @xcite for review ) from each space time points .
therefore , estimating radial flow from the em probes will shed light on the time evolution of the collectivity in the system .
the generation of collectivity in the system depends on eos - hence this can be used to differentiate partonic and hadronic phases as the eos for these two phases are different . in case of em
probes- dilepton has the advantage over the real photons . because the low @xmath2 photons from the hadronic phase receive large transverse kick due to radial flow and consequently appear in the high @xmath2 domain to mingle with those from the qgp phase , making the detection of photons from qgp difficult .
however , for dileptons there are two kinematic variables available - the @xmath2 and the invariant mass ( @xmath0 ) .
while the @xmath2 spectra of dilepton is affected by the flow , the @xmath2 integrated @xmath0 spectra remains unaltered .
this suggests that a careful selection of @xmath2 and @xmath0 windows will be very useful to characterize the qgp and hadronic phases . in the present work
we will demonstrate how the development of radial flow can be estimated through the hanbury - brown twiss ( hbt ) interferometry with virtual photons ( lepton pairs ) for different @xmath0 windows .
the radial flow velocity ( @xmath1 ) can be estimated by considering the ratio of the @xmath2 distribution of single photon to lepton pairs for various @xmath0 windows .
we will briefly discuss the results here and refer to @xcite for details .
single electrons originating from the semi - leptonic decays of heavy mesons carry the information on the interaction of the heavy quarks ( a constituent of the heavy mesons ) with the thermal medium of light quarks and gluons produced in heavy ion collisions . the @xmath4 and @xmath3 can be used to quantify the interaction of the heavy quarks with the qgp .
several ingredients like inclusions of non - perturbative contributions from the quasi - hadronic bound state @xcite , 3-body scattering effects @xcite , the dissociation of heavy mesons due to its interaction with the partons in the thermal medium @xcite and employment of running coupling constants and realistic debye mass @xcite , the inclusion of both elastic and inelastic collisions along with the path length fluctuation have been proposed @xcite to improve the description of the experimental data . within the framework of fokker planck equation ( fpe )
we will evaluate @xmath3 and @xmath4 for these electrons . in the present paper
we discuss the elliptic flow of the matter probed by the single electron from the heavy mesons decays . for the elliptic flow of the matter probed by single photon and lepton pair we refer to @xcite for details . in the next section
we will briefly describe the hbt interferometry with virtual photons .
the azimuthal anisotropy of the system probed by single electrons originating from the heavy flavour decays will be discussed in section 3 .
section 4 is devoted to summary and discussions .
the interferometry of the dilepton pairs actually reflect correlation between two virtual photons , the analysis then can proceed by computing the bose - einstein correlation ( bec ) function for two virtual photons which can be defined as , @xmath5 $ ] , where @xmath6 is momentum of the individual lepton pair , @xmath7 and @xmath8 represent the one- and two- particle inclusive lepton pair spectra respectively , which can be evaluated form the source function for various invariant mass windows of the pair @xcite . for the productions of lepton pairs from qgp the annihilation of thermal quarks and from the hadronic phase
the decays of thermal light vector mesons ( @xmath9 , @xmath10 and @xmath11 ) have been considered . for the space time evolution of the system relativistic hydrodynamical model with cylindrical symmetry @xcite and boost invariance along the longitudinal direction
@xcite has been used . for a system undergoing isentropic expansion ,
the initial temperature ( @xmath12 ) and proper thermalization time ( @xmath13 ) of the system may be constrained by the measured hadronic multiplicity , @xmath14 . for relativistic heavy ion collider ( rhic )
we have taken @xmath15 mev and @xmath16 fm / c .
the eos which controls the rate of expansion / cooling has been taken from the lattice qcd calculations @xcite . the chemical ( @xmath17=170 mev ) and kinetic ( @xmath18=120 mev ) freeze - out temperatures are fixed by the particle ratios and the slope of the @xmath2 spectra of hadrons @xcite . with all these ingredients the correlation function @xmath19 has been evaluated for different ( average ) invariant mass windows , @xmath20 as a function of @xmath21 and @xmath22 @xcite which are related to the transverse momentum of individual pair .
the hbt radii , @xmath23 and @xmath24 corresponding to @xmath25 and @xmath26 extracted from the ( gaussian ) parametrization of @xmath19 . and @xmath27 as a function of @xmath28 .
the dashed ( solidi with asterisk ) line indicates hbt radii for qgp ( total = qgp+hadron ) phase .
right panel : variation of the radial velocity with @xmath28 extracted from the ratio of the @xmath2 distribution of photons to lepton pairs ( see @xcite for details ) .
, title="fig : " ] and @xmath27 as a function of @xmath28 .
the dashed ( solidi with asterisk ) line indicates hbt radii for qgp ( total = qgp+hadron ) phase .
right panel : variation of the radial velocity with @xmath28 extracted from the ratio of the @xmath2 distribution of photons to lepton pairs ( see @xcite for details ) .
, title="fig : " ] the @xmath23 is related to the transverse size of the system whereas the @xmath24 measures both the transverse size and duration of particle emission ( @xcite for review ) .
the @xmath23 shows non - monotonic dependence on @xmath29 ( fig .
[ fig1 ] , left panel ) .
it can be shown that @xmath30 .
the high @xmath29 regions are dominated by the early partonic phase @xcite where the collective flow has not been developed fully consequently the ratio of collective ( @xmath31 ) to thermal ( @xmath32 ) energies is small- hence a larger @xmath23 is obtained for large @xmath0 .
in contrast , the lepton pairs with @xmath33 are emitted from the late hadronic phase where the collective flow or @xmath34 is large , which is reflected as a dip in @xmath23 for @xmath35 .
thus the variation of @xmath23 with @xmath0 can be used as an efficient tool to measure the collectivity in various phases of matter .
we observe that by keeping the @xmath9 and @xmath10 contributions and setting radial velocity , @xmath36 , the dip in @xmath23 vanishes , confirming the fact that the dip is caused by the large radial flow of the hadronic matter .
the @xmath24 probes both the transverse dimension as well as the duration of emission and unlike @xmath23 , @xmath24 does not remain constant even in the absence of radial flow .
the large @xmath29 regions are populated by lepton pairs from early partonic phase where the effect of flow is small and the duration of emission is also small - resulting in smaller values of @xmath24 .
for lepton pair from @xmath35 region the flow is large which could have resulted in a dip as in @xmath23 in this @xmath0 region .
however , @xmath24 probes the duration of emission too , which is large for hadronic phase .
the larger duration overwhelms the reduction of @xmath24 due to flow in the hadronic phase resulting in a bump in @xmath24 in this region of @xmath29 . as mentioned before
the @xmath1 can be estimated from the ratio of the @xmath2 spectra of real photons to lepton pairs .
[ fig1 ] ( right panel ) shows the variation of @xmath1 with @xmath29 both for sps and rhic conditions .
the individual spectra of photons and lepton pairs are constrained by the available experimental data @xcite .
the @xmath1 increases with m up to @xmath37 then drops . from the invariant mass spectra
it is known that the low @xmath0 ( below @xmath9 mass ) and high m ( above @xmath11 peak ) pairs originate from a partonic source @xcite .
the collectivity ( or flow ) does not develop fully in the qgp resulting in smaller values of @xmath1 at both low and high @xmath0 regions .
lepton pairs for @xmath33 originate from the late hadronic source which are largely affected by the flow resulting in higher values of @xmath1 . in summary ,
the value of @xmath1 for @xmath0 below and above the @xmath9-peak is small but around the @xmath9 peak is large - the resulting behaviour is displayed in fig .
[ fig1 ] ( right panel ) .
similar non - monotonic variation of the effective slope parameter of the @xmath2 distribution of lepton pairs for various @xmath0 windows is observed in @xcite .
the evolution of @xmath1 as observed in fig .
[ fig1 ] ( right panel ) is responsible for such behaviour .
the heavy flavors , namely , charm and bottom quarks , play a crucial role in characterizing the qgp ( see also @xcite ) .
as the relaxation time is larger for heavy quarks than light partons , the light quarks and the gluons thermalize faster .
therefore , the propagation of heavy quarks through qgp may be treated as the interactions between equilibrium and non - equilibrium degrees of freedom and the fpe provides an appropriate framework @xcite for such studies . in this work
we would like to evaluate @xmath3 and @xmath4 of heavy flavours within the framework of fpe and contrast the results with the available experimental data .
the evolution of heavy quarks momentum distribution function , while propagating through the qgp are assumed to be governed by the fpe , which reads , @xmath38 \label{expeq}\end{aligned}\ ] ] where the kernels @xmath39 and @xmath40 are given by , @xmath41 , for @xmath42 , @xmath43 and @xmath44 , where @xmath45 and @xmath46 stand for drag and diffusion co - efficients respectively .
the basic inputs required for solving the fp equation are the dissipative co - efficients and initial momentum distributions of the heavy quarks .
the ( effective ) drag and diffusion coefficients have been evaluated by taking in to account both the collisional and radiative processes @xcite . in the radiative process the dead cone and lpm effects
are included . in evaluating the drag
co - efficient we have used temperature dependent strong coupling , @xmath47 @xcite . the debye mass , @xmath48 also a temperature dependent quantity used as a cut - off to shield the infrared divergences arising due to the exchange of massless gluons .
the initial momentum distribution of heavy quarks in pp collisions have been taken from the nlo mnr @xcite code .
the solution of the fpe for the heavy quarks is convoluted with the fragmentation functions of the heavy quarks to obtain the @xmath2 distribution of the @xmath46 and @xmath49 mesons . for heavy - quark fragmentation function
, the peterson function has been used .
the solution of the fpe has been used to predict the @xmath2 spectra of the @xmath46 mesons by following the procedure similar to blast wave method @xcite , the result is compared with experimental data @xcite ( fig . [ fig2]:left panel ) which indicate that the present data can not distinguish between the equilibrium and non - equilibrium scenario .
the @xmath2 distribution of the electrons from the semi - leptonic decays of @xmath46 and @xmath49 meson are evaluated using the standard techniques available in the literature .
the ratio of the @xmath2 distribution of the electron from the decays of heavy flavours produced in heavy ion collisions to the corresponding ( appropriately scaled by the number of collisions ) quantities from the pp collisions is defined as : @xmath50 $ ] , which will be unity in the absence of re - scattering .
the star @xcite and the phenix @xcite collaborations have measured the @xmath51 for non - photonic single electron as a function of @xmath2 for au+au at @xmath52 gev .
the experimental data from both the collaborations show @xmath53 for @xmath54 gev indicating substantial interaction of the heavy quarks with the plasma particles . the spectra evaluated using the formalism described above
reproduces the data reasonably well ( fig .
[ fig2 ] , right panel ) .
distribution of @xmath46 mesons .
@xmath55 indicates the value of @xmath1 which appears as a parameter in the blast wave model .
fp stands for the results obtained from the solution of fp equation .
the experimental data from star collaboration @xcite is compared with the theoretical results .
right panel : variation of @xmath4 with @xmath2 . the initial temperature and thermalization time are taken as 400 mev and 0.2 fm / c respectively .
, title="fig : " ] distribution of @xmath46 mesons .
@xmath55 indicates the value of @xmath1 which appears as a parameter in the blast wave model .
fp stands for the results obtained from the solution of fp equation .
the experimental data from star collaboration @xcite is compared with the theoretical results .
right panel : variation of @xmath4 with @xmath2 . the initial temperature and thermalization time are taken as 400 mev and 0.2 fm / c respectively .
, title="fig : " ] next we discuss the elliptic flow resulting from non - central collisions of nuclei . when a heavy quark propagates along the major axis of an ellipsoidal domain of qgp ( resulting from the non - central collisions ) then the number of interactions it encounters or in other words the amount of energy it dissipates or the amount of momentum degradation that takes place is different from when it propagates along the minor axis
. therefore , the momentum distribution of electrons originating from the decays of heavy flavoured hadrons produced from the fragmentation of heavy quarks propagating through an anisotropic domain will reflect such anisotropy .
the degree of momentum anisotropy will depend on both the spatial anisotropy and more importantly on the coupling strength of the interactions between the heavy quarks and the qgp .
the drag and diffusion coefficients depend on the temperature of the background medium ( qgp ) which evolves in space and time due to expansion .
therefore , the drag and diffusion will also change due to the flow of the background .
the flow of the background has been treated within the ambit of ( 2 + 1 ) dimensional hydrodynamics @xcite .
the coefficient of elliptic flow , @xmath3 is defined as : @xmath56 $ ] .
we evaluate @xmath3 in the current formalism @xcite and compare the results with experimental data @xcite ( fig .
[ fig3 ] , left panel ) .
with @xmath2 for for rhic energy .
right panel : @xmath3 is plotted as a function of @xmath2 for lhc energy for 0 - 10% centrality .
the values of @xmath57 and @xmath58 are taken as 700 mev and 0.08 fm / c respectively .
, title="fig : " ] with @xmath2 for for rhic energy .
right panel : @xmath3 is plotted as a function of @xmath2 for lhc energy for 0 - 10% centrality .
the values of @xmath57 and @xmath58 are taken as 700 mev and 0.08 fm / c respectively .
, title="fig : " ] the prediction for the elliptic flow of the heavy quarks to be measured at large hadron collider ( lhc ) energy through the semi - leptonic decays is depicted in fig .
[ fig3 ] ( right panel ) .
the value of @xmath3 at lhc is similar to that of at rhic .
the prediction for the @xmath59 at lhc has been displayed in fig .
[ fig4 ] separately for @xmath46 and @xmath49 mesons .
the sensitivity of the results on the equation of state ( eos ) , i.e. on the velocity of sound is also considered .
lowering of @xmath60 gives more suppressions as observed in fig .
lower value of velocity of sound , @xmath60 makes the expansion of the plasma slower enabling the propagating heavy quarks to spend more time to interact in the medium and hence lose more energy before exiting from the plasma resulting in more suppression . with @xmath2 for lhc for 0 - 10% centrality . left panel for charm and right panel bottom quarks .
the values of @xmath57 and @xmath58 are taken as 700 mev and 0.1 fm / c respectively .
, title="fig : " ] with @xmath2 for lhc for 0 - 10% centrality .
left panel for charm and right panel bottom quarks .
the values of @xmath57 and @xmath58 are taken as 700 mev and 0.1 fm / c respectively .
, title="fig : " ]
we have shown that the variation of various hbt radii with invariant mass extracted from the correlation functions of two lepton pairs can be used to understand the evolution of collectivity in ultra - relativistic heavy ion collision .
the evolution of the radial flow in the system produced in urhic has been discussed and demonstrated that the non - monotonic variation of @xmath1 with @xmath0 can be used as a signal for parton to hadron transition .
the elliptic flow of the matter probed by the single electrons originating from the heavy flavour decays has been studied . the elliptic flow and the nuclear suppression factor measured at rhic
have been reproduced and predictions for lhc have been given including both the radiative and the collisional processes of energy loss in evaluating the effective drag and diffusion coefficients .
the sensitivity of the results on the eos has also been studied .
* acknowledgment : * the author is grateful to santosh k das , sabyasachi ghosh , bedangadas mohanty , payal mohanty , jajati k nayak and sourav sarkar for collaboration and to tetsufumi hirano for providing hadronic chemical potentials .
this work is partially supported by dae - brns sanction no .
2005/21/5-brns/2455 .
l. d. mclerran and t. toimela , phys . rev .
d * 31 * , 545 ( 1985 ) .
r. rapp and j. wambach , adv .
. phys . * 25 * , 1 ( 2000 ) ; j. alam _
et al . _ , phys .
rep . * 273 * , 243 ( 1996 ) ; j. alam _
et al . _ ,
phys . * 286 * , 159 ( 2001 ) .
b. svetitsky , phys .
d * 37 * , 2484 ( 1988 ) ; j. alam _
et al . _ ,
* 73 * , 1895 ( 1994 ) ; h. van hees , r. rapp , phys .
c,*71 * , 034907 ( 2005 ) ; g. d. moore and d. teaney , phys .
c * 71 * , 064904 ( 2005 ) . | it has been shown that the evolution of collectivity in ultra - relativistic heavy ion collision is manifested in the variation of various hbt radii with invariant mass ( @xmath0 ) extracted from the correlation functions of two lepton pairs .
the value of the radial velocity ( @xmath1 ) can be estimated from the ratio of the @xmath2 distributions of single photons to lepton pairs for various @xmath0 windows .
it has been argued that the variation of radial flow with appropriate kinematic variables can be used as an indicator of a phase transition from initially produced partons to hadrons .
we also consider the elliptic flow ( @xmath3 ) of the matter as probed by the single electron spectra originating from the semileptonic decays of heavy mesons .
the measured values of @xmath3 and the nuclear suppression factor ( @xmath4 ) at rhic energy have been reproduced simultaneously by including both the collisional and radiative processes within the scope of perturbative quantum chromodynamics .
the @xmath4 and @xmath3 have been predicted for lhc energy .
heavy ion collision , quark gluon plasma , photons , dileptons.25.75.-q,25.75.dw,24.85.+p |
let @xmath13 be @xmath0 open , bounded and simply connected sets in @xmath5 with lipschitz boundaries , containing the origin .
we assume that their sizes and lipschitz constants are uniformly bounded .
we set @xmath14 to be the small bodies characterized by the parameter @xmath15 and the locations @xmath16 , @xmath17 .
assume that the lam coefficients @xmath2 and @xmath3 are constants satisfying @xmath18 and the mass density @xmath10 to be a constant that we normalize to a unity .
let @xmath19 be a solution of the navier equation @xmath20 , @xmath21 .
we denote by @xmath22 the elastic field scattered by the @xmath0 small bodies @xmath23 due to the incident field @xmath19 . we restrict ourselves to the scattering by rigid bodies .
hence the total field @xmath24 satisfies the following exterior dirichlet problem of the elastic waves @xmath25 @xmath26 with the kupradze radiation conditions ( k.r.c ) @xmath27 where the two limits are uniform in all the directions @xmath28 .
also , we denote @xmath29 to be the longitudinal ( or the pressure or p ) part of the field @xmath22 and @xmath30 to be the transversal ( or the shear or s ) part of the field @xmath22 corresponding to the helmholtz decomposition @xmath31 .
the constants @xmath32 and @xmath33 are known as the longitudinal and transversal wavenumbers , @xmath34 and @xmath35 are the corresponding phase velocities , respectively and @xmath36 is the frequency .
the scattering problem ( [ elaimpoenetrable]-[radiationcela ] ) is well posed in the hlder or sobolev spaces , see @xcite for instance , and the scattered field @xmath37 has the following asymptotic expansion : @xmath38 uniformly in all directions @xmath39 . the longitudinal part of the far - field , i.e. @xmath40 is normal to @xmath41 while the transversal part @xmath42 is tangential to @xmath41 .
we set @xmath43 . as usual
, we use plane incident waves of the form @xmath44 , where @xmath45 is any direction in @xmath41 perpendicular to the incident direction @xmath46 , @xmath47 are arbitrary constants .
the functions @xmath48 and @xmath49 for @xmath50 are called the p - part and the s - part of the far - field pattern respectively .
[ def1 ] we define 1 .
@xmath51,$ ] 2 .
@xmath52 @xmath53 .
we assume that @xmath54 and @xmath55 is given .
3 . @xmath56 as the upper bound of the used frequencies , i.e. @xmath57 $ ] .
4 . @xmath4 to be a bounded domain in @xmath5 containing the small bodies @xmath58 .
we assume that @xmath59 , with the same diameter @xmath60 , are non - flat lipschitz obstacles , i.e. @xmath61 s are lipschitz obstacles and there exist constants @xmath62 $ ] such that @xmath63 where @xmath64 are assumed to be uniformly bounded from below by a positive constant . in @xcite
, we have shown that there exist two positive constants @xmath65 and @xmath66 depending only on the size of @xmath4 , the lipschitz character of @xmath67 , @xmath55 and @xmath56 such that if @xmath68 then we have the following asymptotic expansion for the p - part , @xmath69 , and the s - part , @xmath70 , of the far - field pattern : @xmath71+m(m-1)\left[\frac{a^3}{d^{2\alpha}}+\frac{a^4}{d^{4-\alpha}}+ \frac{a^4}{d^{5 - 2\alpha}}\right]+m(m-1)^2\frac{a^4}{d^{3\alpha}}\right ) \right ] , \label{x oustdie1 d_m farmainp - near}\\ & \hspace{-1cm}u^\infty_s(\hat{x},\theta)= & \frac{1}{4\pi\,c_s^{2}}(\rm \textbf{i}- \hat{x}\otimes\hat{x})\left[\sum_{m=1}^{m}e^{-i\frac{\omega}{c_s}\hat{x}\cdot\,z_m}q_m\right.\nonumber\\ & & \hspace{-2cm}\left.+o\left(m\left[a^2+\frac{a^3}{d^{5 - 3\alpha}}+\frac { a^4}{d^{9 - 6\alpha}}\right]+m(m-1)\left[\frac{a^3}{d^{2\alpha}}+\frac{a^4}{d^{4-\alpha}}+ \frac{a^4}{d^{5 - 2\alpha}}\right]+m(m-1)^2\frac{a^4}{d^{3\alpha}}\right ) \right ] , \label{x oustdie1 d_m farmains - near } \end{aligned}\ ] ] where @xmath72 , @xmath73 , is a parameter describing the relative distribution of the small bodies .
the vector coefficients @xmath74 , @xmath75 are the solutions of the following linear algebraic system @xmath76 for @xmath77 with @xmath78 denoting the kupradze matrix of the fundamental solution to the navier equation with frequency @xmath36 , @xmath79 and @xmath80 is the solution matrix of the integral equation of the first kind @xmath81 with @xmath82 the identity matrix of order 3 .
consider now the special case @xmath83 and @xmath84 with @xmath85 , @xmath86 , @xmath87 and @xmath88 are positive .
then the asymptotic expansions ( [ x oustdie1 d_m farmainp - near]-[x oustdie1 d_m farmains - near ] ) can be rewritten as @xmath89,\quad \label{x oustdie1 d_m farmainp - near*}\\ u^\infty_s(\hat{x},\theta)&= & \frac{1}{4\pi\,c_s^{2}}(\rm
\textbf{i}- \hat{x}\otimes\hat{x})\big[\sum_{m=1}^{m}e^{-i\frac{\omega}{c_s}\hat{x}\cdot\,z_m}q_m \nonumber \\ & & + o\left(a^{2-s}+a^{3-s-5t+3t\alpha}+a^{4-s-9t+6t\alpha}+a^{3 - 2s-2t\alpha}+a^{4 - 3s-3t\alpha}+a^{4 - 2s-5t+2t\alpha}\right ) \big].\quad \label{x oustdie1 d_m farmains - near * } \end{aligned}\ ] ] as the diameter @xmath60 tends to zero the error term tends to zero for @xmath8 and @xmath90 such that @xmath91 in @xcite , we have shown that @xmath92 , then we have the upper bound @xmath93 hence if the number of obstacles is @xmath94 and @xmath8 satisfies ( [ general - condition - s - t ] ) , @xmath95 , then from ( [ x oustdie1 d_m farmainp - near * ] , [ x oustdie1 d_m farmains - near * ] ) , we deduce that @xmath96 this means that this collection of obstacles has no effect on the homogeneous medium as @xmath9 .
let us consider the case when @xmath97 .
we set @xmath4 to be a bounded domain , say of unit volume , containing the obstacles @xmath98 . given a positive and continuous function @xmath99 , we divide @xmath4 into @xmath100 $ ] subdomains @xmath101 $ ] , each of volume @xmath102}{k(z_m)+1}$ ] , with @xmath103 as its center and contains @xmath104 $ ] obstacles , see fig [ distribution - obstacles ] .
we set @xmath105 , hence @xmath106}_{j=1}[k(z_m)+1]\leq k_{max}[a^{-1}]=o(a^{-1})$ ] .
[ distribution - obstacles ] [ equivalent - medimu ] let the small obstacles be distributed in a bounded domain @xmath4 , say of unit volume , with their number @xmath6 and their minimum distance @xmath107 , @xmath108 , as @xmath95 , as described above . 1 .
if the obstacles are distributed arbitrarily in @xmath4 , i.e. with different capacitances , then there exists a potential @xmath109 with support in @xmath4 such that @xmath110 where @xmath111 is the farfield corresponding to the scattering problem + @xmath112 + @xmath113 + with the radiation conditions @xmath114 2 . if in addition @xmath115 is in @xmath116 , @xmath117 $ ] and the obstacles have the same capacitances @xmath118 , then @xmath119 where @xmath120 in @xmath4 and @xmath121 in @xmath122 .
assume that the lam coefficients @xmath2 and @xmath3 are constants satisfying @xmath18 and the mass density @xmath10 to be a measurable and bounded function which is equal to a constant that we normalize to a unity outside of a bounded domain @xmath4 .
we set @xmath123 to be the upper bound of @xmath10 . in this case
, the total field @xmath124 satisfies the following exterior dirichlet problem of the elastic waves @xmath125 @xmath126 with the kupradze radiation conditions ( k.r.c ) @xmath127 where the two limits are uniform in all the directions @xmath28 and @xmath128 and @xmath129 are respectively the p - part and s - part of the scattered field @xmath130 the scattering problem ( [ elaimpoenetrable - rho]-[radiationcela - rho ] ) is well posed in the hlder or sobolev spaces , see @xcite for instance , and the scattered field @xmath37 has the following asymptotic expansion : @xmath131 uniformly in all directions @xmath39 .
the longitudinal part of the far - field , i.e. @xmath132 is normal to @xmath41 while the transversal part @xmath42 is tangential to @xmath41 .
we set @xmath133 . as in the case of constant background mass density ,
there exist two positive constants @xmath65 and @xmath66 depending only on the size of @xmath4 , the lipschitz character of @xmath67 , @xmath55 , @xmath56 , @xmath123 and @xmath123 such that if @xmath134 then we have the following asymptotic expansion for the p - part , @xmath135 , and the s - part , @xmath136 , of the far - field pattern : @xmath137+m(m-1)\left[\frac{a^3}{d^{2\alpha}}+\frac{a^4}{d^{4-\alpha}}+ \frac{a^4}{d^{5 - 2\alpha}}\right]+m(m-1)^2\frac{a^4}{d^{3\alpha}}\right ) \right ] , \label{x oustdie1 d_m farmainp - near - rho}\\ & \hspace{-1cm}u^\infty_{\rho , s}(\hat{x},\theta)= & v^\infty_{\rho , s}(\hat{x},\theta)\;+~ \left[\sum_{m=1}^{m}g_{\rho , s}^{\infty}(\hat{x } , z_m ) q_{\rho , m}\right.\nonumber\\ & & \hspace{-2cm}\left.+o\left(m\left[a^2+\frac{a^3}{d^{5 - 3\alpha}}+\frac { a^4}{d^{9 - 6\alpha}}\right]+m(m-1)\left[\frac{a^3}{d^{2\alpha}}+\frac{a^4}{d^{4-\alpha}}+ \frac{a^4}{d^{5 - 2\alpha}}\right]+m(m-1)^2\frac{a^4}{d^{3\alpha}}\right ) \right ] , \label{x oustdie1 d_m farmains - near - rho } \end{aligned}\ ] ] where @xmath138 and @xmath139 are the p - part and s - part of the farfields of the green s function @xmath140 , of the operator @xmath141 in the whole space @xmath5 , evaluated in the direction @xmath142 and the source point @xmath143 .
the vector coefficients @xmath144 , @xmath75 are the solutions of the following linear algebraic system @xmath145 for @xmath77 with @xmath146 is the total field satisfying @xmath147 and the scattered field @xmath148 the kupradze radiation conditions ( k.r.c ) .
[ equivalent - medimu - rho ] let the small obstacles be distributed in a bounded domain @xmath4 , say of unit volume , with their number @xmath6 and their minimum distance @xmath107 , @xmath108 , as @xmath95 , as described above . 1 .
if the obstacles are distributed arbitrarily in @xmath4 , i.e. with different capacitances , then there exists a potential @xmath109 with support in @xmath4 such that @xmath149 where @xmath150 is the farfield corresponding to the scattering problem + @xmath151 + @xmath152 + with the radiation conditions .
2 . if in addition @xmath115 is in @xmath116 , @xmath117 $ ] and the obstacles have the same capacitances , then @xmath153 where @xmath120 in @xmath4 and @xmath121 in @xmath122 .
the main contribution of this work is to have shown that by removing from a bounded region of an elastic background , modeled by constant lam coefficients @xmath2 and @xmath3 and a possibly variable density @xmath10 , a number @xmath154 small and rigid bodies of radius @xmath60 distant from each other of at least @xmath155 , @xmath156 , then the perforated medium behaves , as @xmath9 , as a new elastic medium modeled by the same lam coefficients @xmath2 and @xmath3 but with a mass density - like @xmath157 .
the coefficient @xmath158 models the local distribution ( or the local number ) of the bodies while the coefficient @xmath159 , coming from the capacitance of the bodies , describes the geometry of the small bodies as well as their elastic directional diffusion properties ( i.e. the anisotropy character ) .
in addition , we provide explicit error estimates between the far - fields corresponding to the perforated medium and the equivalent one . from this result
we can make the following conclusions : 1 .
assume that the removed bodies have spherical shapes .
for these shapes the corresponding elastic capacitance @xmath118 is of the form @xmath160 ( i.e. a scalar multiplied by the identity matrix ) . in section [ structure - capacitance ] , we describe a more general set of shapes satisfying this property .
hence the equivalent mass density @xmath161 is isotropic while for general shapes it might be anisotropic . to achieve anisotropic densities , a possible choice of the shapes might be an ellipse .
2 . if we choose the local number of bodies @xmath158 large enough or the shapes of the reference bodies , @xmath162 , @xmath163 , having a large capacitance ( i.e. a relative large radius ) so that @xmath164 , then we design elastic materials having negative mass densities .
3 . assume that the background medium is modeled by variable mass density @xmath165 in @xmath4 .
if we remove small bodies from @xmath4 with appropriate @xmath158 and/ or capacitance @xmath159 so that @xmath166 , then the new elastic material will behave every where in @xmath5 as the background medium . hence the new material will not scatter the sent incident waves , i.e. the region @xmath4 modeled by @xmath10 will be cloaked .
the equivalent behavior between a collection of , appropriately dense , small holes and an extended penetrable obstacle modeled by an additive potential was already observed by cioranescu and murat @xcite and also the references therein , where the coefficient @xmath158 is reduced to zero since locally they have only one hole .
their analysis is based on the homogenization theory for which they assume that the obstacles are distributed periodically , see also @xcite and @xcite . in the results presented here
, we do not need such periodicity and no homogenization is used . instead
, the analysis is based on the invertibility properties of the algebraic system ( [ fracqcfracmain ] ) and the precise treatment of the summation in the dominant terms of ( [ x oustdie1 d_m farmainp - near])-([x oustdie1 d_m farmains - near ] ) .
this analysis was already tested for the acoustic model in @xcite .
compared to @xcite , here , in addition to the difficulties coming from the vector character of the lam system , we improved the order of the error estimate , i.e. @xmath167 instead of @xmath168 which , for @xmath169 for instance , reduce to @xmath170 and @xmath171 respectively .
let us finally mention that a result similar to ( [ b - rho ] ) , for the acoustic model , is also derived by ramm in several of his papers , see for instance @xcite , but without error estimates .
compared to his results , and as we said earlier in addition to the vector character of lame model , we provide the approximation by improved explicit error estimates without any other assumptions while , as shown in @xcite for instance , in addition to some formal arguments , he needs extra assumptions on the distribution of the obstacles .
the rest of the paper is organized as follows . in section 2 , we give the detailed proof of theorem [ equivalent - medimu ] .
in section 3 , we describe the one of corollary [ equivalent - medimu - rho ] by discussing the main changes one needs to make in the proof of theorem [ equivalent - medimu ] .
finally , in section 4 , we discuss some invariant properties of the elastic capacitance to characterize the shapes that have a scalar capacitance .
the kupradze matrix @xmath172 of the fundamental solution to the navier equation is given by @xmath173,\end{aligned}\ ] ] where @xmath174 denotes the free space fundamental solution of the helmholtz equation @xmath175 in @xmath5 .
the asymptotic behavior of kupradze tensor at infinity is given as follows @xmath176 with @xmath177 and @xmath82 being the identity matrix in @xmath178 , see @xcite for instance .
as mentioned in @xcite , can also be represented as @xmath179 from which we can get the gradient @xmath180.\end{aligned}\ ] ] using the formulas and we can have the following estimates , for @xmath181 , see @xcite ; @xmath182,\qquad&&%\\ \left|\nabla_y\gamma^\omega(x , y)\right| \leq \frac{1}{4\pi}\left[\frac{c_9}{\vert\;x - y\vert^2}+c_{10}\right],\label{gradkupradzeten1dfnt}\end{aligned}\ ] ] with @xmath183,\qquad c_9:=3\left(\frac{1}{c_s^2}+\frac{1}{c_p^2}\right),\ ] ] @xmath184 @xmath185 and @xmath186 $ ] where we assume the wave number @xmath36 and the lam parameters @xmath2 and @xmath3 to satisfy the condition @xmath187 .
the estimates ( [ gradkupradzeten1dfnt ] ) can be written as @xmath188 for different points @xmath189 , where @xmath190 the following observation will be useful for the proof of theorem [ equivalent - medimu ] .
for @xmath17 fixed , we distinguish between the obstacles @xmath191 , @xmath192 by keeping them into different layers based on their distance from @xmath61 .
let us first assume that @xmath193 for every @xmath143 .
hence each @xmath194 has the ( same ) volume @xmath60 and contains only one obstacle @xmath61 . without loss of generality
, we can take the @xmath194 s as cubes .
hence we can suppose that these cubes are arranged in cuboids , for example unit rubics cube , in different layers such that the total cubes upto the @xmath195 layer consists @xmath196 cubes for @xmath197 $ ] , and @xmath194 is located on the center , see fig [ distribution - obstacles ] .
hence the number of obstacles located in the @xmath195 , @xmath198 layer will be @xmath199=24 n^2 + 2 $ ] and their distance from @xmath61 is more than @xmath200 .
observe that , @xmath201 .
now , we come back to the case where @xmath202 .
first observe that @xmath203}{k(z_m)+1}\leq 1 $ ] . hence with such @xmath194 s , the total cubes located in the @xmath204th layer @xmath204
consists of at most the double of @xmath199 $ ] , i.e. @xmath205 .
we start with the following lemma on the uniform bounds of the elastic capacitances , see ( * ? ? ?
* lemma 3.1 ) and the references there in .
[ capacitance - eig - single ] let @xmath206 and @xmath207 be the minimal and maximal eigenvalues of the elastic capacitance matrices @xmath208 , for @xmath209
. denote by @xmath210 the capacitance of each scatterer in the acoustic case , , @xmath211 and @xmath80 is the solution of the integral equation of the first kind @xmath212 , see @xcite . ]
then we have the following estimate ; @xmath213 for @xmath214 .
the constant @xmath215 is the acoustic capacitance of the reference body @xmath162 which can be estimated above and below by the lipschitz character of @xmath162 , @xcite .
the following lemma provides us with the needed estimate on the invertibility of the algebraic system whose coefficient matrix @xmath216 is given by ; @xmath217 [ mazyawrkthmela]the matrix @xmath216 is invertible and the solution vector @xmath218 of satisfies the estimate : @xmath219 if @xmath220 .
in addition , @xmath221 since @xmath222 , then @xmath223 makes sense if @xmath224 as @xmath225 is proportional to the radius of @xmath162 , then ( [ algebraic - system - invertible - condition ] ) will be satisfied if @xmath36 and the lam parameters @xmath2 and @xmath3 satisfy the condition @xmath226 recalling that @xmath227 is defined in ( [ constant - final - estimates - green ] ) . finally , let us observe that the right hand side of ( [ upper - bound - omega - lame - coe ] ) depends only @xmath88and the lipschitz character of the reference obstacles @xmath162 s .
_ proof of lemma [ mazyawrkthmela ] .
_ we start by factorizing @xmath216 as @xmath228 where @xmath229 , @xmath230 is the identity matrix and @xmath231 .
we have @xmath232 , so it is enough to prove the injectivity in order to prove its invetibility . for this purpose , let @xmath233 are vectors in @xmath234 and consider the system @xmath235 we denote by @xmath236 and @xmath237 the real and the imaginary parts of the corresponding complex vector / matrix .
we also set @xmath238 by @xmath239 . from
we derive the following two identities : @xmath240 and then @xmath241 summing up and we obtain @xmath242 the right - hand side in can be estimated as @xmath243 let us now consider the right hand side in .
first we have @xmath244 where @xmath245 and @xmath246 if @xmath247 and @xmath248 for @xmath249 . hence from ( [ kupradzeten1 ] ) @xmath250 . from the observation before lemma [ capacitance - eig - single ] , we deduce that @xmath251}_{n=1}\hspace{-.1cm}2[(2n+1)^3-(2n-1)^3]\frac{{\mathring{c}}^2}{(4\pi)^2 n^2 \left(\frac{a^{\frac{1}{3}}}{2}\right)^{2 } } \leq 13 { \mathring{c}}^2 m \frac{a^{-\frac{2}{3}}}{\pi^2}\sum^{[2a^{-\frac{1}{3}}]}_{n=1}1 = \frac{26 m_{max}{\mathring{c}}^2 a^{-2}}{\pi^2}\end{aligned}\ ] ] or @xmath252 and then @xmath253 using lemma [ capacitance - eig - single ] , we deduce that @xmath254 from ( [ systemsolve1-small - sub1 - 2-ac-3 ] ) , ( [ systemsolve1-small - sub1 - 2-ac-4 ] ) , ( [ b_n - estimation-1 ] ) and ( [ lower - bound - c - i ] ) , we deduce that @xmath255 if @xmath256 and then the matrix @xmath216 in algebraic system is invertible . from the function @xmath158 , we define a bounded function @xmath257 as follows : @xmath258 . \end{array}\right.\ ] ] hence each @xmath194 contains @xmath259 $ ] obstacles and @xmath260 .
let @xmath261 be the @xmath262 matrix having entries as piecewise constant functions such that @xmath263 for all @xmath17 and vanishes outside @xmath4 . here
, @xmath208 are the capacitances of @xmath162 s . from @xcite
, we can observe that @xmath208 are defined through defined through @xmath264 , and are independent of @xmath60 .
we set @xmath265 consider the lippmann - schwinger equation @xmath266 and set the lam potential @xmath267 the coefficients @xmath268 and @xmath269 are uniformly bounded .
the next lemma concerns the mapping properties of the lam potential .
these properties are proved for the scalar poisson potential in @xcite , for instance .
similar arguments are applicable for the lam potential as well , so we omit to give the details .
[ lip - schw ] the operator @xmath270 is well defined and it is a bounded operator for any bounded domain @xmath271 in @xmath5 , i.e. there exists a positive constant @xmath66 such that @xmath272 we have also the following lemma .
[ invertibility - of - vc ] there exists one and only one solution @xmath273 of the lippmann - schwinger equation ( [ fracqcfracmain - effect - int ] ) and it satisfies the estimate @xmath274 where @xmath275 being a large bounded domain which contains @xmath276 .
_ proof of lemma [ invertibility - of - vc ] _ using the lemma [ lip - schw ] , we see that @xmath277 is fredholm with index zero and then we can apply the fredholm alternative to @xmath278 .
the uniqueness is a consequence of the uniqueness of the scattering problem corresponding to the model @xmath279 where @xmath280 and @xmath281 satisfies the kupradze radiation conditions and @xmath282 is an incident field .
the estimate ( [ est - lipm - sch ] ) can be derived , as it is done in @xcite for the acoustic case , by coupling the invertibility of @xmath277 and the @xmath283 interior estimates of the solutions of the system @xmath284 .
the capacitances of the obstacles @xmath285 , i.e. @xmath286 are bounded by their lipschitz constants , see @xcite , and we assumed that these lipschitz constants are uniformly bounded . hence @xmath269 is bounded in @xmath287 and then there exists a function @xmath159 in @xmath288 ( actually in every @xmath289 ) such that @xmath269 converges weakly to @xmath159 in @xmath287 .
now , since @xmath158 is continuous hence @xmath268 converges to @xmath290 in @xmath291 and hence in @xmath288 .
then we can show that @xmath292 converges to @xmath293 in @xmath288 .
since @xmath294 is bounded in @xmath295 , then from the invertibility of the lippmann - schwinger equation and the mapping properties of the lam potential , see lemma [ invertibility - of - vc ] , we deduce that @xmath296 is bounded and in particular , up to a sub - sequence , @xmath297 tends to @xmath298 in @xmath288 . from the convergence of @xmath299 to @xmath300 and the one of @xmath297 to @xmath298 and ( [ fracqcfracmain - effect - int ] ) , we derive the following equation satisfied by @xmath301 @xmath302 this is the lippmann - schwinger equation corresponding to the scattering problem @xmath303 , @xmath304 , and @xmath37 satisfies the kupradze radiation conditions .
as the corresponding farfields are of the form @xmath305 and the ones of @xmath306 are of the form @xmath307 we deduce that @xmath308 if we assume that @xmath310 $ ] , then we have the estimate @xmath311 , @xmath312 .
since the capacitances of the obstacles are assumed to be equal , we set @xmath159 to be a constant in @xmath4 and @xmath121 in @xmath313 . recall that @xmath314 and @xmath315 are solutions of the lippmann - schwinger equations @xmath316 and @xmath317 from the estimate @xmath311 , @xmath312 , we derive the estimate @xmath318 for each @xmath17 , we rewrite the equation as follows @xmath319 } \gamma^{\omega}(z_m , z_j)k_a(z_j)\bar{c}_j u_a(z_j ) vol(\omega_j)\nonumber\\ & & + \sum_{\substack{j=1 \\
j\neq m}}^{[a^{-1 } ] } \gamma^{\omega}(z_m , z_j)k_a(z_j)\bar{c}_j u_a(z_j ) vol(\omega_j)-\int_{\omega } \gamma^{\omega}(z_m , y)k_a(y){\bold{c}_a}(y ) u_a(y ) dy.\end{aligned}\ ] ] let us estimate the following quantities : @xmath320 } \gamma^{\omega}(z_m , z_j)k_a(z_j)\bar{c}_j u_a(z_j ) vol(\omega_j)-\int_{\omega } \gamma^{\omega}(z_m , y)k_a(y){\bold{c}_a}(y ) u_a(y ) dy\ ] ] and @xmath321 } \gamma^{\omega}(z_m , z_j)k_a(z_j)\bar{c}_j u_a(z_j ) vol(\omega_j).\ ] ] by the decomposition of @xmath4 , @xmath323}_{l=1}$ ] , we have @xmath324}\int_{\omega_l } \gamma^{\omega}(z_m , y){k_a}(y){\bold{c}_a}(y ) u_a(y ) dy.\ ] ] @xmath325 } \left [ \gamma^{\omega}(z_m , z_j)k_a(z_j)\bar{c}_j u_a(z_j ) vol(\omega_j ) -\int_{\omega_j } \gamma^{\omega}(z_m , y)k_a(y){\bold{c}_a}(y ) u_a(y ) dy \right].\label{rearranging - a}\end{aligned}\ ] ] for @xmath326 , we have @xmath327 dy .
\end{aligned}\ ] ] we set @xmath328 then every component @xmath329 of @xmath330 satisfies @xmath331 where @xmath332\,d\beta\nonumber\\ & = & \int_0 ^ 1\sum^3_{j=1}\left[\nabla_y\gamma_{i , j}^{\omega}(z_m , y-\beta(y - z_l))\right ] u_{a , j}(y-\beta(y - z_l))\,d\beta\nonumber\\ & & + \int_0 ^ 1\sum^3_{j=1}\gamma_{i , j}^{\omega}(z_m , y-\beta(y - z_l))\left[\nabla_y u_{a , j}(y-\beta(y - z_l))\right]\,d\beta .
\end{aligned}\ ] ] from and from section [ the - relative - distribution - of - the - small - bodies ] , we derive for @xmath333 @xmath334 where @xmath227 depends only on @xmath36 and some universal constants .
then @xmath335 then , for @xmath326 , and and observing that @xmath336 is a constant matrix in @xmath337 , imply the estimate @xmath338 & \vert y - z_l\vert dy \nonumber\\ & & + \frac{\mathring{c } \bar{c}_l k_a(z_l)}{2\pi\;n a^{\frac{1}{3 } } } \int_{\omega_l } \left[\int_0 ^ 1 { \vert \nabla_y u_{a}(y-\beta(y - z_l))\vert}d\beta \right]\vert y - z_l\vert dy \nonumber\\ \substack{\leq\\ \eqref{est - lipm - sch}}&c_1\frac{\textcolor{black}{[{k_a}(z_l)]}\bar{c}_l}{n^2 a^{\frac{2}{3}}}\,a^\frac{4}{3 } \,\substack{\leq\\ \eqref{sup - cap - all}}\ , c_1\frac{\textcolor{black}{k_{\max}}\mathcal{c}}{n^2 } \,a^\frac{2}{3},\qquad \end{aligned}\ ] ] for a suitable constant @xmath339 .
regarding the integral @xmath340 we do the following estimates : @xmath341\right)}_{=:lm(r , a)}\frac{\mathring{c}}{4\pi } c_2\bar{c}_m{k_a}&(z_m)\nonumber\\ \leq\,&\frac{\mathring{c}}{4\pi}c_2{k_a}(z_m)\bar{c}_m~lm(r^c , a),\qquad\qquad\nonumber\\ & & \mbox{here $ r^c$ is the value of $ r$ where $ lm(r , a)$ attains maximum}.\nonumber\\ & & { \partial_r lm(r , a)=0 \rightarrow 4\pi r-\frac{a}{r^2}-\frac{8}{3}\pi r=0\rightarrow r_c=\left(\frac{3}{4}\pi a\right)^\frac{1}{3}}\nonumber\\ & & { \begin{array}{ccc } lm(r_c , a)&=&2\pi\left(\frac{3}{4}\pi\right)^\frac{2}{3 } a^\frac{2}{3}+\left(\frac{4}{3\pi } \right)^\frac{1}{3}a^\frac{2}{3}-\frac{4}{3}\pi\left(\frac{3}{4}\pi \right)^\frac{2}{3}a^\frac{2}{3}\\ & & \\ & = & \left[\frac{2}{3\pi}\left(\frac{3}{4}\pi\right)^\frac{2}{3}+\left(\frac{4}{3\pi } \right)^\frac{1}{3 } \right]a^\frac{2}{3}=\frac{3}{2}\left(\frac{4}{3\pi } \right)^\frac{1}{3}a^\frac{2}{3 } \end{array } } \nonumber\\ \leq\,&\frac{3}{8\pi}\mathring{c}c_2k_{max}\mathcal{c}\left(\frac{4}{3\pi } \right)^\frac{1}{3}a^\frac{2}{3}.\qquad\qquad\end{aligned}\ ] ] from , we can have @xmath342 } \left [ \vert \gamma^{\omega}(z_m , z_j)k_a(z_j)\bar{c}_j u_a(z_j ) vol(\omega_j)-\int_{\omega_j } \gamma^{\omega}(z_m , y)k_a(y){\bold{c}_a}(y ) u_a(y ) dy \vert_\infty\right].\end{aligned}\ ] ] which we can estimate by @xmath343}_{n=1}2[(2n&+1)^3-(2n-1)^3 ] \bigg [ \vert \gamma^{\omega}(z_m , z_j)k_a(z_j)\bar{c}_j u_a(z_j ) vol(\omega_j)\\ & -\int_{\omega_j } \gamma^{\omega}(z_m , y)k_a(y){\bold{c}_a}(y ) u_a(y ) dy \vert_\infty\bigg]+\vert \int_{\omega_m } \gamma^{\omega}(z_m , y)k_a(y){\bold{c}_a}(y ) u_a(y ) dy \vert_\infty.\end{aligned}\ ] ] and then @xmath344.\ ] ] finally @xmath345 @xmath347 } \gamma^{\omega}(z_m , z_j)k_a(z_j)\bar{c}_j u_a(z_j ) vol(\omega_j ) \\
= \sum_{\substack{l=1 \\
l\neq m\\ z_l \in \omega_m}}^{\textcolor{black}{[k_a(z_m)]}}\gamma^{\omega}(z_m&,z_l)\bar{c}_l u_a(z_l ) a+ \sum_{\substack{j=1
\\ j\neq m}}^{[a^{-1 } ] } \sum_{\substack{l=1 \\ z_l \in \omega_j}}^{\textcolor{black}{[k_a(z_j)]}}\gamma^{\omega}(z_m , z_l)\bar{c}_l u_a(z_l ) a-
\sum_{\substack{j=1 \\ j\neq m}}^{[a^{-1 } ] } \gamma^{\omega}(z_m , z_j)k_a(z_j)\bar{c}_j u_a(z_j ) vol(\omega_j ) \\ = \bar{c}_m
l\neq m\\ z_l \in \omega_m}}^{\textcolor{black}{[k_a(z_m)]}}\gamma^{\omega}&(z_m , z_l ) u_a(z_l ) + \sum_{\substack{j=1 \\
j\neq m}}^{[a^{-1}]}\bar{c}_j a\big[\big(\sum_{\substack{l=1 \\
z_l \in \omega_j}}^{\textcolor{black}{[k_a(z_j)]}}\gamma^{\omega}(z_m , z_l ) u_a(z_l)\big)- \gamma^{\omega}(z_m , z_j)\textcolor{black}{[k_a(z_j ) ] } u_a(z_j)\big ] , \end{aligned}\ ] ] since @xmath348}}{k_a(z_j)}\ ; \mbox { and } \bar{c}_l=\bar{c}_j , \mbox { for } l=1 , ... , \ ; k_a(z_j)$ ] .
we write , @xmath349}}\gamma^{\omega}(z_m , z_l ) u_a(z_l)\end{aligned}\ ] ] and @xmath350}}\gamma^{\omega}(z_m , z_l ) u_a(z_l)\big)- \gamma^{\omega}(z_m , z_j)\textcolor{black}{[k_a(z_j ) ] } u_a(z_j)\big]\nonumber\\ & = & \sum_{\substack{l=1 \\
z_l \in \omega_j}}^{\textcolor{black}{[k_a(z_j)]}}\big(\gamma^{\omega}(z_m , z_l ) u_a(z_l)- \gamma^{\omega}(z_m , z_j ) u_a(z_j)\big).\end{aligned}\ ] ] we need to estimate @xmath351 and @xmath352}\bar{c}_j ae^j_2 $ ] .
+ now by writing @xmath353 . for @xmath354 , using taylor series , we can write @xmath355 with @xmath356 by doing the computations similar to the ones we have performed in ( [ taylorremind1]-[taylorremind - effect ] ) and by using lemma [ invertibility - of - vc ] , we obtain @xmath357}\bar{c}_j a e^j_2 \vert \leq c_4\mathcal{c}k_{max } a^{\frac{1}{3 } } \label{este2j1}\end{aligned}\ ] ] one can easily see that , @xmath358 substitution of in and using the estimates and associated to @xmath322 and the estimates and associated to @xmath346 gives us @xmath359 we rewrite the algebraic system ( [ fracqcfracmain ] ) as @xmath360 where we set @xmath361 , recalling that @xmath362 . taking the difference between and produces the algebraic system @xmath363 comparing this system with and by using lemma [ mazyawrkthmela ] , we obtain the estimate @xmath364 for the special case @xmath365 with @xmath366 , we have the following approximation of the far - field from the foldy - lax asymptotic expansion and from the definitions @xmath367 and @xmath264 , for @xmath17 : @xmath368 } \end{aligned}\ ] ] consider the far - field of type : @xmath369 corresponding to the scattering problem ( [ elas - scat - equi ] ) and set @xmath370 and @xmath371 taking the difference between and we have : @xmath372}\int_{\omega_j } e^{-i\frac{\omega}{c_p}\hat{x}\cdot{y}}{k_a}(y){\bold{c}_a}(y ) u_a(y)\cdot \hat{x}dy -\sum_{j=1}^{[a^{-1 } ] } \sum_{\substack{l=1 \\ z_l \in \omega_j}}^{\textcolor{black}{[k_a(z_j)]}}e^{-i\frac{\omega}{c_p}\hat{x}\cdot z_l}\bar{c}_lu_{a , l}\cdot \hat{x}a\nonumber\\ \nonumber & & \hspace{0cm}+o\left(a \hspace{-.03cm}+\hspace{-.03cm}a^{2 - 5t+3t\alpha}\hspace{-.03cm}+ \hspace{-.03cm}a^{3 - 9t+6t\alpha}\hspace{-.03cm}+\hspace{-.03cm}a^{1 - 2t\alpha}\hspace{-.03cm}+\hspace{-.03cm}a^{1 - 3t\alpha}\hspace{-.03cm}+
\hspace{-.03cm}a^{2 - 5t+2t\alpha}\right ) \\
\nonumber & = & \sum_{j=1}^{[a^{-1}]}{k_a}(z_j)\bar{c}_j\int_{\omega_j } \left[e^{-i\frac{\omega}{c_p}\hat{x}\cdot{y } } u_a(y)\cdot \hat{x } - e^{-i\frac{\omega}{c_p}\hat{x}\cdot z_j}u_a(z_j)\cdot \hat{x}\right]dy\nonumber\\ \nonumber \\ \nonumber & & + \sum_{j=1}^{[a^{-1}]}\bar{c}_ja \left[\sum_{\substack{l=1 \\
z_l \in \omega_j}}^{\textcolor{black}{[k_a(z_j)]}}\left(e^{-i\frac{\omega}{c_p}\hat{x}\cdot z_j } u_a(z_j)\cdot \hat{x}-e^{-i\frac{\omega}{c_p}\hat{x}\cdot z_l}u_a(z_l)\right)+\sum_{\substack{l=1 \\
z_l \in \omega_j}}^{\textcolor{black}{[k_a(z_j ) ] } } e^{-i\frac{\omega}{c_p}\hat{x}\cdot z_l
} \left(u_a(z_l)-u_{a , l}\right)\cdot \hat{x}\right]\nonumber\\ \nonumber & & \hspace{0cm}+o\left(a\hspace{-.03cm}+\hspace{-.03cm}a^{2 - 5t+3t\alpha}\hspace{-.03cm}+\hspace{-.03 cm } a^{3 - 9t+6t\alpha}\hspace{-.03cm}+\hspace{-.03cm}a^{1 - 2t\alpha } \hspace{-.03cm}+\hspace{-.03cm}a^{1 - 3t\alpha}\hspace{-.03cm}+ \hspace{-.03cm}a^{2 - 5t+2t\alpha}\right ) \\ \nonumber & = & \sum_{j=1}^{[a^{-1}]}\int_{\omega_j}{k_a}(z_j)\bar{c}_j \left[e^{-i\frac{\omega}{c_p}\hat{x}\cdot{y } } u_a(y)\cdot \hat{x } - e^{-i\frac{\omega}{c_p}\hat{x}\cdot z_j}u_a(z_j)\cdot \hat{x}\right]dy\nonumber\\ & & + \sum_{j=1}^{[a^{-1}]}\bar{c}_ja \sum_{\substack{l=1 \\
z_l \in \omega_j}}^{\textcolor{black}{[k_a(z_j)]}}\left(e^{-i\frac{\omega}{c_p}\hat{x}\cdot z_j } u_a(z_j)\cdot \hat{x}-e^{-i\frac{\omega}{c_p}\hat{x}\cdot z_l}u_a(z_l)\cdot \hat{x}\right)+\sum_{j=1}^{m}e^{-i\frac{\omega}{c_p}\hat{x}\cdot z_j}\bar{c}_ja \left[u_a(z_j)-u_{a , j}\right]\cdot \hat{x}\nonumber\\ \nonumber & & \hspace{0cm}+o\left(a\hspace{-.03cm}+\hspace{-.03 cm } a^{2 - 5t+3t\alpha}\hspace{-.03cm}+\hspace{-.03cm}a^{3 - 9t+6t\alpha } \hspace{-.03cm}+\hspace{-.03cm}a^{1 - 2t\alpha}\hspace{-.03cm}+ \hspace{-.03cm}a^{1 - 3t\alpha}\hspace{-.03cm}+\hspace{-.03 cm } a^{2 - 5t+2t\alpha}\right)\\ & \substack{= \\ \eqref{mazya - fnlinvert - small - ac-3-effect - dif } } & \sum_{j=1}^{[a^{-1}]}{k_a}(z_j)\bar{c}_j\int_{\omega_j } \left[e^{-i\frac{\omega}{c_p}\hat{x}\cdot{y } } u_a(y)\cdot \hat{x } - e^{-i\frac{\omega}{c_p}\hat{x}\cdot z_j}u_a(z_j)\cdot \hat{x}\right]dy\nonumber\\ & & + \sum_{j=1}^{[a^{-1}]}\bar{c}_ja \sum_{\substack{l=1 \\
z_l \in \omega_j}}^{\textcolor{black}{[k_a(z_j)]}}\left(e^{-i\frac{\omega}{c_p}\hat{x}\cdot z_j } u_a(z_j)\cdot \hat{x}-e^{-i\frac{\omega}{c_p}\hat{x}\cdot z_l}u_a(z_l)\cdot \hat{x}\right)+o\left(\mathcal{c}^2 k_{max } ( a^{\frac{1}{3}}+a^{1-t})\right)\nonumber\\ & & + o\left(a\hspace{-.03cm}+\hspace{-.03cm}a^{2 - 5t+3t\alpha}\hspace{-.03cm}+ \hspace{-.03cm}a^{3 - 9t+6t\alpha}\hspace{-.03cm}+\hspace{-.03 cm } a^{1 - 2t\alpha}\hspace{-.03cm}+\hspace{-.03cm}a^{1 - 3t\alpha } \hspace{-.03cm}+\hspace{-.03cm}a^{2 - 5t+2t\alpha}\right ) . \end{aligned}\ ] ] now , let us estimate the difference @xmath373}{k_a}(z_j)\bar{c}_j\int_{\omega_j } \left[e^{-i\frac{\omega}{c_p}\hat{x}\cdot{y } } u_a(y ) - e^{-i\frac{\omega}{c_p}\hat{x}\cdot z_j}u_a(z_j)\right]dy$ ] .
write , @xmath374 . using taylor series , we can write @xmath375 with @xmath376 \right]_k\,d\beta\nonumber\\ & = & \int_0 ^ 1\left[\nabla_ye^{-i\frac{\omega}{c_p}\hat{x}\cdot(y-\beta(y - z_j))}\right]_k u_{a}(y-\beta(y - z_j))\,d\beta\nonumber\\ % & & + \int_0 ^ 1e^{-i\frac{\omega}{c_p}\hat{x}\cdot(y-\beta(y - z_j))}\left[\nabla_yc(y-\beta(y - z_j))\right ] y(y-\beta(y - z_j))\,d\beta\nonumber\\ & & + \int_0 ^ 1e^{-i\frac{\omega}{c_p}\hat{x}\cdot(y-\beta(y - z_j))}\left[\nabla_y u_{a}(y-\beta(y - z_j))\right]_k\,d\beta . \end{aligned}\ ] ] we have @xmath377 then @xmath378 using we get the estimate @xmath379}{k_a}(z_j)\bar{c}_j\int_{\omega_j } \left[e^{-i\frac{\omega}{c_p}\hat{x}\cdot{y}}(y ) u_a(y ) - e^{-i\frac{\omega}{c_p}\hat{x}\cdot z_j}u_a(z_j)\right]dy\right \vert_\infty \leq\ ] ] @xmath380}{k_a}(z_j)\bar{c}_j \left(\frac{\omega}{c_p } \int_{\omega_j}\vert y - z_j\vert_\infty\int_0 ^ 1 |u_{a}(y-\beta(y - z_j))|\ , d\beta\,dy\right)\,\ ] ] @xmath381}{k_a}(z_j)\bar{c}_j \left(\int_{\omega_j}\vert y - z_j\vert\int_0 ^ 1\vert\nabla_y u_{a}(y-\beta(y - z_j))\vert_\infty \,d\beta\,dy\right)\ ] ] @xmath382}{k_a}(z_j)\bar{c}_jc_1\ , a\,a^\frac{1}{3}\,\left(\frac{\omega}{c_p } + c_5\right ) \leq k_{max } \mathcal{c}c_1\left(\frac{\omega}{c_p } + c_5\right)\,a^{\frac{1}{3}}.\ ] ] in the similar way , using ( [ mazya - fnlinvert - small - ac-3-effect - dif ] ) , we have , @xmath383}\bar{c}_ja \sum_{\substack{l=1 \\ z_l \in \omega_j}}^{\textcolor{black}{[k_a(z_j)]}}\left(e^{-i\frac{\omega}{c_p}\hat{x}\cdot z_j } u_a(z_j)-e^{-i\frac{\omega}{c_p}\hat{x}\cdot z_l}u_a(z_l)\right)\right\vert_\infty & \leq & o\left(k_{max } \mathcal{c}(a^{\frac{1}{3}}+a^{1-t})\right ) . \end{aligned}\ ] ] using the estimates and in , we obtain @xmath384 since @xmath385 is of order @xmath386 , and @xmath387 is of the order @xmath388 , we should have @xmath389 .
hence @xmath390 hence for @xmath391 , we have @xmath392 equating @xmath393 , we find that @xmath394 and then @xmath395 and @xmath396 .
in addition , since @xmath397 , then @xmath398 . hence , the error is @xmath399\cdot \hat{x } & = & o\left ( a^{\frac{1}{3 } } + a^{\frac{3}{2}-3t}\right),\;~~ \frac{1}{3 } \leq t \leq \frac{7}{12}. \end{aligned}\ ] ] combining the estimates ( [ elastic - difference - farfield - effect-1**1 - 1 ] ) and ( [ appro-0-a ] ) , we deduce that @xmath400\cdot \hat{x}=o(a^{\min \{\gamma,\ ; \frac{1}{3},\ ; \frac{3}{2}-3t\}}),\ ; a<<1 , \;~~ \frac{1}{3 } \leq t \leq \frac{7}{12}\ ] ] uniformly in terms of @xmath401 .
for an obstacle @xmath402 of radius @xmath403 , @xmath404 and @xmath405 . similarly , we set @xmath406 and @xmath407 .
we see that @xmath408 satisfies @xmath409 with the kupradze radiation conditions .
since @xmath410 , @xmath411 is bounded in @xmath289 , for @xmath412 , by interior estimates , we deduce that @xmath413 , @xmath411 is bounded in @xmath414 , for @xmath412 , and hence , in particular , the normal traces are bounded in @xmath415
. then we can show that the norms of the operators @xmath416 and @xmath417 are of the order @xmath418 at least . 1 . using these properties and arguing as in @xcite
, we derive the asymptotic expansions ( [ x oustdie1 d_m farmainp - near - rho])-([x oustdie1 d_m farmains - near - rho ] ) . indeed ,
apart from the computations done in @xcite , the main arguments needed to extend those results to the case of variable density is the fredholm alternative for the corresponding integral operators and the application of the neumann series expansions . after splitting @xmath419 as @xmath420 , these two arguments are applicable as soon as we have ( [ single - layer - rho])-([double - layer - rho ] ) .
2 . the justification of the invertibility of the algebraic system ( [ fracqcfracmain ] ) depends only on ( 1 ) the distribution of small bodies and ( 2 ) the background medium through the singularities of the fundamental solution ( of the form @xmath421 ) . however , this type of singularity is true for general background elastic media . with the singularity of @xmath78 and the smoothness of @xmath422
then the same arguments can be used to justify the invertibiliy of the algebraic system ( [ fracqcfracmain - rho ] ) . using the above mentioned decomposition of the green s function @xmath419 , the properties of the lippmann schwinger integral equation are also valid replacing @xmath78 by @xmath419 , and hence the results in section [ subsection - piecewise - constant ] are valid . finally , and again using the decomposition of @xmath419 , the computations in section [ the approximation - by - as ] can be carried out using @xmath419 .
we start with the following lemma on the symmetry structure of the elastic capacitance .
_ proof of lemma [ lemma - cap - adj ] . _ we know that the matrix @xmath427 solves the invertible integral equation @xmath428 or precisely @xmath429 , where @xmath430 and @xmath431 is the @xmath432 column of @xmath433 .
let @xmath60 be any constant vector in @xmath5 , then the vector @xmath434 satisfies @xmath435 we set @xmath436 .
then @xmath437 satisfies the problem @xmath438 , in @xmath424 and @xmath439 on @xmath440 .
in addition , we have the jump relation @xmath441 on @xmath440 where @xmath442 is the elastic conormal derivative .
hence @xmath443 now , let @xmath60 and @xmath444 be arbitrary constant vectors in @xmath5 . to both @xmath60 and @xmath444
, we correspond @xmath437 and @xmath445 as above . using the green formulas inside and outside of @xmath424
, we deduce that @xmath446 recalling that every quantity here is real valued .
[ capacitance - rotation ] let @xmath447 be a unitary transform in @xmath448 , @xmath424 be bounded lipschitz domain in @xmath448 , @xmath449 and @xmath450 .
let @xmath118 and @xmath451 be the corresponding elastic capacitance matrices due to the denisty matrices @xmath452 and @xmath453 , as defined in , respectively .
then we have @xmath454 .
_ proof of lemma [ capacitance - rotation ] .
_ first recall the relation @xmath455 , see ( * ? ? ?
* lemma 6.11 ) .
from , we have that @xmath456 now from the uniqueness of solutions of , we deduce that @xmath457 and then @xmath458 from the definition of the capacitance , see ( [ barqcimsurfacefrm1main ] ) , and , we have @xmath459 [ capacitance - rotation-2drltn ] let @xmath447 be a unitary transform in @xmath448 , @xmath460 , and let @xmath424 be a bounded lipschitz domain in @xmath448 , @xmath449 , and @xmath450 .
let @xmath118 and @xmath451 be the corresponding elastic capacitance matrices due to the denisty matrices @xmath452 and @xmath453 , as defined in , respectively . 1 . @xmath461-case .
if the shape of @xmath424 is rotationally invariant for any rotation by one angle @xmath462 , then @xmath118 is a scalar multiplied by the identity matrix .
2 . @xmath463-case .
if the shape of @xmath424 is rotationally invariant for any two of the rotations around the @xmath464 or @xmath465 axis by one angle @xmath462 and @xmath466 respectively , then @xmath118 is a scalar multiplied by the identity matrix .
_ proof of proposition [ capacitance - rotation-2drltn ] . _ in @xmath467 case , the rotation matrix by an angle @xmath468 is given by @xmath469 as the shape is invariant by this rotation then @xmath470 . since @xmath471 is unitary then @xmath472 and then implies @xmath473 we deduce from ( [ cap - rotation-1 ] ) and the symmetry of matrix @xmath118 the following relations : @xmath474 @xmath475 @xmath476 we rewrite ( [ relations-2d-1 ] ) and ( [ relations-2d-3 ] ) respectively as @xmath477 @xmath478 taking @xmath479 as a multiplicative factor in ( [ relations-2d-1 - 1 ] ) and ( [ relations-2d-3 - 1 ] ) we see that if @xmath462 , i.e. @xmath480 , then we have @xmath481 let us now consider the 3d case .
first , let us assume that the shape is invariant under the rotation about the @xmath482 and an angle @xmath462 .
this rotation matrix is given by @xmath483
since @xmath471 is unitary , @xmath472 , then gives us ; @xmath484 @xmath485 we observe the equality of the @xmath486 matrices : @xmath487 to show that @xmath118 is scalar multiplied by the identity matrix we need to prove that @xmath489 , for instance , and @xmath490 .
for this purpose , we use another rotation . taking the rotation around the @xmath465-axis axis . ] by one angle @xmath491 and proceeding as we did for the rotation about the @xmath492-axis , we show that @xmath493 1 . for the spherical shapes , in particular ,
the capacitance is a scalar multiplied by the identity matrix .
ellipsoidal shapes are invariant only under rotations with angle @xmath494 ( or trivially @xmath495 ) . for these shapes
, the capacitance might not be a scalar multiplied by the identity matrix but a diagonal matrix instead . to justify this property ,
the arguments in @xcite can be useful . | we deal with the elastic scattering by a large number @xmath0 of rigid bodies of arbitrary shapes with maximum radius @xmath1 with constant lam coefficients @xmath2 and @xmath3 .
we show that , when these rigid bodies are distributed arbitrarily ( not necessarily periodically ) in a bounded region @xmath4 of @xmath5 where their number is @xmath6 and the minimum distance between them is @xmath7 with @xmath8 in some appropriate range , as @xmath9 , the generated far - field patterns approximate the far - field patterns generated by an equivalent ( possibly variable ) mass density .
this mass density is described by two coefficients : one modeling the local distribution of the small bodies and the other one by their geometries . in particular ,
if the distributed bodies have a uniform spherical shape then the equivalent mass density is isotropic while for general shapes it might be anisotropic . in addition , we can distribute the small bodies in such a way that the equivalent mass density is negative .
finally , if the background density @xmath10 is variable in @xmath4 and @xmath11 in @xmath12 , then if we remove from @xmath4 appropriately distributed small bodies then the equivalent density will be equal to unity in @xmath5 , i.e. the obstacle @xmath4 characterized by @xmath10 is approximately cloaked .
* keywords * : elastic wave scattering , small - scatterers , effective mass density . |
nearby star forming regions display a continuum of properties , ranging from isolated young stellar objects and loose `` aggregates '' @xcite to dense clusters containing hundreds if not thousands of stars within parsec sized regions @xcite .
the diverse range of observed stellar environments reflect the various physical processes that initiate star formation over the lifetime of molecular clouds . in principle , the predominant mechanisms that lead to star formation can be inferred by reconstructing the star formation rate in molecular clouds as a function of time and stellar mass . while a comprehensive picture of how star formation evolves in time
requires obtaining masses and ages for the individual stars that have formed within a cloud , constraints on the star formation history can nonetheless be obtained from a census of the embedded stellar population to establish the number of stars that have formed and where in the cloud they tend to be located .
the stellar population associated with a cloud is most often identified using large scale surveys for an emission feature commonly associated with pre - main - sequence stars , such as far - infrared emission ( e.g. beichman et al .
1986 ) , near - infrared excesses ( e.g. cambrsy et al .
1998 ) , h@xmath4 ( e.g. nakano , wiramihardja , & kogure 1995 ) , or x - rays ( e.g. walter et al . 1988 ) .
while such studies generally produce reliable catalogs of young stellar objects , they may not provide a complete census of the stellar population in molecular clouds .
surveys for far - infrared emission , near - infrared excesses , and h@xmath4 in particular generally detect pre - main - sequence objects surrounded by an optically thick circumstellar disk or envelope .
the evolution of circumstellar material as a function of stellar age and mass remains poorly quantified , which makes it difficult to infer the extent of the stellar population not currently in this phase of stellar evolution and not detectable by these diagnostics . indeed , x - ray observations , which probe stellar chromospheric activity and
can readily detect stars as old as @xmath1100 myr in nearby ( @xmath5200 pc ) star forming regions , have detected a widespread population of objects in and around molecular clouds .
the interpretation of the x - ray results remain controversial , though , as the spatially extended x - ray population may represent older pre - main - sequence stars that have dispersed from molecular clouds , stars that formed in surrounding cloudlets @xcite , or young main sequence stars that have formed in the solar neighborhood in the past 100 myr @xcite .
furthermore , x - ray observations in the pre - chandra era lack the required combination of sensitivity , coverage , and angular resolution to provide an accurate census of the stellar population contained in both isolated star forming regions and dense clusters over entire molecular clouds .
an alternative approach to infer the underlying stellar population in molecular clouds is to statistically determine the excess number of stars observed toward the cloud relative to the expected field star contribution .
such star count techniques are best suited for molecular clouds located at high galactic latitudes where the field star contamination is less severe , and for near - infrared wavelength surveys where embedded stars are more readily detected .
while near - infrared star count analyses have the obvious limitation that individual young stellar objects can not be uniquely identified for further study , they represent a powerful probe of the stellar population in that near - infrared emission generally originates from the stellar photosphere ( except for the heavily accreting objects ) .
the intrinsic brightness of a star then depends foremost on the stellar mass ( and age for pre - main - sequence objects ) , and the parameter space probed by a set of observations can be estimated with the aid of pre - main - sequence evolutionary models .
further , current observations are generally sensitive to stars spanning a broad range of masses and ages , including objects in nearby molecular clouds with masses below the hydrogen burning limit , and have the angular resolution needed to resolve many of the stars in dense clusters .
thus near - infrared star count studies have the potential obtain a more unbiased census of the embedded stellar population in molecular clouds .
the first extensive near - infrared observations to statistically characterize the spatial distribution of young stars in a molecular cloud was the @xmath6 band survey of orion b ( l1630 ) by lada et al .
( 1991 ; see also li , evans , & lada 1997 ) .
surveys of other regions with similar aims have since been conducted for orion a ( l1641 ; strom , strom , & merrill 1993 ) , ngc2264 @xcite , the rosette @xcite , ngc1333 @xcite , and ophiuchus @xcite among others ( see recent reviews by allen & hillenbrand 2000 ; meyer & lada 2000 ; clarke , bonnell , & hillenbrand 2000 ) .
a synthesis of these results suggest that more than half of the stellar population in molecular clouds is contained in clusters with various sizes and densities @xcite .
however , most near - infrared observations to date cover only a fraction of the total cloud area and nearly all have been predisposed toward regions within clouds known _ a priori _ to contain prominent star formation activity ( and likely clusters ) . thus current near - infrared observations do not accurately characterize the spatial distribution of stars over entire molecular clouds .
the 2 micron all sky survey ( _ 2mass _ ) removes many of the limitations of prior observations and has provided the first sensitive , uniformly calibrated @xmath7 , @xmath8 , and @xmath9 near - infrared survey over entire molecular clouds . in this paper
, we analyze data from the _
2mass _ second incremental release to investigate the spatial distribution of young stars in the perseus , orion a , orion b , and monoceros r2 ( monr2 ) molecular clouds in order to determine the relative importance of cluster and isolated stellar systems .
these clouds were chosen for this study since they are nearby with distances less than 1 kpc to maximum sensitivity to the low mass stellar population , are located at high galactic latitudes ( @xmath10 ) to minimize field star contamination , and have available @xmath11co maps that are required for the analysis presented here .
the results from this study are presented as follows . in section [ data ] , we review the characteristics of the _ 2mass _ data and select a subset of the point source catalog for analysis . in section [ analysis ] , the near - infrared star counts toward the perseus , orion a , orion b , and monr2 molecular clouds are used to identify any cluster and distributed stellar populations .
the implications of these results for the star formation history of molecular clouds is discussed in section [ discussion ] , and the conclusions are summarized in section [ summary ] .
the near - infrared data analyzed in this study were obtained from the _ 2mass_second incremental data release .
the observational goal of _ 2mass _ is to survey @xmath395% of the sky simultaneously in the @xmath7 , @xmath8 , and @xmath9 bands .
the data are obtained by scanning a series of @xmath12 long tiles aligned in declination . each position in a tile is observed first with a short , 51 msec integration to recover photometry for bright stars ( between @xmath1 5 - 8@xmath13 ) , and then with a relatively long , 1.3 sec integration for deeper photometry .
adjacent images within a tile overlap such that each position on the sky is nominally observed 6 times .
the pixel scale of the observations is 2 , but since overlapping images within a tile are designed to subsample the raw camera pixels , the coadded images produced in data reduction from the long integration frames have a pixel scale of 1 .
the _ 2mass _ second incremental release point source catalog contains a list of positions and magnitudes for point sources in @xmath147% of the sky that have a photometric uncertainty @xmath14 0.155@xmath13 ( i.e. a signal to noise ratio @xmath15 ) in at least one of the three observed bands .
this study analyzes a subset of the catalog that includes the perseus , orion a , orion b , and monr2 molecular clouds . at the time of the second incremental release , _
2mass _ data are available for the entire area of the monr2 and orion a molecular clouds , @xmath180% of perseus , and @xmath170% of orion b. the galactic coordinates of these clouds and their adopted distances are summarized in table [ tbl : clouds ] . the _
2mass _ point source data need to be analyzed at magnitude limits bright enough to ensure uniform completeness across the region studied .
the magnitude thresholds for the three bands were set by computing as a function of magnitude the average signal to noise ratio and the fraction of stars that have a signal to noise ratio @xmath16 10 for point sources between galactic longitudes of 130 and 250 and galactic latitudes of -50 and + 50 . based on this analysis , the faint magnitudes thresholds for analyzing the star counts
were set at 15.8@xmath13 , 14.9@xmath13 , and 14.3@xmath13 at @xmath7 , @xmath8 , and @xmath9 band respectively . at these thresholds
, the average signal to noise for the point sources is between 14 - 15 , and greater than 99% , 99% , and 97% of the sources at these magnitude limits have a photometric uncertainty corresponding to a signal to noise ratio @xmath16 10 .
stars brighter than @xmath15@xmath13 are saturated even in the short integration data , and as a conservative limit , it was required that stars be fainter than 6.0@xmath13 in the appropriate band to be included in the analysis . at the galactic coordinates of the clouds studied here , the average source densities at the faint magnitude thresholds are @xmath52arcmin@xmath0 , which implies a typical source separation of @xmath1740 .
thus source identification will only be confusion limited in the densest clusters which can have substantially higher surface densities than these average values .
individual survey tiles may have poorer sensitivity than the adopted magnitude thresholds if spatial structure in the sky background from airglow emission added substantially to the sky noise ( especially at @xmath8 band ) at the time of the observation or if the sky background was much higher than average ( especially at @xmath9 band ) .
data adversely affected by airglow were identified and removed in the present analysis by finding all tiles in the second incremental release scan database in which the observed sky noise was 20% higher than the expected noise computed from the sky background level . the percentage of the total number of tiles ( @xmath19400 total ) removed due to this criteria amounted to 0% , 8% , 0.1% at @xmath7 , @xmath8 , and @xmath9 band respectively . to remove lower sensitivity data because of a high sky background level ,
tiles in which the average signal to noise ratio is less than 10.0 at the adopted magnitude thresholds were also discarded .
the percentage of tiles removed by this criteria were 0.14% , 0.07% , and 0.25% at @xmath7 , @xmath8 , and @xmath9 band .
finally , some point sources identified around bright stars ( @xmath18 @xmath5 9.5@xmath13 ) were omitted when drafting the second incremental release catalogs so that image artifacts from bright objects would not reduce the data reliability .
the masked areas are primarily due to diffraction spikes and circular regions centered on bright saturated stars .
the dimensions of these masks depend on the magnitude of the bright star and effectively create holes in the star count maps . using the quantitative description of these masks provided in the
_ 2mass _ explanatory supplement , the fraction of each pixel in the star count map that has been partially or completely masked was computed as a function of magnitude and was used to correct the star counts maps on a pixel by pixel basis .
a @xmath9 band stellar surface density map for the area between galactic longitudes of 130 and 250 and galactic latitudes of @xmath19 and @xmath20 is shown in figure [ ksqrt ] .
the surface density map was generated by binning stars with @xmath9 band magnitudes between 6.0@xmath13 and 14.3@xmath13 in @xmath21 cells .
the galactic plane is clearly visible as a band of stars extending across the center of the image .
the location of the perseus , orion a , orion b , and monr2 molecular clouds are labeled in figure [ ksqrt ] , but because of the low @xmath9 band extinction , the outline of the clouds are only faintly seen in absorption against the backdrop of field stars . to gain a better perspective on the location of various molecular clouds ,
figure [ jk ] presents a map of the average @xmath22 stellar color for the same region shown in figure [ ksqrt ] .
large @xmath22 colors , represented by darker gray scales in figure [ jk ] , are attributed to background field stars and embedded young stellar objects reddened by dust associated with molecular clouds . the taurus , perseus , orion a , orion b , and monr2 molecular clouds are clearly visible in figure [ jk ] , as well as numerous clouds in the galactic plane .
the outline of the clouds agree well with that inferred from optical star counts @xcite and far - infrared emission @xcite .
the observed star counts toward a molecular cloud consist of unreddened foreground field stars , background field stars reddened by dust within the cloud , and young stars associated with the cloud itself .
the spatial distribution of young stars in the cloud can then be obtained by removing the field stars from the star count map .
ideally one would like to specifically identify the individual young stars so that they can be furthered studied .
however , near - infrared photometry alone can not uniquely distinguish field stars from the stellar population associated with the cloud .
stars with relatively blue near - infrared colors can either be foreground field stars , background field stars appearing through holes in the cloud , or young stars on the surface of the cloud .
similarly , red stars can either be background reddened field stars or pre - main - sequence objects embedded in the cloud .
the degeneracy between the field and embedded stellar population can be only partially reduced by considering both the magnitudes and colors .
therefore , instead of attempting to uniquely identify which stars are associated with the cloud , the embedded stellar population was inferred statistically by first constructing a semi - empirical model of the expected field star surface density toward the cloud , and then subtracting the field star model from the observed stellar surface density map . in principle , the star count analysis can be done by simultaneously analyzing the five observable quantities for each star ( i.e. galactic longitude and latitude and the observed @xmath7 , @xmath8 , and @xmath9 band magnitudes ) . in the analysis conducted here
, the three near - infrared bands are analyzed separately to determine if each band provides a consistent picture of the embedded stellar population .
the following sections describe how the semi - empirical field star model was generated using _
2mass _ observations of regions away from the molecular clouds and published @xmath11co(10 ) maps as a tracer of the cloud extinction .
the field star subtracted stellar surface density map is then used to identify stellar clusters and any stars distributed more uniformly throughout the molecular cloud .
the expected number of field stars toward the molecular clouds in the absence of extinction from the cloud itself was estimated using _
2mass _ observations of regions outside the molecular cloud boundaries .
this was accomplished by fitting legendre polynomials as a function of galactic longitude and latitude to @xmath7 , @xmath8 , and @xmath9 surface density maps for the region between @xmath23 to 240 and @xmath24 to @xmath25 binned in @xmath21 cells ( see figure [ ksqrt ] ) .
this bin size was chosen so that large scale gradients in the field star counts are well resolved while maintaining a manageable image size .
a global fit was performed to the large scale surface density maps as opposed to fitting a localized region around each cloud since several blocks of _ 2mass _ tiles were not available at the time of this study , especially near the orion b molecular cloud . without these tiles ,
a localized polynomial fit is poorly constrained near the cloud boundaries .
lines of sight in the surface density map that pass through an obvious stellar cluster ( see figure [ ksqrt ] ) or that intercept a molecular cloud as traced by molecular emission @xcite or red @xmath22 colors ( see figure [ jk ] ) were excluded from the fit .
at least a 0.5 border around each cloud was also excluded which prevented any potential young stars surrounding the clouds from influencing the polynomial fit .
the remaining @xmath1180,000 pixels in each surface density map were then fitted with legendre polynomials .
the order of the polynomial fit was increased until the systematics in the residuals as a function of position were no longer evident .
a seventh order polynomial fit in both longitude and latitude were ultimately used .
the results from the polynomial fit are examined in figure [ fit ] , which shows as a function of galactic latitude the observed mean stellar surface density , the mean residuals after subtracting the polynomial fit , and the rms in the residuals for each band .
the rms of the residuals varies between @xmath1 0.1 - 0.3arcmin@xmath0 depending on the band and the galactic latitude , and is consistent with that expected from poisson statistics within the @xmath21 bin sizes .
the polynomial fit described above was interpolated to yield the expected surface density of field stars as a function of galactic longitude and latitude toward the molecular clouds in the absence of extinction from the cloud itself .
the uncertainties associated with interpolating the polynomial fit were assessed by performing a similar fit to star counts at the same galactic longitude but at positive latitudes where are there essentially no clouds ( see figure [ jk ] ) .
the molecular clouds masks used in fitting the star counts at negative latitude were also used at positive latitudes in in order to preserve the geometry and number of masked regions .
the accuracy of the interpolated polynomial fit was assessed by subtracting the fit from the observed star counts in the masked regions and computing the rms of the residuals in areas 6 deg@xmath26 in size , which is a typical area of the larger clouds analyzed here .
it was found that the average difference over this sized region between the interpolated fit and the observed star counts is @xmath50.003arcmin@xmath0 in each band with a rms deviation of @xmath1 0.002arcmin@xmath0 .
dust within the molecular clouds obscures background field stars and depresses the total field star counts relative to the interpolated polynomial fit .
the number of field stars that are obscured depends on the extinction along the line of sight and the frequency distribution of magnitudes for the background field star population , as foreground field stars will not be further reddened by the cloud .
the extinction as a function of position through the molecular clouds was estimated using published @xmath11co maps of the monr2 , orion a , orion b , and perseus molecular clouds kindly provided by j. bally ( see padoan et al . 1999 ; miesch & bally 1994 ; bally et al .
these maps have a full - width - at - half - maximum beam size of 100 with a sampling of 60 .
while the molecular gas will contain substructure within the 100@xmath27 beam , these maps resolve the large scale structural features present in these clouds and should accurately trace the global spatial distribution of extinction .
assuming that the @xmath11co(10 ) emission is optically thin and in local thermodynamic equilibrium , the @xmath11co column density can be estimated from the j=1 - 0 integrated intensity using the formula @xmath28 where @xmath29 is the excitation temperature and @xmath30 the integrated @xmath11co(10 ) intensity .
the derived @xmath11co column densities will be in error by less than a factor of 2 by assuming @xmath29=10 k as long as the actual excitation temperature is less than 30 k. such high excitation temperatures are expected only toward the embedded ob stars which occupy a small fraction of the total cloud area in any of these regions .
the @xmath11cocolumns densities can be converted into visual extinctions using the empirical @xmath31@xmath32 correlations that has been observed in many clouds @xcite .
these studies have shown that the ratio of @xmath31 to @xmath32 is typically within the range of @xmath1@xmath33 cm@xmath0 mag@xmath35 and that @xmath11co is detectable only when the visual extinction reaches a threshold value of @xmath10.5 - 1.5 magnitude due to photodissociation of the molecules at low column densities @xcite . to determine the sensitivity of the field star model on the derived extinctions , three different set of assumptions were adopted to convert the @xmath11co column densities into visual extinctions .
the `` nominal '' extinction model assumes a @xmath36 ratio of @xmath37 cm@xmath0 mag@xmath35 as derived by @xcite and that the extinction needed to detect @xmath11co(10 ) is @xmath38@xmath13 .
this calibration was adopted since it has the best statistical accuracy among published observations and the parameters are in the middle of values generally observed in molecular clouds .
similarly , a `` low '' extinction model was derived by assuming a @xmath36 ratio of @xmath39 cm@xmath0 mag@xmath35 and a visual extinction threshold of 0.5@xmath13 , and a `` high '' extinction model was obtained by assuming a @xmath31/@xmath32 ratio of @xmath40 cm@xmath0 mag@xmath35 and a visual extinction threshold of 1.5@xmath13 .
the low and high extinction models adopt parameters at the extrema of the observed values and are assumed to represent the plausible range of visual extinction and are not formal 1@xmath41 confidence intervals .
the masses of the clouds under these three extinction assumptions are summarized in table [ tbl : clouds ] .
the observed correlation between @xmath11co column density and visual extinction is valid for @xmath42@xmath13 @xcite . for @xmath32 @xmath1710@xmath13 , the @xmath11co(10 )
emission is saturated and is not a reliable tracer of the extinction .
the @xmath11co maps used here have a 3@xmath41 detection level corresponding to a visual extinction of @xmath10.2@xmath13 ( not including the visual extinction threshold ) for the nominal extinction model .
the fraction of the perseus , orion a , orion b , and monr2 cloud area with an inferred visual extinction @xmath14 5@xmath13 is 95% , 88% , 92% , and 98% respectively .
therefore , @xmath11co(10 ) emission should accurately trace the extinction over nearly the entire cloud area .
the last step in constructing the field star model is to estimate the fraction of background field stars that become obscured by dust within the molecular clouds .
this fraction was estimated using the extinction maps described above and the @xcite galactic star count model .
this model agrees with the observed star counts to within 15% for high galactic latitude regions ( @xmath43 ) that do not contain additional reddening along the line of sight from molecular clouds . since perseus , orion a , orion b , and monr2 are all located at galactic latitudes between @xmath44 and @xmath45
, we expect that this model should provide an accurate estimate of the fraction of field stars that are foreground to the molecular clouds .
the @xcite model predicts that the fraction of the total field star counts contributed by foreground field stars ranges from 0.05 - 0.08 in perseus to 0.2 - 0.4 in monr2 in the @xmath7 , @xmath8 , and @xmath9 bands .
the predicted absolute level of the foreground field star contamination is between 0.04arcmin@xmath0 ( perseus ) and @xmath50.3arcmin@xmath0 ( monr2 ) . for each line of sight toward the molecular clouds with a @xmath11co(10 ) detection , the expected number of field stars were computed using the wainscoat star model with and without the @xmath11co derived extinction added at the appropriate cloud distance .
the fraction of the field star population in this model that are reddened to magnitudes outside the adopted thresholds ( see section [ data ] ) was then computed , and the expected field star contribution inferred from the polynomial fits was reduced by the same fraction for that line of sight .
thus the total number of unreddened field stars is still determined by _ 2mass _ observations , and only the relative fraction of field stars obscured by extinction from dust in molecular clouds is model dependent . in performing the calculations
, we adopted the interstellar reddening vector from @xcite and the cloud distances summarized in table [ tbl : clouds ] .
uncertainties in the molecular cloud distances and the three dimensional structure of the clouds contribute to uncertainties in the field star model since extinction from the molecular cloud is applied only to the background field stars .
the distance dependent uncertainties on the field star subtraction were quantified using the @xcite star count model . adopting an average visual extinction of 3@xmath13
as implied by the @xmath11co maps , the @xmath6 band field star surface density computed from the wainscoat model will vary by @xmath1@xmath46 0.002arcmin@xmath0 for perseus , @xmath46 0.004arcmin@xmath0 for orion a and orion b , and @xmath46 0.014arcmin@xmath0 for monr2 by changing the cloud distance by @xmath46 20% from the nominal assumed value .
the corresponding uncertainties at @xmath7 and @xmath8 bands are higher by a factor of @xmath12 - 3 .
the uncertainties are largest for monr2 since it is the most distant cloud analyzed here and the foreground field stars contain a greater fraction of the total field star counts than toward perseus , orion a , or orion b. in addition , since molecular clouds are three dimensional structures , there may be systematic variations in the distances to different parts of the cloud .
for the orion molecular clouds in particular , @xcite argue that the near edge of the orion a and orion b molecular clouds is at a distance of @xmath1320pc , and the far edge is at a distance of @xmath1500pc .
the difference in the field star model between these two distances for a visual extinction of 3@xmath13 is @xmath1 0.006arcmin@xmath0 .
the stellar population associated with the cloud can be inferred by subtracting the field star model from the observed star counts . instead of using the binned star count maps that were produced for the field star polynomial fit , ( e.g. figure [ ksqrt ] ) ,
stellar surface density maps were re - generated for the molecular clouds from the observed star counts using adaptive kernel density estimation @xcite . in this method ,
each star is represented by a kernel function , such as a gaussian , where the width of the kernel depends on the local stellar density as described below .
the sum of the individual kernel functions yields the stellar density map for the cloud .
the advantage of adaptive kernel density estimation compared to a fixed kernel size or a binned star count map is that the resolution of the map varies with the local stellar density which enables compact clusters to be more readily identified .
the kernel width for each star was computed using the procedure outlined in @xcite .
first , a pilot stellar density map , @xmath47 , was constructed using the standard kernel density estimate as @xmath48 where @xmath49 is the bandwidth that controls the amount of smoothing , @xmath6 is the kernel function , and @xmath50 is the number of stars .
the results are not sensitive to the choice of the kernel function , and a radially symmetric gaussian kernel was adopted such that @xmath51 for the pilot density estimate , the bandwidth , @xmath49 , was fixed at the angular size corresponding to 0.25 pc in anticipation that clusters and their associated dense molecular cores have typical sizes of @xmath11pc . a local bandwidth , @xmath52 , was then computed for each star as @xmath53^{-\alpha},\ ] ] where @xmath54 is the geometric mean of @xmath55 . following @xcite ,
@xmath4 was set to be 0.5 .
thus in regions of low stellar surface density such that @xmath56 is small , the local bandwidth , @xmath57 , is relatively large and the data are heavily smoothed .
conversely , in regions of high density such that @xmath58 is large , @xmath57 is relatively small and the data are smoothed less . the adaptive kernel density estimate was then computed as @xmath59 as for the pilot density estimate , the bandwidth , @xmath49 , was chosen to be 0.25 pc . the expected number of field stars for each pixel in the map ( see section [ model ] )
was then subtracted from the adaptive kernel surface density image to obtain the stellar population associated with the cloud .
@xmath9 band surface density maps for the perseus , orion a , orion b , and monr2 molecular clouds are presented in figures [ perseus_kernel][monr2_kernel ] along with corresponding _ iras _ 60 images , @xmath11co(10 ) integrated intensity maps @xcite , and an image of the average @xmath22 color for points sources in the _
2mass _ second incremental release .
based on the extent of the molecular clouds indicated by the @xmath11co images , the _ 2mass _ observations at the time of this study were not available for the southern portion of the perseus molecular cloud and the northern third of the orion b molecular cloud .
further , the orion a @xmath11co map from @xcite does not encompass the full extent of the molecular emission near @xmath60 ( see nagahama et al .
1998 ) .
stellar clusters were identified by forming closed contours at the @xmath61 level above the local background level in the field star subtracted density maps .
the noise in the star counts , @xmath41 , was determined by assuming poisson statistics for the field stars at the appropriate latitude and longitude .
clusters are defined as closed contours that have a peak stellar surface density @xmath62 and a total number of stars within the closed contour @xmath63 with respect to the expected field star population .
these thresholds are arbitrary , and were selected using lines of sight away from the molecular cloud as a control field that should be dominated by field star population and contain few if any clusters .
while the 6@xmath41 peak density requirement biases the algorithm from finding extended , low density clusters , the resulting list of the clusters should be fairly reliable .
the clusters identified in the field star subtracted density map are listed in table [ tbl : clusters ] along with the number of cluster members , the angular area within the 2@xmath41 contour , and any associations with known star forming regions .
all of the identified clusters are located at least partially in projection against the molecular cloud boundaries .
three of the clusters in orion a are outside the spatial extent of the @xmath11co image .
as discussed further below , lower limits to the cluster membership are provided for a few clusters since part of the cluster area has been masked in the _
2mass _ point source catalog due to neighboring bright stars .
the clusters are labeled in the surface density maps in figures [ perseus_kernel][monr2_kernel ] , and contour maps of the @xmath9 stellar surface for each cluster are presented in figure [ contours ] .
the major axis of the cluster measured at the 2@xmath41 contour level ranges from @xmath11 pc to @xmath116 pc for the orion nebula cluster ( onc ) .
@xmath64 vs. @xmath65 color - color diagrams for stars within the projected area of each cluster are shown in figure [ jhhk ] .
each cluster contains a number of stars with red near - infrared colors , supporting the notion that these clusters are indeed embedded within the molecular clouds .
most of the clusters shown in figure [ contours ] have been previously identified and studied . in the perseus molecular cloud
, the ic348 cluster has been the subject of several photometric and spectroscopic investigations @xcite . in the orion a molecular cloud ,
the onc @xcite is by the far largest , richest cluster identified here and extends for @xmath12 ( 16pc ) along a filament of molecular gas prevalent in molecular line @xcite and dust continuum @xcite images .
the boundaries of the onc as defined here include the l1641n cluster @xcite and a couple of stellar density enhancements associated with _ iras _
point sources @xcite . in the orion b molecular cloud , ngc2024
@xcite and ngc2068 @xcite are identified as clusters , although ngc2068 is at the edge of the density map and has not yet been completely imaged by _
2mass_. finally , the major cluster identified in the monr2 molecular cloud is the monr2 cluster itself @xcite .
the number of cluster members listed in table [ tbl : clusters ] are lower limits to the total cluster membership in these clouds . as with any survey ,
the completeness of the cluster membership is limited by the ability to resolve stars in crowded regions , the sensitivity of the observations to stars of various ages and masses , the high extinction in many clusters that may obscure some stars , and the difficulties in detecting point sources against bright nebular backgrounds that often accompany young star forming regions .
further , as a feature of the _
2mass _ second incremental release point source catalog , part of the cluster area in ic348 , l1641 c , the onc , and ngc2024 have been masked out due to the presence of nearby bright stars ( see figure [ contours ] ) .
the fraction of the cluster area masked out in the ic348 , l1641 c , and onc clusters is estimated to be @xmath110% , 5% , and 5% respectively .
the masked region in the onc includes the dense trapezium region that contains @xmath1600 hundred stars brighter than @xmath6=14@xmath13 within the central @xmath21 of the cluster @xcite . to assess the fraction of the cluster population masked out in ngc2024 , the _
2mass _ observations were compared with the ngc2024 @xmath6 band survey by @xcite , which has comparable resolution and sensitivity as the _
2mass _ observations .
they identified 309 stars within the cluster boundaries , compared to @xmath3201 stars found here .
the density and molecular cloud maps presented in figures [ perseus_kernel]-[monr2_kernel ] show that the _ 2mass _ observations at the time of this study were not complete toward the perseus and orion b molecular clouds .
the prominent star forming regions contained within these unscanned areas include the ngc1333 cluster in perseus @xcite , the ngc2071 cluster in orion b @xcite , and the remaining part of ngc2068 in orion b. previous observations with comparable sensitivity and resolution as the _
2mass_data suggest that the ngc1333 cluster contains @xmath194 stars @xcite and ngc2071 @xmath1105 stars @xcite .
the ngc2068 cluster contains @xmath1147 stars in addition to those already found identified @xcite . since the smallest cluster identified with the _
2mass _ data contains 15 stars , each of these regions would almost certainly have been identified as a cluster if they have been included in the _
2mass _ second incremental release .
several star forming regions in the orion clouds found in previous studies did not meet the cluster definition adopted here , including ngc2023 , hh34 , v380 ori , the cohen - kuhi group , kms 35 , kms 36 , and additional low luminosity _ iras _ sources @xcite .
these star forming regions contains between 5 and 34 stars each , which is near the limit of the smallest clusters define in this study . the total cluster membership in orion a and orion
b would not change appreciably if these regions were formally defined as clusters in this study .
given the high extinction that is generally associated with star forming regions , deeply embedded , rich clusters ( @xmath17100 stars ) could exist that remain undetected at near - infrared wavelengths .
any such clusters should appear in the _ iras _ 60 image which can probe deeper into molecular clouds than @xmath9 observations .
in perseus , the brightest _ iras _
sources besides ic348 and ngc1333 are associated with l1448 , l1455 , and a ring of bright far - infrared emission southwest of ic348 that is possibly related to a h ii region @xcite .
the luminosities of these _ iras _
point sources are @xmath520l@xmath66 @xcite , which is too low to contain a significant cluster of young stars unless the stellar masses are strongly biased toward low mass objects . in orion a and orion b , each of the bright far - infrared emitting regions corresponds to a stellar cluster and therefore it is unlikely that a rich cluster remains unidentified in these clouds . in monr2 , three bright _ iras _
sources , corresponding to reflection nebula vdb 70 , vdb 72 ( ngc2182 ) , and vdb 74 @xcite , have no apparent clusters in the _ 2mass _ data . the relatively blue stellar colors and the weak @xmath11co emission indicate that the molecular gas associated with these nebula is diffuse and can not contain a deeply embedded cluster . two additional bright _ iras _ sources ( 06124 - 0621 and 06128 - 0624 ) are present in the northeast corner of the monr2 molecular cloud that are coincident with moderately bright @xmath11co emission .
part of the _
2mass _ point source data in this region has been masked due to a neighboring bright star .
visual inspection of the _
2mass _ image atlas suggests that _ iras _ 06128 - 0624 may be associated with a cluster of @xmath120 stars , but the low far - infrared luminosity of both sources ( @xmath530l@xmath66 ) suggests that neither _ iras _ source is associated with a rich cluster .
therefore , it it seems likely that the richest clusters in each of these molecular clouds have been identified from _
2mass _ and published observations . in summary , the total number of cluster members identified in the _
2mass _ data after field star subtraction is @xmath3299 , @xmath31992 , 246 , and 543 for perseus , orion a , orion b , and monr2 , respectively . if we include observations from the literature to take into account clusters not included in the _
2mass _ second incremental release ( ngc1333 and ngc2071 ) , clusters that have been only partially mapped with _
2mass _ so far ( ngc2068 ) , and clusters that have been partially masked because of neighboring bright stars ( ngc2024 ) , the total number of stars in clusters increases to 393 for perseus and 606 for orion b. corrections to the total stellar population in ic348 , l1641 c , and the onc due to masked out regions are not accurately known and have not been applied .
the number of clusters ranges from 2 in perseus to 7 in orion a , and at least half of the cluster stellar population in each cloud can be attributed to a single cluster .
the fraction of the molecular cloud surface area as defined by @xmath11co(10 ) emission occupied by the clusters ( including the cluster properties incorporated from the literature ) is @xmath13% , 12% , 7% , 3% in perseus , orion a , orion b , and monr2 respectively .
stars associated with the molecular cloud but located outside the boundaries of the stellar clusters are defined as the distributed stellar population .
the distributed population may include stars that formed in isolation , stars that formed in clusters in the past but have since dispersed into the molecular cloud , small groups of stars not meeting the adopted surface density requirements to be identified as clusters , and stars that are associated with identified clusters but lie beyond the adopted cluster boundary . the angular extent of the distributed population is unknown , so it is assumed that the stars may be located anywhere where @xmath11co(10 ) emission is detected as shown in figures [ perseus_kernel]-[monr2_kernel ] .
the signature of a distributed population is a positive average stellar surface density toward lines of sight outside the cluster boundaries but within the molecular cloud after subtracting the field star contribution to the observed star counts .
histograms of the field star subtracted , @xmath7 , @xmath8 , and @xmath9 stellar surface densities toward the perseus , orion a , orion b , and monr2 molecular clouds are shown in figures [ perseus_hist]-[monr2_hist ] for the nominal extinction model .
the open histograms are for all lines of sight toward the molecular clouds , and the hatched histograms are for lines of sight within the cloud that do not intercept a cluster .
each of the histograms are similar in that they peak at a surface density of @xmath10arcmin@xmath0 with a tail toward higher values .
the high surface density tails in the histograms are generally associated with clusters , although the @xmath7 and @xmath8 histograms for orion a do contain a few lines of sight with high surface densities ( @xmath67arcmin@xmath0 ) that have been assigned to the distributed population .
these lines of sight are regions around bright stars in the onc that were masked out at @xmath9 band where the cluster area was defined , but not at @xmath7 and @xmath8 bands where a smaller region was masked out . for each cloud ,
the histograms extend to negative surface densities .
negative surface densities for individual pixels are expected since poisson fluctuations in the star counts implies that the number of field stars will be less than the expected average field star surface density for some lines of sight , and exceed the field star model for others .
the number of lines of sight with a negative surface density and the width of the histograms in figures [ perseus_hist]-[monr2_hist ] depend on the kernel size used to generate the surface density maps and the surface density of the field star population . since the nominal kernel width is set to a constant linear size of 0.25pc , the angular size of the kernel will be smallest for the more distant cloud , monr2 , and largest for the nearest cloud , perseus .
the perseus star count map will then be more heavily smoothed and have smaller fluctuations in the observed star counts about the model field star population than monr2 , as observed . the mean surface density of the distributed population was obtained by averaging the field star subtracted surface density map for regions outside the cluster boundaries but within the @xmath11co map area as represented graphically by the hatched histograms in figures [ perseus_hist]-[monr2_hist ] .
these mean surface densities are summarized in table [ tbl : distributed ] for each band and the three extinction models .
the formal 1@xmath41 uncertainties in the mean stellar surface densities computed from the distribution of points in the hatched histogram are @xmath10.005arcmin@xmath0 in each band for perseus , orion a , and orion b , and @xmath10.009arcmin@xmath0 for monr2 . by comparison ,
the range in the surface densities due to the uncertainties in the extinction model are @xmath1@xmath46 0.10 , 0.05 , and 0.03arcmin@xmath0 at @xmath7 , @xmath8 , and @xmath9 respectively ( see table [ tbl : distributed ] ) . as mentioned in section [ model ] , however , the uncertainties due to the extinction model are not formal 1@xmath41 confidence intervals but represent the plausible range of values based upon converting the @xmath11co integrated intensity into visual extinction .
thus compared to the other field star model uncertainties discussed in section [ model ] , the largest source of uncertainty in the mean stellar surface density for the distributed population is converting the @xmath11co integrated intensity to visual extinction . the range of possible surface densities for the distributed population from the three extinction models is shown graphically in figures [ perseus_cuts]-[monr2_cuts ] .
these figures show the field star subtracted surface density averaged over galactic latitude for each cloud as a function of galactic longitude for the nominal extinction model .
the surface density of the distributed population for the nominal extinction model is shown by the solid horizontal line , and the dashed lines show the corresponding surface densities derived from the low and high extinction models .
the top panel in each figure shows the average visual extinction also averaged over galactic latitude .
the narrow peaks in the surface density at discrete galactic latitudes represent the compact stellar clusters , although the large angular extent of the orion nebula cluster is clearly visible .
figures [ perseus_cuts]-[monr2_cuts ] indicate that within a given cloud , the @xmath7 , @xmath8 , and @xmath9 data give a consistent value for the distributed population surface density within the uncertainties of the field star subtraction .
the most stringent constraints on the distributed population is provided at @xmath9 band , which is the least sensitive of the three bands to the assumed extinction model .
therefore , the characteristics of the distributed population discussed below will be based primarily on the @xmath9 band data .
table [ tbl : distributed ] indicates that for the nominal extinction model , the perseus , orion a , and monr2 molecular clouds contain a positive mean stellar surface density that is indicative of a distributed population , while orion b has a negative surface density .
for the low extinction model , the perseus molecular cloud has also a negative implied surface density for the distribution population .
a negative mean surface density for the distributed population is unphysical and indicates that too many field stars were subtracted from the observed star counts for that particular extinction model .
the number of stars contained in the distributed population over the entire cloud as mapped in @xmath11co was computed by multiplying the average surface density by the cloud area ( see table [ tbl : clouds ] ) , including those regions not encompassed by the _
2mass_second incremental release .
the magnitude of the distributed population inferred from the @xmath9 band observations ranges from zero stars in orion b to @xmath1730 in orion a. however , given the uncertainty in the extinction , a distributed population consisting of @xmath1790 stars can not be ruled out even in orion b. these estimates for the number of stars in the distributed population refer only to stars that fall within the adopted magnitude thresholds .
the fraction of the distributed stellar population that may exist but have magnitudes outside these thresholds is considered in section [ dis : sensitivity ] .
the surface density of the distributed population can be compared with previous observations of the orion a and orion b molecular clouds .
@xcite found that the surface density for the distributed population in orion b of @xmath10.007arcmin@xmath0 for stars with @xmath68@xmath13 over a 0.7 deg@xmath26 region .
@xcite observed a 0.37 deg@xmath26 region away from known clusters at @xmath7 , @xmath8 , and @xmath6 band in orion b , and inferred that the surface density of stars in the distributed population with a near - infrared excess in the @xmath64 vs. @xmath69 color - color diagram is @xmath2 0.014arcmin@xmath0 .
both of these results are consistent with values obtained here for orion b within the uncertainties of the field star subtraction .
@xcite analyzed the stellar population in orion a based on a near - infrared imaging survey encompassing a total of 0.77 deg@xmath26 of the cloud area .
in addition to identifying several clusters , they measured a distributed population of @xmath11500 stars down to a 5@xmath41 detection limit of @xmath70@xmath13 , implying a surface density for the distributed population of @xmath10.54 starsarcmin@xmath0 . by comparison , the maximum surface density for the distributed population in orion a implied by the _
2mass _ data is 0.099 starsarcmin@xmath0 .
the discrepancy between the @xcite results and those obtained here are unclear .
while @xcite analyzed their data at a @xmath7 band magnitude limit 0.7@xmath13 fainter than adopted here , figures 9 and 11 in that paper suggests that most of the distributed population they found should have @xmath7 band magnitudes brighter than 16@xmath13 and be detectable with the magnitudes limits adopted here .
given that the _
@xmath7 , @xmath8 , and @xmath9 star counts analyzed here are based on more extensive observations and provide consistent results , we will use these data to characterize the distributed population in orion a.
a complete picture of the star formation history of molecular clouds requires extensive spectroscopic and photometric observations of the cluster and distributed populations .
constraints on the star formation history of molecular clouds can be obtained though by using the star count results from the previous section and examining the relative contribution of clusters and distributed populations to the total stellar population .
the ages of several clusters in nearby molecular clouds have been inferred by constructing hr diagrams and comparing the stellar effective temperatures and luminosities with pre - main - sequence evolutionary tracks , including @xmath71 ophiuchus @xcite , ic348 @xcite , the orion nebula cluster @xcite , ngc2024 @xcite , l1641 n @xcite , and l1641 south @xcite .
a literal interpretation of the hr diagrams indicate that most of the stars in these clusters have formed within the last 1 - 2 myr , although some stars may be as old as 10 myr . by comparison
, @xmath72co observations of ob associations @xcite and the observed fraction of molecular clouds with current star formation suggest that the cloud lifetime is between 10 myr and 100 myr @xcite .
even longer lifetimes though are required in coagulation models for cloud formation @xcite .
relative to the expected lifetime of molecular clouds then , embedded clusters represent a recent `` burst '' of star formation .
a distributed stellar population can not be interpreted unambiguously , as it can represent either older clusters that have since dispersed or stars of all ages that formed in isolation throughout the cloud .
the relative number of the total stellar population ( cluster plus distributed ) currently contained in clusters then places a lower limit on the fraction of stars that have formed in the recent history of a molecular cloud . before examining the implications of the star count results for the star formation history of the molecular clouds studied here ,
it is important to recognize the sensitivity of the _
2mass _ observations to stars of various masses and ages .
these dependencies are examined in figure [ sensitivity ] , which shows @xmath9 band iso - magnitude curves for stars as a function of stellar age ( @xmath73 years ) and stellar mass ( @xmath74m@xmath66 ) at the distance of perseus , orion a / b , and monr2 .
the magnitudes were computed using the pre - main - sequence evolutionary tracks from dantona & mazzitelli ( 1997,98 ) assuming zero extinction and no near - infrared excess as outlined in @xcite .
only the @xmath9 band sensitivity limits are considered here since , as shown in section [ distributed ] , the @xmath9 band star counts provide the tightest constraint on the surface density of the distribution population .
figure [ sensitivity ] shows , for example , that at the distance of the perseus molecular cloud , 0.08m@xmath66 stars are brighter than the adopted @xmath9 band magnitude threshold for ages @xmath510 myr .
( young brown dwarfs with masses @xmath2 0.08m@xmath66 , if present , are also detected at these distances for young ages .
the shape of the initial mass function for these objects remains uncertain at this time , however , and these low masses are not considered here . ) at older ages , low mass pre - main - sequence stars become systematically fainter than the sensitivity limits of the survey and would no longer be detectable . assuming that clouds have ages
@xmath5 100 myr , figure [ sensitivity ] shows that all stars more massive than @xmath10.3m@xmath66 on the surface of the perseus molecular clouds will be detectable , with corresponding limits of @xmath10.5m@xmath66 for orion a and orion b and @xmath10.7m@xmath66 for monr2 .
the fraction of the total stellar population that is detectable with _ 2mass _ depends on the mass and age distribution of the constituent stars .
these distributions are crudely known for only a handful of clusters and are completely unknown for the distributed population .
therefore , we have assumed that the stars have been forming at a constant rate in time throughout the cloud with a miller - scalo initial mass function @xcite between 0.08 and 10m@xmath66 .
the magnitudes for this model stellar population were calculated using the pre - main - sequence tracks up to 3m@xmath66 as described above .
stars more massive than 3m@xmath66 are not included in the @xcite models , but since their pre - main - sequence phase lasts for only @xmath52 myr @xcite , the magnitudes of these stars as they appear on the zero age main sequence were adopted for all ages .
the average visual extinction computed from the @xmath11co maps ( 3.0@xmath13 , 3.2@xmath13 , and 2.6@xmath13 for perseus , orion a / b , and monr2 respectively ) were applied to the computed magnitudes .
figure [ fraction ] shows the fraction of the model stellar population that has apparent @xmath9 magnitudes between 6.0@xmath13 and 14.3@xmath13 as a function of the time .
while the model is intended mainly to describe a distributed population forming over an extended period of time , the results can be applied to clusters as well .
this figure indicates that for clusters , which have typical ages @xmath52 myr as discussed above , @xmath1780% of the cluster members with stellar masses
@xmath16 0.08m@xmath66 are detectable with _ 2mass _ for each of the molecular clouds studied here .
( note , however , that clusters often have high extinction assumed for this model . ) on the other hand , for a distributed population that has been forming stars for 50 myr , 67% of the stars would be detectable in perseus , 45% in orion a and orion b , and only 26% in monr2 .
this model demonstrates that if the distributed population has been forming for over tens of million of years , an appreciable fraction of the stars would no longer be detectable within the adopted magnitude thresholds .
given the above calculations , the observed surface density can be used to place constraints on the time averaged star formation efficiency in molecular clouds , where the star formation efficiency is defined as the fraction of the cloud mass that has been converted into stars .
the predicted surface density of stars can be computed using the cloud masses and angular areas summarized in table [ tbl : clouds ] , and assuming that the stars have been forming at a constant rate in time with a miller - scalo initial mass function between 0.08m@xmath66 and 10m@xmath66 .
the solid curves in figure [ sfe ] show the predicted @xmath9 surface density for stars with magnitudes of 6.0@xmath13 @xmath75@xmath13 for the nominal extinction model assuming star formation efficiencies of 1% , 2% , 5% , and 8% .
the dashed lines in figure [ sfe ] show the observed stellar surface density for the distributed population only .
similarly , the dotted lines show the surface density of the clusters only , averaged over the entire cloud area in order to determine how efficient mass is being converted into clusters on a global scale .
the star formation efficiencies in the dense cores that the clusters formed in will naturally be higher . as with figure [ fraction ] , the older ages are intended to apply only to the distributed population and not the clusters .
figure [ sfe ] indicates that the star formation efficiency implied by the distributed population alone is @xmath21% for perseus , 1 - 5% for orion a , 0% for orion b , and 1 - 4% for monr2 , where the range is due to the uncertain ages of the distributed population .
the star formation efficiency implied by the clusters only is @xmath12% in perseus , 4% in orion a , 1% in orion b , and 1% in monr2 . the total star formation efficiency implied by the sum of the cluster and distribution populations then ranges from a low of @xmath11% in orion b to a high of 9% in orion a if the distribution population is as old as 100 myr .
since the observed ages of clusters is @xmath52 myr , the fraction of the total stellar population contained in clusters alone reflects the relative contributions of recent star formation activity compared to isolated star forming regions and older stars that constitute the distributed population .
the fraction of the detected stellar population contained in clusters using the nominal extinction model is 80% , 73% , 100% , and 56% for perseus , orion a , orion b , and monr2 respectively , where the contributions to the stellar population from previously known clusters not contained in the _
2mass _ second incremental release have been included ( see section [ clusters ] ) .
the cluster fraction may be as high as 100% , 87% , 100% , and 74% for these clouds if the low extinction model is adopted , and as low as 34% , 59% , 39% , and 41% with the high extinction model . as emphasized above
, however , an appreciable fraction of the distributed population may not have been detected if the stellar population has been forming for more than several million years .
thus the observed fraction of stars in clusters may underestimate the contribution from the distributed population .
the number of stars in the distributed population that are too faint to detect with the present observations was estimated in figure [ fraction ] assuming a constant star formation rate and a miller - scalo imf . applying these assumptions and model calculations , figure [ clusterfraction ]
shows the implied fraction of the total stellar population contained in clusters as a function of age for the distributed population .
the solid curve shows the fraction of stars in clusters for the nominal extinction model , and the dotted curves represent the cluster fraction for the low and high extinction models . after taking into account the ages of the distributed population , the fraction of stars contained in clusters under the nominal extinction model is @xmath1750% in perseus , orion a , and orion b for ages less than 100 myr , and @xmath1725% for monr2 . even for the high extinction model and an old age ( 100 myr )
, the fraction of stars in clusters is still between 17 - 41% for the four clouds .
thus in each of these molecular clouds , a substantial fraction of the total stellar population , if not the majority , is contained in young stellar clusters .
the substantial fraction of stars found in clusters regardless of the age of the distributed population is surprising given the canonical assumed lifetimes of molecular clouds .
as discussed above , embedded , dense clusters typically have ages @xmath52 myr .
the fact that no substantially older clusters are found within molecular clouds indicates that these older clusters have dispersed into the molecular cloud , have destroyed the surrounding molecular gas and no longer appear associated with a cloud , or these clouds are just now forming their first clusters .
if we assume a constant star formation rate and that clusters simply disperse after a few million years , then the fraction of stars contained in clusters should decrease as @xmath76 , where @xmath77 is the age of the molecular cloud . with these assumptions and assuming a cloud lifetime of 50 myr just as an example , the fraction of stars in clusters should be @xmath14% .
by contrast , the observed fraction of stars in clusters for the clouds analyzed here is substantially higher as discussed above , and would imply a cloud lifetime of @xmath57 myr under the constant star formation rate assumption even given the uncertainties in the field star subtraction .
thus in each of the clouds , the distributed population contains fewer stars than expected if star formation has been occurring in these regions at a constant rate for tens of millions of years .
this situation is reminiscent of the post - t tauri star problem in the taurus molecular cloud , where most stars within the confines of the molecular gas have ages @xmath53 myr and few stars have been identified that are older than 10 myr @xcite .
various hypotheses have been proposed to explain the lack of older stars in taurus and other nearby molecular clouds , including the kinematic dispersal of older stars from the molecular cloud boundaries , imposing a star formation rate that changes with time , and postulating that molecular clouds are inherently short lived .
we briefly discuss these possibilities as they pertain to the perseus , orion a , orion b , and monr2 molecular clouds .
the small number of old stars that have been identified in nearby molecular clouds has led to the suggestion that a substantial fraction of the stellar population has dispersed from the molecular gas and that the current stellar census is incomplete @xcite . assuming a one dimensional velocity of 1 kms@xmath35
, a star could drift as far as 10 pc ( 1.2 at the distance of orion a and orion b ) in 10 myr .
however , molecular clouds are likely gravitationally bound structures , and presumably the kinematics of the newly formed stars reflect that of the molecular gas . since the star formation efficiency implied by the star counts is @xmath59% , the dynamical evolution of the embedded stars will be dominated by the gas as long as the cloud remains intact . therefore
, many , if not most , of the stars must remain within the cloud .
nevertheless , the size of any dispersed stellar population can by gauged from h@xmath4 and x - ray surveys of regions surrounding the cloud .
this has been in fact done for the orion a and orion b molecular clouds , but the results remain inconclusive . both h@xmath4 @xcite and x - ray @xcite surveys of this region find a source density of @xmath11 star deg@xmath0 in regions surrounding the orion molecular clouds and ob association .
up to one - third of the x - ray sources may be unrelated foreground stars @xcite perhaps related to gould s belt @xcite , but in any event , the surface density of sources is too low to contribute significantly to the total population in the orion a and orion b molecular clouds relative to the clusters .
h@xmath4 emission is likely not sensitive to the majority of the older stars likely to constitute a old dispersed population , though , and at the distance of the orion clouds , the rosat all sky survey is able to detect only to the more luminous x - ray t tauri stars @xcite . in more nearby clouds , however , follow up studies of the widely distributed rosat x - ray sources indicate that many these objects likely do not represent a dispersed population , but rather solar - type zero age main sequence stars or older pre - main - sequence stars unrelated to the existing molecular clouds ( magazz et al.1997 ; see also briceo et al .
1997 , favata , micela , & sciortino 1997 ) .
further , a large scale x - ray and objective - prism survey of taurus failed to find any new t tauri stars older than 5 myr @xcite .
if most of the stars that have formed over the lifetime of the perseus , orion a , orion b , and monr2 molecular clouds are still associated with the molecular gas , then the high fraction of the total stellar population contained in clusters implies that either the star formation rate is currently higher than it has been in the past @xcite , or that molecular clouds have relatively young ages @xcite . @xcite
re - analyzed published hr diagrams for a number of nearby star forming regions with a single set of pre - main - sequence models and argued that star formation over entire molecular clouds ( e.g. taurus and chamaeleon i ) and individual clusters ( e.g. the orion nebula cluster ) started at a low rate @xmath17 10 myr ago and has increased dramatically within the past 1 - 3 myr . in their scenario ,
molecular clouds remain relatively dormant for much of their cloud lifetime since individual dense cores are supported against gravitational collapse by magnetic fields , and the time scale to dissipate the magnetic support is on the order of @xmath110 myr ( see also palla & galli 1997 ) .
@xcite , on the other hand , suggested that the lack of old stars in taurus this is a consequence of the molecular filaments that make up the cloud having formed only within the past few million years in turbulent flows in the interstellar medium . in this scenario , taurus is an intrinsically young cloud that has not had time to form old stars ( see also elmegreen 2000 ) .
the difference between the two pictures for the evolution of molecular clouds has important consequences for molecular cloud evolution . in the @xcite scenario ,
molecular clouds are in a quasi - static state for most of their lifetime , while the @xcite model suggest clouds form and evolve dynamically in the interstellar medium .
if the @xcite model is correct , most molecular clouds should not contain a substantial stellar population since they are relatively dormant for most of their cloud lifetimes .
the difficulty with this scenario is that it would imply that we are observing a number of molecular clouds in the solar neighborhood just as they are forming a substantial fraction of their total stellar population in young clusters .
this would include not only the 4 molecular clouds studied here , but also chamaeleon i , lupus , and taurus .
this is not necessarily an observational bias in that these clouds are frequently studied because they have active star formation , as star formation and stellar clusters appear to be ubiquitous in molecular clouds .
more than half of the clouds within 200 pc of the sun are forming an appreciable number of t tauri stars ( feigelson 1996 and references therein ) , and most giant molecular clouds within 3 kpc of the sun are forming ob stars and presumably an accompanied clusters of lower mass stars @xcite . in this respect , imposing a short lifetime for molecular clouds would qualitatively account for the lack of a substantial older stellar population in molecular clouds without the need to impose a higher than average star formation rate at the current time @xcite .
a more systematic survey of a volume limited sample of molecular clouds seems warranted though to better establish the frequency of clusters within molecular clouds as a more comprehensive observational test of these two pictures of cloud evolution .
we investigate the spatial distribution of young stars associated with the perseus , orion a , orion b , and monr2 molecular clouds using @xmath7 , @xmath8 , and @xmath9 star counts from the _
2mass _ second incremental release .
the stellar population associated within these clouds is determined statistically by subtracting a semi - empirical model for the field star contamination from the observed stellar surface density maps .
the field star model is constructed using _
2mass _ observations of the regions surrounding the clouds to measure the total field star surface density , published @xmath11co(10 ) maps to determine the cloud extinction , and the @xcite galactic star count model to estimate the fraction of field stars that are background to the clouds .
the stellar population is categorized into compact stellar clusters and a distributed stellar population .
individual stellar clusters are identified as peaks in the field star subtracted , stellar surface density maps , and the distributed stellar population is defined as any excess star counts outside the cluster boundaries but within the molecular cloud as traced by @xmath11co(10 ) emission . the number of clusters within the studied molecular clouds , including clusters identified in the literature in regions not yet observed with _
2mass _ , varies from 2 in perseus to 7 in orion a. the total number of stars contained within the clusters is @xmath3393 in perseus , @xmath31992 in orion a , 606 in orion b , and 543 in monr2 .
more than half of the cluster members in each cloud is contained in just a single rich cluster .
a distributed stellar population is detected in the orion a and monr2 molecular clouds within the uncertainties of the fieldstar subtraction at a @xmath9 band surface density between 0.013arcmin@xmath0 and 0.084arcmin@xmath0 .
model calculations suggest , however , that the surface density of the distributed population in the clouds may be underestimated by as much as a factor of three if the stars have been forming with a miller - scalo initial mass function between 0.08m@xmath66 and 10m@xmath66 at a constant star formation rate for up to 100 myr . after taking into consideration the possible evolutionary status of the distributed population ,
the star formation efficiencies implied by the sum of cluster and distributed populations varies between @xmath11 - 9% for the four clouds .
the percentage of the total observed stellar population contained in clusters alone is 80% , 73% , 100% , and 56% for perseus , orion a , orion b , and monr2 respectively for the nominal extinction model , but may be up to a factor of two lower depending again on the age of the distributed population .
nonetheless , the high fraction of stars currently contained in clusters is surprising given that embedded clusters typically have ages ( @xmath52 myr ) that are substantially younger than that often assumed for molecular clouds ( 10 - 100 myr ) .
these results indicate that either each of these molecular clouds have been forming stars at a higher rate in the last couple of million years than in the prior history , or that the ages of molecular clouds are younger than generally assumed .
jmc would like to thank lynne hillenbrand for numerous discussions and comments on this work .
he is also grateful to john bally for providing the @xmath11co molecular line maps .
this publication makes use of data products from the two micron all sky survey , which is a joint project of the university of massachusetts and the infrared processing and analysis center , funded by the national aeronautics and space administration and the national science foundation .
jmc acknowledges support long term space astrophysics grant nag5 - 8217 and the owens valley radio observatory , which is supported by the national science foundation through nsf grant ast 9981546 .
alcal , j. m. , chavarra - k .
c. , & terranegra , l. 1998 , , 330 , 1017 alcal , j. m. , covino , e. , torres , g. , sterzik , m. f. , pfeiffer , m. j. , & neuhuser , r. 2000 , , 353 , 186 alcal , j. m. , terranegra , l. , wichmann , r. , chavarra - k .
c. , krautter , j. , schmitt , j. h. m. m. , moreno - corral , m. a. , de lara , e. , & wagner , r. m. 1996 , , 119 , 7 ali , b. , & depoy , d. l. 1995 , , 109 , 709 allen , l. e. 1996 , ph.d .
thesis , university of massachusetts allen , l. e. , & hillenbrand , l. a. 2000 , to appear in `` proceedings of the ringberg castle conference on orion '' , pasp conference proceedings , ed .
m. mccaughrean , in press andersson , b - g , wannier , p. g. , moriarty - schieven , g. h. , & bakker , e. j. 2000 , , 119 , 1325 aspin , c. , & sandell , g. , 1997 , , 289 , 1 aspin , c. , sandell , g. , & russell , a. p. g. 1994 , , 106 , 165 aspin , c. , & walther , d. m. 1990 , , 235 , 387 bachiller , r. , & cernicharo , j. 1986 , , 166 , 283 ballesteros - paredes , j. , hartmann , l. , & vzquez - semadeni , e. 1999 , , 527 , 285 bally , j. , stark , a. a. , wilson , r. w. , & langer , w. d. 1987 , , 312 , l45 barsony , m. , kenyon , s. j. , lada , e. a. , & teuben , p. j. 1997 , , 112 , 109 bash , f. n. , green , e. , & peters , w. l. iii 1977 , , 217 , 464 beichman , c. a. , myers , p. c. , emerson , j. p. , harris , s. , mathieu , p. , benson , p. j. , & jennings , r. e. 1986 , , 307 , 337 bessell , m. s. , & brett , j. m. 1988 , , 100 , 1134 blitz , l. 1991 , in _ the physics of star formation _ , eds . c. j. lada & n. kylafis , ( dordrecht : kluwer ) , 3 briceo , c. , calvet , n. , kenyon , s. , & hartmann , l. 1999 , , 118 , 1354 briceo , c. , hartmann , l. w. , stauffer , j. r. , gagn , m. , stern , r. , & caillault , j - p .
1997 , , 113 , 740 brown , a. g. a. , de geus , e. j. , & de zeeuw , p. t. 1994 , , 289 , 101 cambrsy , l. 1999 , , 345 , 965 cambrsy , l. , copet , e. , epchtein , n. , de batz , b. , borsenberger , j. , fouqu , kimeswenger , s. , tiphne , d. 1998 , , 338 , 977 carpenter , j. m. , meyer , m. r. , dougados , c. , strom , s. e. , & hillenbrand , l. a. 1997 , , 114 , 198 chen , h. , & tokunaga , a. t. 1994 , , 90 , 149 chen , h. , tokunaga , a. t. , strom , k. m. , & hodapp , k. w. 1993 , , 407 , 639 clarke , c. j. , bonnell , i. a. , & hillenbrand , l. a. 2000 , in protostars and planets iv , ed . v. mannings , a. p. boss , and s. s. russell ( tucson : university of arizona press ) , in press cohen , j. g. , grogel , j. a. , perrson , s. e. , & elias , j. h. 1981 , , 249 , 481 comern , f. , rieke , g. h. , & rieke , m. j. 1996 , , 473 , 294 dame , t. m. , ungerechts , h. , cohen , r. s. , de geus , e. j. , grenier , i. a. , may , j. , murphy , d. c. , nyman , l .- a . , & thaddeus , p. 1987
, , 322 , 706 dantona , f. & mazzitelli , i. 1997 , mmsai 68 , 807 .
1998 , private communication de zeeuw , p. t. , hoogerwerf , r. , de bruijne , j. h. j. , brown , a. g. a. , & blaauw , a. 1999 , , 117 , 354 elmegreen , b. g. 1991 , in _ the physics of star formation _ , eds . c. j. lada & n. kylafis , ( dordrecht : kluwer ) , 35 .
2000 , , 530 , 277 favata , f. , micela , g. , & sciortino , s. 1997 , , 326 , 647 feigelson , e. d. 1996 , , 468 , 306 frerking , m. a. , langer , w. d. , & wilson , r. w. 1982 , , 262 , 590 genzel , r. , reid , m.j . , moran , j.m . , & downes , d. 1981 , apj 224 , 884 gomez , m. , hartmann , l. , kenyon , s. j. , & hewett , r. 1994 , , 105 , 1927 greene , t. p. , & meyer , m. r. 1995 , , 450 , 233 guillout , p. , sterzik , m. f. , schmitt , j. h. m. m. , motch , c. , & neuhuser , r. 1998 , , 337 , 113 gyulbudaghian , a. l. , glushkov , yu .
i. , & denisyuk , e. k. 1978 , apj , 224 , l137 hayakawa , t. , mizuno , a. , onishi , t. , yonekura , y. , hara , a. , yamaguchi , r. , & fukui , y. 1999 , , 51 , 919 herbig , g. h. 1978 , in problems of physics and evolution of the universe , ed . l. mirzoyan ( yerevan : acad .
armenian s. s. r. ) , 171 herbig , g. h. 1998 , , 497 , 736 herbst , w. , & racine , r. 1976 , , 81 , 840 hillenbrand , l. a. 1997 , , 113 , 1733 hillenbrand , l.a . , & carpenter , j. m. 2000 , , 540 , 236 hillenbrand , l.a .
, & hartmann , l. w. 1998 , , 492 , 540 hodapp , k. w. , & deane , j. 1993 , , 88 , 119 howard , e. m. , pipher , j. l. , & forrest , w. j. 1994 , , 425 , 707 johnstone , d. , & bally , j. 1999 , , 510 , l49 jones , b. f. , & walker , m. f. 1988 , , 95 , 1755 kwan , j. , & valdes , f. 1983 , , 271 , 604 lada , c. j. , alves , j. , & lada , e. a. 1996 , , 111 , 1964 lada , c. j. , lada , e. a. , clemens , d. p. , & bally , j. 1994 , , 429 , 694 lada , c. j. , young , e. t. , & greene , t. p. 1993
, , 408 , 471 lada , e. a. , depoy , d. l. , evans , n. j. ii , & gatley , i. 1991 , , 371 , 171 lada , e. a. , & lada , c. j. 1995 , , 109 , 1682 ladd , e. f. , lada , e. a. , & myers , p. c. 1993 , , 410 , 168 langer , w. d. , wilson , r. w. , goldsmith , p. f. , & beichman , c. a. 1989 , , 337 , 355 lee , y. , snell , r. l. , & dickman , r. l. 1994 , , 432 , 167 leisawitz , d. 1990 , , 359 , 319 li , w. , evans , n. j. ii , & lada , e. a. 1997 , , 488 , 277 lis , d. c. , serabyn , e. , keene , j. , dowell , c. d. , benford , d. j. , phillips , t. g. , hunter , t. r. , & wang , n. 1998 , , 509 , 299 luhman , k. l. 1999 , , 525 , 466 luhman , k. l. , & rieke , g. h. 1999 , , 525 , 440 luhman , k. l. , rieke , g. h. , lada , c. j. , & lada , e. a. 1998 , , 508 , 347 maddalena , r. j. , moscowitz , j. , thaddeus , p. , & morris , m. 1986 , , 303 , 375 magazz , a. , martn , e. l. , sterzki , m. f. , neuhuser , r. , covino , e. , & alcal , j. m. 1997 , , 124 , 449 mccaughrean , m. j. , & stauffer , j. r. 1994 , , 108 , 1382 meyer , m. r. 1996 , ph.d .
thesis , university of massachusetts meyer , m. r. , & lada , e. a. 2000 , to appear in `` proceedings of the ringberg castle conference on orion '' , pasp conference proceedings , ed .
m. mccaughrean , in press miesch , m. s. , & bally , j. 1994 , , 429 , 645 miller , g. e. , & scalo , j. m. 1979 , , 41 , 513 nagahama , t. , mizuno , a. , ogawa , h. , & fukui , y. 1998 , , 116 , 336 nakono , m. , wiramihardja , s.d . , & kogure , t. 1995 , , 47 , 889 neuhuser , r. , sterzik , m. f. , schmitt , j. h. m. m. , wichmann , r. , & krautter , j. 1995 , , 297 , 391 padoan , p. , bally , j. , billawala , y. , juvela , m. , & nordlund , a. 1999 , , 525 , 318 palla , f. , & galli , f. 1997 , , 476 , l35 palla , f. , & stahler , s. w. 1993 , , 418 , 414 .
, in press phelps , r. l. , & lada , e. a. 1997 , , 477 , 176 piche , f. 1993 , , 105 , 324 prosser , c. f. , stauffer , j. r. , hartman , l. , soderblom , d. r. , jones , b. f. , werner , m. w. , & mccaughrean , m. j. 1994 , , 421 , 517 reach , w. t. , wall , w. f. , & odegard , n. 1998 , , 507 , 507 silverman , b. w. , 1986 , _ density estimation for statistics and data analysis _ , ( london : chapman & hall ) sterzik , m. f. , alcala , j. m. , meuhaeuser , r. , & schmitt , j. h. m. m. 1995 , , 297 , 418 strom , k. m. , kepner , j. , & strom , s. e. 1995 , , 438 , 813 strom , k. m. , strom , s. e. , & merrill , k. m. 1993 , , 412 , 233 tatematsu , k. , umemoto , t. , heyer , m. h. , hirano , n. , kameya , o. , & jaffe , d. t. 1998 , , 118 , 517 van den bergh , s. 1966 , , 71 , 990 wainscoat , r. j. , cohen , m. , volk , k. , walker , h. j. , & schwartz , d. e. 1992 , , 83 , 111 walter , f. m. , brown , a. , mathieu , r. d. , myers , p. c. , & vrba , f. v. 1988 , , 96 , 297 cccccccc@cc perseus & 160 & @xmath78 & 320 & 6.4 & 1.0 & 1.3 & 1.9 & 1 & 4 + orion a & 211 & @xmath44 & 480 & 6.2 & 2.5 & 3.3 & 4.8 & 2 & 5 + orion b & 206 & @xmath79 & 480 & 7.2 & 2.4 & 3.2 & 4.7 & 2 & 6 + monr2 & 214 & @xmath80 & 830 & 2.6 & 2.3 & 3.1 & 4.6 & 3 & 6 + ccc@c@c@rcl + 1 & 160.5044 & -17.8011 & 03:44:37.2 & + 32:09:19 & @xmath3299 & 1.19 & ic 348 + + + 2 & 208.9740 & @xmath8119.3458 & 05:35:20.9 & @xmath8105:20:29 & @xmath31744 & 3.86 & orion nebula cluster + 3 & 210.8003 & @xmath8119.5051 & 05:37:52.9 & @xmath8106:57:09 & 23 & 0.56 & v592 ori + 4 & 210.9800 & @xmath8119.3276 & 05:38:49.6 & @xmath8107:01:32 & 36 & 0.61 & l1641 c + 5 & 212.4710 & @xmath8119.0181 & 05:42:27.1 & @xmath8108:09:12 & 129 & 1.10 & l1641 south + 6 & 212.9816 & @xmath8119.1469 & 05:42:50.1 & @xmath8108:38:37 & 15 & 0.42 & iras 05404 - 0839 + 7 & 214.0556 & @xmath8119.6177 & 05:42:53.5 & @xmath8109:45:38 & 23 & 0.47 & iras 05404 - 0946 + 8 & 214.2706 & @xmath8119.7824 & 05:42:38.4 & @xmath8110:00:51 & 22 & 0.47 & iras 05401 - 1002 + + + 9 & 205.3080 & @xmath8114.2957 & 05:46:47.0 & + 00:06:16 & @xmath345 & 0.72 & ngc 2068 + 10 & 206.5122 & @xmath8116.3719 & 05:41:37.3 & @xmath8101:53:39 & @xmath3201 & 1.01 & ngc 2024 + + + 11 & 213.3381 & @xmath8112.6029 & 06:07:08.1 & @xmath8106:03:53 & 15 & 0.41 & iras 06046 - 0603 + 12 & 213.6955 & @xmath8112.5926 & 06:07:47.8 & @xmath8106:22:20 & 371 & 1.85 & monr2 + 13 & 213.8745 & @xmath8111.8410 & 06:10:49.1 & @xmath8106:11:38 & 134 & 1.13 & ggd 12 - 15 + 14 & 214.1337 & @xmath8111.4173 & 06:12:48.0 & @xmath8106:13:56 & 23 & 0.61 & ggd 17 + c@ccc@ccc perseus + & -0.048 & 0.013 & 0.101 & & 310 & 2300 + & -0.031 & 0.006 & 0.064 & & 150 & 1500 + & -0.014 & 0.004 & 0.034 & & 100 & 780 + + orion a + & -0.054 & 0.009 & 0.099 & & 210 & 2200 + & -0.017 & 0.021 & 0.080 & & 470 & 1800 + & 0.013 & 0.032 & 0.064 & 300 & 730 & 1400 + + orion b + & -0.10 & -0.018 & 0.096 & & & 2500 + & -0.057 & -0.008 & 0.066 & & & 1700 + & -0.033 & -0.009 & 0.031 & & & 790 + + monr2 + & -0.031 & 0.070 & 0.205 & & 660 & 1900 + & 0.001 & 0.056 & 0.136 & 10 & 530 & 1300 + & 0.020 & 0.045 & 0.083 & 190 & 420 & 780 + | we use the _ 2mass _ second incremental release point source catalog to investigate the spatial distribution of young stars in the perseus , orion a , orion b , and monr2 molecular clouds . after subtracting a semi - empirical model of the field star contamination from the observed star counts ,
stellar surface density maps are used to identify compact clusters and any stellar population found more uniformly distributed over the molecular cloud .
each cloud contains between 2 to 7 clusters , with at least half of the cluster population found in a single , rich cluster .
in addition , a distributed stellar population is inferred in the orion a and monr2 molecular clouds within the uncertainties of the field star subtraction with a surface density between 0.013 - 0.083arcmin@xmath0 .
sensitivity calculations suggest , however , that the number of stars in the distributed population may be underestimated by a factor of 2 or more if stars have been forming with a miller - scalo imf at a constant star formation rate for longer than 10 myr . after considering the possible evolutionary status of the distributed population , the global star formation efficiency implied by the sum of the distributed and cluster populations ranges between 1 - 9% among the four clouds .
the fraction of the total stellar population contained in clusters for the nominal extinction model ranges from @xmath150 - 100% if the distributed population is relatively young ( @xmath2 10 myr ) , to @xmath125%-70% if it is relatively old ( @xmath1 100 myr ) .
the relatively high fraction of stars contained in clusters regardless of the age of the distributed population , in conjunction with the young ages generally inferred for embedded clusters in nearby molecular clouds , indicates that a substantial fraction of the total stellar population in these regions has formed within the past few million years in dense clusters .
this suggests that either the star formation rate in each these clouds has recently peaked if one assumes clouds have ages @xmath3 10 myr , or molecular clouds are younger than typically thought if one assumes that the star formation rate has been approximately constant in time . |
the large magellanic cloud ( lmc ) is the nearest actively star - forming galaxy to us , and an invaluable laboratory for the study of stellar and galactic evolution . with a total mass of @xmath810@xmath9 m@xmath10 @xcite ,
the lmc lies just at the borderline between dwarf and giant galaxies , in a regime where the scaling relations of basic galaxy properties ( metallicity , mean stellar age , internal structure ) with mass undergo a fundamental qualitative change @xcite .
additionally , the lmc is deeply affected by the tidal forces stemming from its interactions with the small magellanic cloud and the milky way .
such interactions have been plausibly connected to major events in the lifetime of galaxies , including the creation of bulges , bars , and/or thick disks ; and to starburst activity ( e.g. * ? ? ?
to accurately measure the histories of star formation and chemical evolution of the lmc is a major challenge for astrophysicists .
while general characteristics of its evolution such as the relatively greater number of stars aged a few gigayears compared to the milky way @xcite have been known for decades , the details are only now able to be measured .
for instance , it was not until the construction of the cleanest possible deep color - magnitude diagrams ( cmds ) from the wfpc2 camera aboard the hubble space telescope that it became apparent that the variations in field star formation rate have been largely decoupled from variations in the cluster formation rate @xcite .
even these modern analyses are not without their difficulties , chiefly the extreme disparity in angular size between the lmc ( @xmath1110@xmath12 square arcminutes ) and the wfpc2 field of view ( @xmath7 5 square arcminutes ) .
a major factor limiting the precision of star - formation history ( sfh ) measurements based on deep cmds is the age - metallicity degeneracy the fact that age and metallicity can be played off against one another to recreate closely similar distributions of stars in cmds ( e.g. , * ? ? ?
this is exacerbated by the current lack of knowledge of the lmc s age - metallicity relation ( amr ) .
star clusters , the most obvious tracers of such a relation , are famously scarce in the lmc for ages between 310 gyr @xcite .
this age gap spans over half the age of the universe ; it includes the likely epoch of galactic disk formation around redshift @xmath13 @xmath711.5 , and probably spans four lmc - milky way orbital periods ( e.g. , * ? ? ? * ; * ? ? ?
@xcite draw a connection between the tidal capture of the small magellanic cloud ( smc ) by the lmc and the end of the cluster age gap this capture event had previously been thought unlikely ( e.g. , * ? ? ?
* and references therein ) . for these reasons
we have begun to measure the chemical abundances of the field stars in the lmc : to fill in the cluster age gap , to measure the variation of metallicity with radius across the lmc , and to reliably distinguish the bar , disk , and possible thick disk or halo populations from each other . because the shape of the metallicity distribution function ( mdf ) is not known _ a priori _
, it is important to measure the largest possible sample .
the brightest common stars in the age range from @xmath7114 gyr are red giants ; in the lmc , they have magnitudes i @xmath1414.8 . because high - dispersion spectroscopy of faint red giants is extremely expensive in telescope time
, we rely on spectra of moderate resolution to derive abundances good to @xmath50.10.2 dex by comparison to star clusters of known metallicity .
the near - infrared calcium ii triplet ( @xmath158500 ) is the most widely used such technique .
it has been very successfully applied to lmc star clusters , beginning with a landmark paper by @xcite ( hereafter ossh ) .
the results from ossh have become the standard reference for abundances of lmc clusters , and have been used as the basis for simulations of the lmc s chemical evolution ( e.g. , * ? ? ?
* ) ( hereafter pt98 ) .
the cluster - based amr has in turn been taken as a guide to the amr in derivations of the star formation history based on deep color - magnitude diagrams of the field populations ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
it was not feasible to obtain large samples of field star abundances prior to the advent of efficient multi - object spectrographs , first on 4-meter class telescopes , and more recently at 8-meter class telescopes .
we began to obtain abundances for red giants in two fields of the inner lmc disk using long - slit spectroscopy in ( * ? ? ?
* paper i ) , and expanded the sample six - fold using the hydra multiobject spectrograph at the cerro tololo inter - american observatory 4-meter telescope ( * ? ? ?
* paper ii ) .
that study found the mean metallicity of red giants at a radius of roughly one disk scale length to be [ fe / h ] = @xmath00.45 , with fewer than 10% of the rgb stars more metal - poor than [ fe / h ] = @xmath01 .
we found evidence for a composite kinematic structure of the lmc disk , indicating either a velocity dispersion increasing with age , or a segregation by metallicity into thin and thick disks .
paper ii found the inner disk mdf to differ strongly from the only previous field star study , which had targeted a radially distant location ( projected radius @xmath78@xmath16 * ? ? ? * ) , in having a much smaller fraction of metal - poor stars and a sharper peak around the median .
we used our abundance data and photometry to constrain the age - metallicity relation of the inner disk , finding a slow increase in mean abundance with time .
in this paper , we present our measurements of the chemical abundances and radial velocities of a large number of red giant stars in the highest surface brightness region of the lmc : its bar .
bar fields could not be targeted using hydra because the wide ( 2 arcsecond ) diameter of its fibers and the difficulties of sky subtraction using fiber systems .
we begin by giving some background on the lmc and on the field studied here ; this bar field includes the fields singled out for detailed study with wfpc2 by the wfpc2 team @xcite and by our guest observer program ( go7382 ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
section 2 describes our observing program at the 8.2 meter yepun ( vlt - ut4 ) telescope at the european southern observatory s paranal observatory , and the derivation of metallicities and radial velocites from our spectra .
we also discuss how we combine optical photometry and the new metallicity information to estimate the stellar ages of the target red giants . in section 3
we present the bar field mdf , comparing to the solar neighborhood , other regions of the lmc , and the lmc star clusters .
we explore the connection between kinematics and metallicity and how the line of sight velocity dispersion changes with abundance .
section 3.3 compares the derived amr to the predictions of chemical evolution models based on both smooth and bursting histories of star formation .
our results are summarized and the implications discussed in section 4 . to orient the reader and allow us to place our results in a broader context
, we show a schematic diagram of the lmc in figure [ fig - map ] .
the map is an albers equal - area conic projection = 6@xmath17 , @xmath18 = @xmath090@xmath19 , with reference latitudes @xmath20 = @xmath090@xmath19 , @xmath21 = 0@xmath19 . ]
@xcite of equatorial coordinates spanning @xmath711@xmath19 @xmath22 12@xmath19 .
the major large - scale features of the lmc stellar and gas distributions are shown .
the solid ellipses follow the smoothed near - infrared isopleths as fit by @xcite , with semimajor axes of 1 , 1.5 , 2 , 3 , 4 , 6 , and 8 degrees . within the 2@xmath19 isopleth ,
the red starlight is dominated by the bar ; outside this radius , the isopleths are elongated towards the milky way @xcite , and virtually all the stars have disk - like kinematics @xcite .
the rotational center of this disk @xcite is marked by the black square .
as many authors have remarked , the neutral hydrogen distribution is significantly offset from the center of the starlight ; we show the kinematic centroid of the @xcite as a black triangle .
major features are plotted with dashed lines , following the maps in @xcite : the main disk , roughly 9 kpc across , the archetypal supergiant shell lmc4 near the northern edge of the disk , and several diffuse , tidal arms that spread out from the southeast and the west side of the disk . as noted by @xcite ,
the distinctly resembles a barred , late - type spiral for velocities between 260280 km s@xmath3 , with two arms connected by a bridge of gas .
we have plotted this bridge in figure [ fig - map ] , which serves to show that there is no strict morphological relation between the optically - identified bar and any major structure in the gas distribution .
additional reference points are given by star symbols marking the locations of the two most active star - forming regions in the lmc : the 30 doradus complex northeast of the bar , and the n11 region near the northwest edge of the disk .
the distribution of stars particularly the most recent generations and gas is incredibly complex and structured on all scales , but this sketch is sufficient to identify the major morphological features that bear on our results .
the galactic center is toward the south ; in this representation , the smc is to the lower right , the orbit of the clouds carries them to the northeast ( towards the galactic plane ) , and the lmc s rotation is clockwise . the features detailed above place our metallicity and kinematics results into larger context .
we have been concerned with five fields , marked by the appropriate alphanumeric tags in figure [ fig - map ] .
each field contains between 36 and 373 field red giants for which spectra at the triplet have been obtained .
the inner disk fields have been studied by ( * ? ? ?
* d1 , paper i ) and ( * ? ? ?
* d1 and d2 , paper ii ) using the ctio 4-meter telescope .
results for the transitional disk field ( `` td '' ) , so called because of its location near the edge of the disk , and the eastern field ( `` e '' ) will be reported in a future paper ( cole et al .
2004 , in preparation ) .
the bar ( `` b '' ) field is effectively located at the heart of the lmc , 0@xmath233 from the center - of - rotation of carbon stars @xcite .
a small number of stars in the outer ( `` o '' ) field were observed by @xcite ; these have remained for more than a decade the only abundance measurements of field giants lying outside the gas - rich disk .
the remainder of this paper is concerned with the bar field .
figure [ fig - bar ] shows the bar field in more detail , oriented with north at the top and east to the left .
the image is in the infrared band of the second digitized sky survey , obtained from the canadian astrophysics data centre .
it is centered at ( @xmath24 , @xmath25 ) = ( 5@xmath1724@xmath26 , @xmath069@xmath1949 ) and spans 30 arcminutes . within this area
are identified 30 star clusters ( blue ellipses ) and five areas of h@xmath27 emission ( magenta ellipses ) , labelled following the atlas of @xcite .
four unlabelled red ellipses mark obscuring dust clouds identified in @xcite .
ob associations are notably absent from the field despite its high stellar density .
minor regions are present , but easily avoided .
the large emission region n132 is a fossil region around the oxygen - rich supernova remnant n132d , which probably arose from a type ib supernova some 2500 yr ago @xcite .
n132d is injecting dense knots of oxygen - rich material into its immediate vicinity @xcite , a vivid reminder that the abundances of the red giants measured here are only indirectly related to the present - day gas phase abundances .
the region around n132 contains a few blue supergiants in projection @xcite ; this type of stellar population is far more common further west along the bar and northeast towards 30 doradus . in many places
the contribution of intermediate - age and old stars is impossible to isolate cleanly .
features of the diffuse interstellar medium are omitted for readability , but the field is comparatively simple in structure compared to much of the lmc .
the neutral hydrogen in this field is not broken into high- and low - velocity components , as it is in large regions to the northwest and southeast ; the column density through this field is roughly n@xmath28 = 8@xmath2210@xmath29 @xmath30 @xcite .
two relatively minor molecular clouds , identified based on carbon monoxide line emission at 2.6 mm by @xcite , are found in the southwest corner of the field and much of the east - central portion .
areas included for study in this paper or related work are also marked .
the yellow wfpc2 footprints show the fields in which we have obtained deep images in v and i band in order to create color - magnitude diagrams reaching magnitude @xmath826 ( * ? ? ?
* cole et al .
2004 , in preparation ) .
similar photometric data were obtained by the wfpc2 team @xcite in the area marked by the green wfpc2 footprint .
the spectroscopic data presented in this paper were obtained within the seven fields shown by the black squares , each of which is 6.8 arcminutes on a side .
the heavy black rectangle outlines the region between 5@xmath1722@xmath26 @xmath1 @xmath24 @xmath1 5@xmath1726@xmath315 , @xmath070@xmath1905 @xmath1 @xmath25 @xmath1 @xmath069@xmath1935 ; this is a convenient simple border for the irregular region comprised of tiled fors2 fields .
overall , the bar field is dense with stars ( surface brightness @xmath32 = 20.7 mag / arcsec@xmath33 : * ? ? ?
* ) , but is not a region of strong current star formation or high dust obscuration .
@xcite estimate the foreground reddening towards the bar field to be e(b@xmath0v ) = 0.06 mag . with the column densities of reported in @xcite , and applying the n@xmath28e(b@xmath0v )
relation from @xcite , we make a first estimate of the mean reddening to the giants in our field as e(b@xmath0v ) = 0.08 @xmath50.02 mag .
some differential reddening is almost certainly present , as will be discussed below .
relatively free of recent activity , the bar field is expected to be an ideal place to study the intermediate - age and old stars of the central regions of the lmc .
while the bar field is as close as practical to the centroid of old ( red ) stars in the lmc , it is offset from the centroid of bright blue stars , which are more well - aligned with the than with the red starlight ( e.g. * ? ? ?
the flocculent bridge of gas seen in the channel maps of @xcite passes just north of our field , and is misaligned with the optical bar . this misalignment is shared with the distributions of very massive main - sequence stars , the brightest red supergiants , and dust - shrouded protostars the effect is perhaps best seen in the 2mass starcount maps published by @xcite .
this population - dependent bar structure is interpreted as being due to evolution of the disk structure over time resulting from a combination of internal and external perturbations ( e.g. * ? ? ?
owing to time - dependent evolution , it is not particularly meaningful to discuss the `` stellar '' bar of the lmc , because stars of different ages are distributed differently .
the familiar optical bar is perhaps best regarded as a `` fossil '' bar , while the less distinct bar traced by extreme population i objects and ( possibly ) disturbances in the velocity field ( see , e.g. , * ? ? ?
* ) can be thought of as a `` stelliparous '' _ parere _ = to bring forth .
] bar . by comparing the morphologies of different types of stars ( e.g. * ? ? ? * ) and star clusters ( e.g. * ? ? ?
* ) , we can roughly assign stars more massive than @xmath768 m@xmath10 , and clusters of swb type @xcite earlier than iii to the stelliparous bar , and older stars to the fossil bar .
this puts the age break between the two systems at roughly 70200 myr .
this is intriguingly close to the epoch of the last major interaction with the small magellanic cloud , an event which could have dramatically redistributed the angular momentum of the lmc disk .
once the bar feature is formed , it can persist for many orbital periods , with the stars born and trapped in the bar comprising a dynamical subsystem of the galaxy ( e.g. * ? ? ?
* ; * ? ? ?
throughout the rest of this paper , the term `` bar field '' will be used to refer to the region shown in figure [ fig - bar ] . where a broader context is intended
, we will use the terms fossil bar and stelliparous bar to distinguish these different morphological systems .
it is important to note that the bar is historically defined purely on the basis of optical appearance ( figure [ fig - map ] ) , and the use of the term does not necessarily imply that a kinematic distinction can be made between populations with different angular momenta and energies at the same location .
when population ages are referred to , we will use the terms very young ( 00.2 gyr ) , young ( 0.21 gyr ) , intermediate - age ( 110 gyr ) , and old ( 10 gyr ) . for stellar populations with velocity dispersions of 1020 km s@xmath3
, 0.2 gyr is enough time to diffuse throughout the roughly 2.5 kpc length of the fossil bar ; since our results primarily concern intermediate - age and old stars , they can be taken as representative of the older populations of the fossil bar .
the near - infrared triplet ( cat ) coming from the ( 3@xmath34d4@xmath34d ) transition is an extremely useful set of lines for the measurement of radial velocities and metallicities in k giants @xcite . the triplet line strength in old
, metal - poor red giants can be empirically calibrated for metallicity by removing the influence of surface gravity via a simple linear equation in v magnitude @xcite .
this empirical calibration has recently been shown to be applicable to stars nearly as metal - rich as the sun and as young as 2.5 gyr ( * ? ? ?
* paper iii ) .
the empirical calibration of the cat to v and [ fe / h ] for red giants is supported by theoretical arguments @xcite as well as by examination of large spectral libraries @xcite .
the three triplet lines , at @xmath35 = 8498 , 8542 , 8662 , are among the strongest spectral features in k giants , and fall neatly between regions of strong telluric h@xmath36o absorption .
this has made it an extremely popular method for the measurement of abundances in interemediate - age and old stars in dwarf galaxies throughout the local group ; the pace of this work has greatly accelerated with the advent of multiobject spectrographs and 810 meter class telescopes .
we picked our targets to be far enough down the giant branch that we could avoid m stars while still fully sampling the color range of the red giant branch ( rgb ) .
our color - magnitude diagram of the region , obtained at ctio @xcite is shown in figure [ fig - sel ] , with our targets highlighted by heavy points .
we picked our stars to have 15.5 @xmath1 i @xmath1 16.5 , and to have v@xmath0i colors bracketed by two isochrones with log(t ) = 9.40 ( age = 2.5 gyr ) from @xcite .
the selection region is bounded on the blue side by the z = 0.0001 ( [ fe / h ] = @xmath02.3 ) isochrone and on the red by the z = 0.019 ( [ fe / h ] = 0.0 ) isochrone .
the selected targets effectively span the observed width of the rgb and so the the exact color limits are not critical to the results .
some stars brighter than the i = 15.5 cutoff were observed when a slit would otherwise have gone unassigned .
astrometry and photometry for the stars in the central block of four fors2 fields ( see figure [ fig - bar ] ) were taken from ctio data published in @xcite .
these core fields were the originally - intended spatial extent of our spectroscopic survey .
however , the southeast quadrant of the core , around ngc 1950 , was unexpectedly crowded with very bright stars and appeared to have a large young population ( possibly in an unbound corona of cluster stars ) .
thus we added three additional flanking fields around the region with ctio photometry . for these flanking fields , we selected targets based on data from the ogle - ii survey @xcite .
comparison of stars with measurements from both sources found @xmath37 for analysis using optical photometry , we use the ctio photometry , or the transformed ogle - ii photometry .
the position of each target was confirmed using fors2 preimages obtained in service mode several weeks prior to the observing run , and the slits were assigned using the fors instrument mask simulator ( fims ) software distributed by eso .
the spectroscopic observations were made in visitor mode at the yepun ( vlt - ut4 ) 8.2-m telescope at eso s paranal observatory , on the nights of 2426 december 2002 .
we used the fors2 spectrograph in multi - object ( mos ) mode , with the 1028z@xmath3829 grism and og590@xmath3832 order blocking filter . in this configuration , the fors2 field
is covered by a mechanical assembly of 19 slit jaws , each 2022 arcseconds long , that can be arbitrarily positioned along the horizontal ( east - west ) axis of the field .
we chose to use a constant slit - width of 1 arcsecond for ease of calibration .
the spectral images were recorded on two 2k@xmath222k mit / ll ccds , which have a read noise of 2.7 electrons and an inverse gain of 0.8 @xmath39 adu@xmath3 .
the physical pixels were binned 2@xmath222 , yielding a plate scale of 0.25 arcsec pixel@xmath3 .
the resulting spectra cover 1700 , with a central wavelength near 8500 and dispersion 0.85 pixel@xmath3 ( resolution 23 ) .
the fors2 field is 6.8 arcmin across , but is limited to 4.8 arcmin usable width in the dispersion direction in order to keep important spectral features from falling off the ends of the ccd .
the log of observations is given in table [ tab - obs ] , which gives the field names and centers , time of observation , the seeing measured by the differential image motion monitor ( dimm ) , and the number of rgb targets recovered from each setup .
the table includes the same information for the 12 galactic star clusters that were used as metallicity calibrators ( _ q.v .
each lmc setup is identified by a number corresponding to its position , and a unique suffix of one or more letters that refers to the slit configuration files produced by fims .
each configuration was observed twice , with offsets of 3 arcsec between exposures , to ameliorate the effects of cosmic rays , bad pixels , and sky fringing .
the total exposure time in each setup was 2@xmath22600 sec , yielding typical signal - to - noise values of s / n @xmath1130 per pixel .
the seeing varied between 0@xmath405 @xmath1 fwhm @xmath1 1@xmath404 during the run , with a median value around 0.8 arcsec .
calibration exposures were taken in daytime , under the fors2 instrument team s standard calibration plan .
these comprised lamp flat - fields with two different illumination configurations and he - ne - ar lamp exposures for each slit configuration .
two lamp settings are required for the flat - fields because of parasitic light in the internal fors2 calibration assembly ( t. szeifert 2003 , private communication ) .
owing to the large number of setups in our program , twilight flats were impractical .
all basic data reduction steps were performed under iraf .
we fitted and subtracted the scaled overscan region , trimmed the image , and divided by the appropriately combined lamp flats within the ccdred package .
spectroscopic extractions were performed with hydra , an iraf package for handling multislit spectra .
our targets were bright enough that the object trace could be extracted directly from the science exposures . across the @xmath41-axis of the ccd , the curvature of the trace along the @xmath42-direction varied significantly , but could in all cases be fit with a low - order polynomial . because of the high spectral density and signal - to - noise of night - sky emission lines primarily oh @xcite and o@xmath36 @xcite we used these lines to dispersion correct each spectrum directly instead of using the arc lamps .
typically , the 30 or so strongest emission lines were used in the wavelength solutions , giving a typical root - mean - square ( rms ) scatter of 0.040.08 .
because of the scatter in target positions across the dispersion direction of the field , individual spectra can reach wavelengths as blue as 7200 , or as red as 10,100 ; most were centered close to 8500 , covering the approximate range 7600 @xmath1 @xmath43 @xmath1 9400 .
extraction to one - dimensional spectra was performed within the apall tasks .
sky subtraction was achieved using one - dimensional fits to the background perpendicular to the dispersion direction . because the targets are bright compared to the sky , and the slits are long compared to the seeing disk , this presented few difficulties .
an exception was when the stars fell near the ends of the slitlets ; in these cases the sky region was chosen interactively and adjusted to produce the cleanest extracted object spectrum in the region around the cat .
some stars very close to the top or bottom of the ccd frames showed high sky residuals and were excluded from subsequent analysis .
the dispersion - corrected spectra were combined using scombine to minimize the effects of bad pixels and cosmic rays .
each spectrum was continuum normalized by fitting a polynomial to the spectrum , excluding the cat and regions of strong water vapor absorption .
sample spectra are shown in figure [ fig - spec ] .
each extracted rgb star and its fors2 field identifier are listed in table [ tab - dat ] , with the vi magnitudes from the sources listed above .
the stars are identified by their number in the 2mass point source catalogue @xcite , except where an unambiguous identification was not possible ; in these cases the number from the ogle - ii catalogue @xcite is used .
if a target lies in or around a feature of interest in figure [ fig - bar ] or has unusual spectral characteristics , this is noted as well .
the full table is available in the electronic version of the journal .
we are interested in stellar radial velocities in order to reject possible foreground milky way stars , and to search for correlations between the moments of the velocity distribution and metallicity that could help distinguish between different stellar populations .
we performed fourier cross - correlation @xcite between our target spectra and the spectra of template stars of known radial velocity .
24 red giants in galactic star clusters were used as templates ; these were a subset of the stars used in our metallicity calibration ( paper iii ) , ensuring a good spectral match between templates and program stars .
we used the iraf fxcor task to perform the cross - correlation , and the radial velocities were found from the average of velocity offsets from each template , weighted by the random error and the height of the correlation peaks .
the observed velocities were then corrected to the heliocentric reference frame for subsequent analysis . because the stellar image was smaller than the slit width in most cases , there were in many cases slight misalignments between the slit centers and the stellar centroids .
this effect propagates into a potentially large systematic error in the observed radial velocity ( e.g. * ? ? ?
we can correct for this velocity offset if we know the magnitude of the offset in pixels between the centroid of the stellar image and the centerline of the slit on the ccd .
images taken through the slit mask , without the grism , prior to each exposure were used to determine this offset .
each through - slit exposure images a patch of sky @xmath721 arcsec long by 1 arcsec wide onto the ccd for each target .
the stellar centroid is determined by a simple profile fit to the through - slit image , while the position of the slit itself is measured from 1-dimensional fits to the profile of the sky , excluding the stellar flux .
the typical offset was less than 0.3 pixels , compared to the slit width of 4 pixels .
when a nonzero offset in the dispersion direction was found , we applied corrections to the measured radial velocities based on the dispersion solution measured from the night - sky emission lines . with our spectral resolution of @xmath729.5
km s@xmath3 pixel@xmath3 , the resulting velocity corrections ranged from @xmath44 = 0 to 32 km s@xmath3 , with a mean correction of @xmath00.05 km s@xmath3 , and a mean absolute correction of 8.5 km s@xmath3 .
we estimate that our centroiding accuracy is roughly a quarter of a pixel , or @xmath77 km s@xmath3 , and we therefore add this in quadrature to the error in the cross - correlation for our final error estimates .
the heliocentric velocities and their associated errors are given in table [ tab - dat ] .
the mean radial velocity of our sample is v@xmath10 = 257 km s@xmath3 , with a root - mean - square dispersion of 25 km s@xmath3 about the mean .
we found no stars with velocities characteristic of the milky way disk ( v@xmath10 @xmath45 100 km s @xmath3 ) , and the observed velocity range of 174 km s@xmath3 @xmath1 v@xmath10 @xmath1 336 km s@xmath3 is entirely consistent with the known range of lmc radial velocities ( e.g. * ? ? ?
some galactic halo giants have similar velocities , but since they are far fewer in number than disk stars , the contamination rate is negligible .
the histogram of heliocentric radial velocities is shown in figure [ fig - vel ] .
for comparison to the expected distribution , a thin disk model for the velocity is overplotted : the mean is derived from the equations in @xcite to be 260 km s@xmath3 , and the dispersion of 24 km s@xmath3 is taken from @xcite .
we used the program ew , described in paper iii , to measure the equivalent widths of the cat lines by fitting each of the three lines by the sum of a gaussian plus a lorentzian , constrained to have a common line center .
this was deemed necessary to account for the very strong damping wings of the lines .
the profile fits were integrated over the line bandpasses @xcite to yield the pseudo - equivalent widths .
error estimates were obtained by measuring the root - mean - square scatter of the data about the profile fits .
the summed equivalent widths of the three lines ranged from 3.5 @xmath1 @xmath46w @xmath1 10 , with typical errors of @xmath72% .
table [ tab - dat ] gives these values for each target . because the relation between cat equivalent width and metallicity is empirically defined , and because there have been hints that the calibration becomes nonlinear at the high - metallicity end ( e.g. * ? ? ?
* ) , we observed red giants in 12 galactic star clusters to define the relation .
the clusters span the metallicity range @xmath02.0 @xmath1 [ fe / h ] @xmath1 @xmath00.1 , and the age range 2.5 gyr @xmath1 age @xmath1 @xmath812 gyr ( see paper iii for details ) . while many of the bar red giants are probably younger than 2.5 gyr ( q.v . ) , extrapolation of the calibration to ages @xmath451 gyr does not seem unreasonable @xcite .
the calibration relies on the empirical fact that red giants of a single metallicity follow a linear relation between @xmath46w and their v magnitude above the horizontal branch , ( v@xmath0v@xmath47 ) : + @xmath48 ) \equiv \sigma w + 0.73 ( v - v_{hb}),\\\ ] ] which then leads to the following relation between the reduced equivalent width , w@xmath49 , and [ fe / h ] : + @xmath50 = ( -2.966 \pm0.032 ) + ( 0.362 \pm0.014 ) w^{\prime},\\\ ] ] with rms scatter @xmath2 = 0.07 dex .
the distribution of target stars in the ( v@xmath0v@xmath47 ) , @xmath46w plane is shown in figure [ fig - cal ] , with isometallicity lines shown for reference .
when comparing to abundances of galactic star clusters , it is important to remember that this calibration is derived with respect to the abundance scale derived by ( * ? ? ? *
cg97 ) for globular clusters , and the compilation of @xcite for open clusters .
the globular cluster and open cluster abundance scales are thought to be consistent ; work is in progress to obtain a homogeneous set of calcium abundances from high - dispersion spectroscopy for a large sample of clusters so that in the future measurements can be calibrated to a single system @xcite . to derive [ fe / h ] ,
we adopt the horizontal branch magnitude v@xmath47 = 19.22 , based on our wfpc2 and ctio photometry ( paper ii ) morphologically , this feature is really a red clump and not a horizontal branch in the strict sense ( see figure [ fig - sel ] ) ; as shown in paper iii , this does not affect our metallicity determinations .
the red clump has a v magnitude dispersion of @xmath50.12 mag , which we propagate through into our metallicity error estimates .
the total random 1@xmath2 error on each metallicity measurement is 0.10.2 dex , with an average value of @xmath50.14 .
the derived metallicities and their estimated 1@xmath2 errors are given in table [ tab - feh ] .
the mean of the sample is [ fe / h ] = @xmath00.45 , with a dispersion of @xmath50.31 dex .
however , because of the long tail of metal - poor stars , the median is a better statistical estimator of the of the typical metallicity , which is [ fe / h ] = @xmath00.40 .
the interquartile range is @xmath00.51 to @xmath00.28 , and the 10th and 90th percentiles of the distribution are , respectively , @xmath00.70 and @xmath00.20 .
the distribution is plotted in figure [ fig - mdf ] .
an expanded view of the rgb region of the cmd is shown in figure [ fig - rgbz ] .
spectroscopically observed stars are color - coded by metallicity , with the ranges chosen for clarity of display .
it is easily seen that the most metal - poor and metal - rich stars roughly divide themselves in color but stars near the peak of the metallicity distribution , between @xmath00.6 @xmath1 [ fe / h ] @xmath1 @xmath00.3 , span the full color width of the rgb .
there is even some overlap between stars with [ fe / h ] @xmath7 @xmath00.2 and @xmath7 @xmath00.8 .
this is a vivid demonstration that where a large age range is present in a stellar population , the mean color of the rgb is an innacurate measure of the metallicity . some of our faintest , reddest stars are far more metal - poor than would be expected , and could be differentially reddened .
there are two additional points to take from figure [ fig - rgbz ] : first , that although we searched well to the blue of the bulk of the rgb based on the expectation that metal - poor red giants should be bluer than their metal - rich counterparts , most of the bluest rgb stars are in fact relatively metal - rich ; and second , the broad color range of stars at the peak of the mdf is indicative of an extremely large range of stellar ages accompanied by little chemical enrichment over time .
this encourages us to quantify the age distribution and age - metallicity relation .
we adopt the procedure described in paper ii to derive isochrone ages for each of our target stars .
we use a program developed by one of us ( aac ) to place isochrones of arbitrary metallicity in the color - magnitude diagram , and linearly interpolate in the logarithm of the age to find the ( [ fe / h ] , log(t ) ) pair that reproduces the stellar metallicity and location in the cmd .
this is not a precise technique , because the effective temperature of a red giant is principally controlled by its convective envelope opacity , which is largely a function of the abundance of heavy elements in the star ( e.g. , * ? ? ?
however , the common perception that the temperature of a red giant is largely independent of its mass is inaccurate .
@xcite demonstrated that because red giants are almost entirely convective in their interior they share a common envelope structure . using analytic homology relations , they showed that when radiation pressure is unimportant , red giants must evolve approximately along tracks described by a relation of the form + @xmath51 using + @xmath52 we derive that , insofar as the conditions in the helium core have only a small effect on the outer envelope of the red giant , + @xmath53 thus from basic physical considerations , we expect that for constant luminosity t@xmath54 @xmath55 m@xmath56 ; this is similar to the dependence discovered in numerical models by @xcite and recovered in modern isochrone sets ( e.g. , * ? ? ?
the magnitude of dt@xmath54/dm varies with the mass and luminosity of the giant , but is of order 300500 k / m@xmath10 . in the context of figure 8
, this means that two stars of the same abundance and magnitude will have the same color if they have the same mass ( and hence , age ) .
if one star is more massive ( younger ) than another , it will be bluer .
this color difference is translated into an age difference using the published isochrones .
given representative values of d(v@xmath0i)/dt@xmath54 ( e.g. , * ? ? ?
* ) and the ( strongly decreasing ) function dm / dt , we find that for perfectly known metallicity and distance , v@xmath0i photometry good to 2% could be sufficient to provide an age with @xmath810% accuracy . in practice ,
random errors in the metallicity measurement dominate the uncertainty , and our typical age errors are of order 60100% ( 0.20.3 in the log ) . it is important not to assign undue weight to the age estimate of an individual star , but to use large samples of stars to beat down the random error and thereby glean some information about the mean age - metallicity relation .
we tested our technique on a small sample of star clusters in paper ii , and found reasonable agreement with main - sequence turnoff ages , albeit with large scatter .
studies of dwarf spheroidal galaxies ( e.g. * ? ? ?
* ; * ? ? ?
* ) are generally supportive of the idea that _ some _ age information can be gleaned from a combination of accurate rgb metallicities and colors . in the most detailed published study of four dwarf spheroidals @xcite
, the derived age distributions showed broad agreement with the star - formation histories derived from main - sequence turnoff photometry .
however , a significant decrease in sensitivity was noted for abundances much below [ fe / h ] @xmath57 @xmath01 .
an in - progress study of fifty - eight red giants in the carina dwarf spheroidal by one of us ( tsh ) will make an extremely interesting test case ; first indications are that the most metal - rich stars ( [ fe / h ] @xmath11 @xmath01.5 ) are on average a factor of four younger than the stars with [ fe / h ] @xmath58 @xmath02
. however , finite errors on the observed metallicities ( see * ? ? ?
* ) make it unlikely that rgb - derived ages will ever clearly resolve the discrete epochs of star - formation derived from main - sequence turnoff photometry as in @xcite .
a potentially major contributor to the error budget is the uncertainty in relative abundances of the various elements heavier than helium .
because the lmc has experienced a different chemical evolution history than the milky way , the scaled - solar abudance ratios can not be assumed to apply .
evidence for changing values of [ o / fe ] with [ fe / h ] in lmc field red giants has been presented by @xcite and the amount of data is increasing rapidly ( hill et al .
, in preparation ) .
@xcite and @xcite have also measured the changing abundance of @xmath27-process elements relative to iron in several massive lmc star clusters , finding important differences between the lmc globulars and the standard milky way population ii abundance mixture . because the @xmath27 elements are major electron donors in red giant envelopes , [ @xmath27/fe ] ratios play a major role in determining the stellar t@xmath54 ( e.g. , * ? ? ?
therefore it is vital in attempting to age date a red giant with isochrones that the correct relative abundance blend is used . for practical purposes
, we represent all deviations from the scaled - solar abundance mixture using the parameter [ @xmath27/fe ] , and adopt the approximate relation between the overall abundance of electron donors , [ m / h ] , and [ @xmath27/fe ] given by @xcite : @xmath59 } \approx \mathrm{[fe / h ] } + \log ( 0.638\ ; 10^{[\alpha / fe]}\ ; + 0.362).\\\ ] ] because of the limited amount of data available , we make the simplifying assumption that [ @xmath27/fe ] = [ o / fe ] .
combining the data from @xcite and the preliminary results from hill et al .
( in preparation ) , we approximate the trend of [ o / fe ] with [ fe / h ] by a bilinear relation : @xmath60 = \left\ { \begin{array } { r@{\quad:\quad}l } 0.05 - 0.10\ , [ \mathrm{fe / h } ] & [ \mathrm{fe / h } ] \leq -1 \\ -0.413 - 0.563\ , [ \mathrm{fe / h } ] & [ \mathrm{fe / h } ] > -1 \end{array } \right.\ ] ] there seems to be scatter of 0.10.2 dex about the mean [ o / fe ] at given [ fe / h ] , but this is not definitely larger than the measurement uncertainty .
the values of [ @xmath27/fe ] adopted for purposes of the age calculation are included in table [ tab - feh ] .
we adopt the lmc `` standard '' distance of 50.1 kpc , based on the distance modulus adopted by the hst key project to determine h@xmath61 @xcite ; adopting the lmc disk structure from @xcite , the bar field is 0.1 kpc more distant than the lmc center of mass , giving a distance modulus ( m@xmath0m)@xmath61(bar ) = 18.50 .
this value is slightly higher than ( but in good agreement with ) more recent distance determinations based on eclipsing variables @xcite and rr lyrae stars @xcite . because of this good agreement
, we adopt an errorbar of @xmath50.06 mag in the distance modulus .
we adopt a reddening value e(b@xmath0v ) = 0.06 @xmath50.03 based on the discussion in @xcite ( also see * ? ? ?
* ; * ? ? ?
the padua isochrones include stars as old as log(age / gyr ) = 10.25 ( age = 17.8 gyr ) .
these were calculated in order to match the horizontal branch morphology of the oldest globular clusters , given what was known about their distances and the parameterization for stellar mass loss that went into the models .
there is now very strong evidence from the first year of data from the wmap satellite that the universe is 13.7 gyr old @xcite . in the meantime ,
both revisions in the cluster distance scale ( e.g. * ? ? ?
* ) and updated stellar interior calculations ( e.g. * ? ? ?
* ) have produced a strong expectation that the oldest globular clusters are no more than @xmath713.5 gyr old .
while these argue for an age bias in the padua isochrones at the old end of the scale , there is much support for their accuracy at intermediate ages .
studies of clusters @xcite and the field @xcite have derived ages for the oldest open clusters and the oldest stars in the solar neighborhood in the range of 911 gyr .
these are in excellent agreement with independent measurements of the age of the milky way disk : @xmath79 gyr from thorium cosmochronometry @xcite and from the faint end of the white dwarf luminosity function @xcite . to bring the results for our oldest stars into agreement with the known age of the universe ,
while at the same time preserving the success of isochrone measurements of the age of the galactic disk , we adopt the expedient of simply rescaling any ages older than 10 gyr . in the absence of detailed information
, we use a linear function @xmath62 the average age shift for the ten stars affected is @xmath01.3 gyr , well within the uncertainty in absolute age - dating of any old stellar population , and negligible compared to the measurement error in our method . because we adopt broad age bins in our analysis of the age - metallicity relation , the exact prescription for enforcing consistency between isochrone ages and the age of the universe has very little effect on our results
. there is ambiguity in the derived ages , because the evolutionary status of a star of given l , t@xmath54 , [ fe / h ] , and [ @xmath27/fe ] is not known _ a priori_. we have assumed that all stars are first - ascent red giants , except where the isochrones invalidate the assumption ; in these cases the age has been derived assuming the star in on the asymptotic giant branch .
if an assumed rgb star is in fact on the agb , our derived age will be roughly 30% too young .
the derived age estimates and random errors , expressed in logarithmic scale , are given in table [ tab - feh ] .
there is a concentration of stars at 13.7 gyr caused by eleven stars that were too red for the oldest isochrone , the ages of which were set equal to the age of the universe ( log a = 10.13 ) . for our choice of isochrone
set , reddening , and [ @xmath27/fe ] ratios , the median age of rgb stars in the bar field is 2 gyr .
the interquartile age range is 1.43.4 gyr , and 90% of the rgb stars in this field are younger than 6 gyr .
the bar field mdf ( fig .
[ fig - mdf ] ) is a basic datum that should provide strong new constraints on the inferred history of the lmc bar based on color - magnitude diagram or spectral synthesis studies .
the mean and dispersion of [ fe / h ] = @xmath00.45 ( systematic error @xmath7 @xmath50.1 dex ) , @xmath2 = 0.31 are not very meaningful statistical descriptors of the data , owing to the strong asymmetry of the distribution .
a maximum - likelihood analysis was used to fit two gaussian distributions to the unbinned data ; the resulting curve is plotted over the histogram in figure [ fig - mdf ] .
the curve is split into a narrow , metal - rich distribution containing 89% of the stars , with the remainder in a broad , metal - poor distribution .
the major population is described by mean @xmath63 = @xmath00.37 and @xmath64 = 0.15 , and the minor component by @xmath65 = @xmath01.08 and @xmath66 = 0.46 .
@xmath64 is barely larger than our measurement error , suggesting either that we have been too conservative in our error estimates or that the intrinsic astrophysical spread in metallicity is less than @xmath50.05 dex for this component .
the fraction of metal - poor stars is much smaller than in the abundance distribution of long - lived main - sequence stars in the solar neighborhood ( e.g. , * ? ? ?
* ) . however , there are strong systematic effects due to the rgb lifetime that make this comparison inappropriate .
the bar field red giant mdf is better compared to the solar neighborhood data for g and k type giants brighter than v @xmath7 5.5 obtained by @xcite .
the two distributions are shown in figure [ fig - mcw ] .
the solar neighborhood mdf shows a similar narrow peak at high metallicity , but is even more deficient in stars with [ fe / h ] @xmath1 @xmath00.8 .
the peak of the mcwilliam ( _ metal - rich bar _ ) sample is at [ fe / h ] = @xmath00.17 ( _ @xmath00.37 _ ) and the dispersion is @xmath2 = 0.16 ( _ 0.15 _ ) dex .
we can make a direct comparison to the abundance distribution of the lmc cluster system taken from the ossh paper .
ossh give abundance measurements for 70 clusters located across the body of the lmc , taken from similar spectra to those obtained here .
their equivalent widths are calibrated to globular and open clusters on a metallicity scale different from the one we use ( paper iii ) . by making a least - squares fit to the calibrating cluster abundances
, we adopt an _ estimated _ cluster abundance scale @xmath67_{clus } } \approx -0.212 + 0.498\ , \mathrm{[fe / h]_{ossh } } - 0.128\ , \mathrm{[fe / h]^2_{ossh}}.\\\ ] ] this recalibration explicitly includes the open clusters m67 and melotte 66 , and so supersedes that presented in paper ii , which was taken from cg97 , with ad hoc modifications above [ fe / h ] = @xmath00.5 .
note that some recent measurements of a subset of lmc clusters at high resolution and signal - to - noise support a recalibration of the ossh measurements , while others support the original ossh results ( e.g. , * ? ? ?
* ; * ? ? ?
these recent studies primarily concern clusters with [ fe / h ] @xmath57 @xmath00.9 . the histogram of recalibrated cluster abundances is plotted with our bar field data in figure [ fig - o91 ] .
the cluster mdf is distinctly bimodal , which presents a visual contrast with the long metal - poor tail of the bar field mdf .
the main peak of the cluster mdf , corresponding to the @xmath713 gyr old clusters , matches up well with the peak of the bar field mdf , but may be more asymmetric towards lower metallicities .
just such a relationship would be predicted by a model of star formation in which cluster formation events are shorter and more intense than field star formation episodes ( e.g , * ? ? ?
. however , we caution against overinterpreting the comparison in figure [ fig - o91 ] , because of the uncertainties introduced by the different metallicity scales .
two additional factors are the accelerated pace of stellar evolution at low metallicity , and the decrease with increasing age of the mass range sampled by our rgb selection region .
this effect ( discussed in detail in paper ii ) effectively biases us against detection of the older and more metal - poor stars in the field .
this must be taken into consideration when comparing the cluster and field star mdfs .
for example , 5% of the field rgb stars in the bar have metallicities in the range @xmath01.5 @xmath1 [ fe / h ] @xmath1 @xmath00.9 ; the same fraction of clusters ( 4 out of 70 , or 6 @xmath53% ) fall in this range . because of the bias against metal - poor field giants , the true relative fraction of astrated mass in this metallicity range is likely to be some 23 times higher , erasing any suggestion that the bar field mdf has a bimodality similar to the cluster mdf .
this effect should apply even more strongly to the most metal - poor ( oldest ) field stars , bringing the _ observed _ fraction of metal - poor stars with [ fe / h ] @xmath58 @xmath01.5 ( 4% ) approximately into line with the observed cluster fraction below @xmath01.5 , that is , 8 out of 70 ( 11 @xmath54% ) . because of the strong role of stellar age in determining the number of rgb stars in our selection window per unit stellar mass created
, a detailed comparison of the field star and cluster metallicity distributions must await a joint analysis of the color - magnitude diagram and metallicity distribution together .
independently of metallicity scales and sample biases , the bar field mdf has a different shape ( unimodal with tail ) than does the cluster sample ( bimodal with slight overlap ) .
this confirms the trend found in the inner disk in paper ii , and extends it into the center of the lmc .
we find that the bar field is closer in shape to the cluster mdf than the inner disk samples in paper ii .
for example , the fraction of field stars falling into the cluster metallicity gap is smaller in the bar field ( 5% ) than in disk 1 ( 13% ) or disk 2 ( 11% ) , despite the bluer color extent of the sample selection region in the current study .
this is probably indicative of the higher fraction of intermediate - age stars in the bar compared to the disk , expected on both observational @xcite and theoretical @xcite grounds .
the mean metallicity of our sample of bar field red giants is [ fe / h]@xmath68 = @xmath00.45 , essentially indistinguishable from that of the d2 field , [ fe / h]@xmath69 = @xmath00.46 . both are more than 0.1 dex more metal - rich than the d1 field that has [ fe / h]@xmath70 = @xmath00.59 .
the offset is many times the formal random error and is highly significant .
all three fields have similar dispersions about the mean metallicity , @xmath71}$ ] = 0.32 @xmath50.01 .
the difference between d1 and the other fields may be related to the location of the d2 field at the end of the bar and the d1 field in a region of much lower surface brightness , far outside the bar - distorted isophotes .
more data , in widely varying locations , is required before any firm conclusion about the possibility of spatial variation in mean metallicity can be reached .
because of the different selection effects , and the imminent addition of data from other locations in the lmc ( the transitional disk and eastern fields , see figure [ fig - map ] ) , we defer a detailed comparison of the bar and disk fields to a future paper . the radial velocity of our sample is entirely consistent with the disk rotation curve derived by @xcite from carbon star velocities , mostly at projected angular radii greater than 2@xmath16 . because our field is located almost directly at the rotation center of the disk , the rotation signature is expected to be small
. therefore we can not rule out the presence of a non - rotating ( halo or bulge ) or slowly - rotating ( thick disk ) disk component with these data .
stars on bar orbits are expected to show large streaming motions along the long axis of the bar ( e.g. * ? ? ?
. these could amount to several tens of kilometers per second , which would produce a signature in the radial velocity data as long as the bar does not lie in or nearly in the plane of the sky .
detailed predictions for the kinematic signature of off - center bars in dwarf galaxies are unavailable , but it seems likely that stars on bar orbits could be contributing to the non - gaussianity in the observed velocity field ( fig .
[ fig - vel ] ) .
such non - circular motions could also be contributing to the line of sight velocity dispersion , which is higher than that predicted by the thin disk model of @xcite .
the velocity dispersion along the line of sight is @xmath2 = 24.7 @xmath50.4 km s@xmath3 , in excellent agreement with the general sample of lmc stars measured by @xcite .
the dispersion is slightly higher than the value of 20 km s@xmath3 reported for the global average of lmc carbon stars by @xcite , and 60% higher than the 16 km s@xmath3 line of sight velocity dispersion of gas reported by @xcite . while the measured dispersion suggests a moderately thick structure , it is not high enough to imply that the majority of stars occupy a dynamically hot population such as a bulge or halo .
it was suggested by @xcite that such a structure could account for some observations of the morphology and structure of the inner lmc and give the appearance of a bar . for many years
, studies of the intermediate - age and old populations in the lmc have found increased velocity dispersions with age , up to a limit of roughly 3035 km s@xmath3 ( e.g. * ? ? ?
* ; * ? ? ?
even the oldest star clusters appeared to form a thick disk rather than a spheroid @xcite .
by contrast , @xcite have found a velocity dispersion of 53 km s@xmath3 for the rr lyrae type variables in the area of the bar .
these values roughly bracket what might be generally expected for a kinematically hot halo in the potential of the lmc ( m @xmath7 10@xmath9 m@xmath10 ) .
we can test our sample for similar effects by dividing into several subsamples . following the procedure in paper
ii , we show the line of sight velocity dispersion of samples in various metallicity ranges in table [ tab - kin ] . the plot of radial velocity vs. metallicity is given in figure [ fig - kin ] , showing that while the metal - rich and metal - poor stars share a total velocity range of over 100 km s@xmath3 , the bulk of the stars are far more concentrated towards the mean than the stars more metal - poor than [ fe / h ] @xmath7 @xmath01 .
the mean radial velocity barely changes with metallicity .
the dispersion starts at 16.7 @xmath51.6 km s@xmath3 for the most metal - rich ( and presumably youngest ) stars , increases dramatically by the next metallicity bin , and then gradually grows with decreasing metallicity until the last bin , when another large jump brings the line of sight dispersion of the stars below [ fe / h ] = @xmath01.15 to 40.8 @xmath51.7 km s@xmath3 .
note that this is completely in line with the expected line of sight velocity dispersion for a halo population , but our sample has neither the size nor the spatial extent to measure any deviations from a rotating disk among this minority population .
@xcite proposed that the optically - identifed lmc bar is actually a triaxial bulge .
if we take its luminosity to be of order @xmath810@xmath72 l@xmath73 , then the faber - jackson relation would predict a velocity dispersion in the neighborhood of @xmath770 km s@xmath3 for a classical bulge .
such structures are not associated with late - type galaxies like the lmc ; on the other hand , barred galaxies are strongly connected with the presence of box- or peanut - shaped pseudobulges ( e.g. , the very thorough review by * ? ? ? * ) .
the creation of pseudobulges , which are dynamically colder than classical bulges , is linked to secular dynamical evolution of the disk and bar .
a pseudobulge in the lmc would likely have a velocity dispersion of @xmath73040 km s@xmath3 , in agreement with what we observe for our most metal - poor subsample of stars .
however , pseudobulges are created from the general disk and bar stellar populations , and so there is no expectation that they should be preferentially more metal - poor than their surroundings .
indeed , @xcite have found that the stellar populations of pseudobulges are indistinguishable from those of the disks in which they are embedded .
this strongly suggests that the small fraction of stars we observe at high velocity dispersion and low metallicity does not owe its existence to the secular heating of the disk by the bar .
irrespective of nomenclature ( `` halo '' , `` bulge '' , or `` pseudobulge '' ) the dynamically hot population population in the central lmc is a very minor contributor to the total stellar surface density . the velocity dispersion we measure is comparable to our d1 and d2 results from paper ii .
we do not attempt to break our bar sample into thin and thick disk components , because of the unknown influence of the bar on disk structure and because the distribution in figure [ fig - vel ] is not particularly well - fit by two gaussians .
the probability is that the stellar populations are characterized by a continuum of velocity dispersions , rising with age due to gravitational scattering .
it is interesting to note that the most metal - rich stars are not significantly hotter than the neutral ism , suggesting either that the stellar disk was not strongly heated by the most recent encounter with the smc , @xmath7200 myr ago , or that the continuing gravitational interactions with the milky way and smc have kept the from cooling below this level .
the derived age - metallicity plot is shown in figure [ fig - amr ] .
typical errorbars are shown at the bottom of the plot for clarity . at a given age
, there is a large scatter in metallicity .
part of this is certainly a real scatter , and part of it stems from systematic effects such as differential reddening or incorrect assumptions about the [ @xmath27/fe ] ratio .
these effects are illustrated in figure [ fig - sys ] . in this diagram
the locations of six `` test '' rgb stars with i = 16 have been plotted in the age - metallicity plane .
arrows show how the derived ages would change if the stellar spectra were contaminated with that of a red clump star of [ fe / h ] = @xmath00.4 ( blue arrows ) , if the star was reddened by an additional 0.1 mag in e(b@xmath0v ) ( green arrows ) , or if the [ @xmath27/fe ] ratio had been assumed to be solar instead of following the lmc trend ( e.g. , * ? ? ?
note that the metallicity measurements are robust against these effects , which primarily affect the age estimates .
we find 14 stars with [ fe / h ] @xmath61 and age @xmath58 10 gyr .
similar populations are much less common in the data for the d1 and d2 fields ( paper ii ) .
this is not an artifact of the different selection criteria adopted : only one of these stars lies blueward of the adopted edge of the selection region in the earlier paper .
as we might expect , most of the stars that the d1 and d2 selection regions would have missed are quite young , with an average age below 0.5 gyr .
the presence of these stars at the blue side of the rgb indicates that young stars are much more prevalent in the bar than in the disk at a radius of one scale length .
the results of figure [ fig - sys ] suggest that some fraction of the apparent intermediate - age , metal - poor stars may be unresolved blends of genuinely ancient , metal - poor stars with intermediate - age red clump stars near the peak of the mdf . a higher percentage of blended stars in the bar than in the disk would naturally be expected because of the much higher stellar surface density in the bar field .
some evidence in favor of this interpretation can be taken from the radial velocities and age estimates of the two stars that have [ fe / h ] @xmath61 that were flagged during data reduction as having faint companions ( table [ tab - dat ] ) .
2mass point sources 05225632 - 6942269 and 05253235 - 6943137 have ages , respectively , of 1.9 @xmath50.8 gyr and 7.6 @xmath55 gyr , much younger than average for their metallicities of @xmath01.19 and @xmath01.61 . depending on the properties of the faint companion objects
, their true ages could be much older .
because of the large random errors ( up to a factor of two ) and the possibility that systematic effects may ruin some individual estimates , the age information is most usefully interpreted when binned up to increase signal - to - noise and suppress the effects of outliers .
we sort the stars into five equal - age bins 2.7 gyr wide , and show the mean metallicity in each bin in figure [ fig - amrb ] .
the five faintest , reddest stars , measured perpendicular to the rgb ridgeline ( see fig . [ fig - rgbz ] ) , have been excluded from the averaging because differential reddening is suspected .
the vertical errorbars on each point show the rms scatter about the mean in each bin , and the horizontal bars denote the extent of the bin .
the area of each point is proportional to the number of stars in the bin , ranging from 10 in the oldest bin to 255 in the youngest .
figure [ fig - amrb ] shows that the metallicity steadily increases with time : quickly at ancient times , and then by @xmath570.5 dex over the past 10 gyr . the metallicity scatter
_ appears _ to decrease with time , from @xmath50.5 dex in the oldest bin to @xmath50.2 dex in the youngest . for the oldest bin ,
the likely culprit is the requisite rapidity of chemical evolution from [ fe / h ] @xmath7 @xmath03 at the end of the population iii phase to @xmath7 @xmath01 within the first 4 gyr .
the fact that the errors in our age estimate are large for the oldest stars probably introduces additional scatter by creating mixing between age bins .
the age - metallicity relation we derive is compared to the results from studies that have focused on specific sub - populations of stars in table [ tab - pops ] .
the first two lines recapitulate the two gaussian fits to the mdf derived in section [ sec - mdf ] .
based on the derived ages , the two metallicity components are split into young and old populations , although there is obviously considerable overlap between them .
the remainder of the table shows the mean metallicities and dispersions of various stellar populations , arranged by increasing age .
where we have been able to trace the published abundances back to a scale similar to that used by ossh , we have applied equation 6 to bring the values into line with our data .
the picture is one of rapid evolution at early times , followed by a very slight increase over the past 10 gyr .
b dwarfs and cepheids are taken to be representative of the very young stellar populations ; their mean abundance is only very slightly higher than the peak of the metal - rich component of the field giant mdf in the bar .
evidently the chemical evolution of the lmc has been quite modest over the past @xmath810@xmath74 years .
very few tracers of chemical evolution at intermediate age have been available to date ; the sample of planetary nebulae measured by @xcite contains very few objects older than the oldest intermediate - age star clusters . the only star cluster with an age of 410 gyr is eso121 - 3 , with [ fe / h ] = @xmath00.93 ( ossh ; * ? ? ?
although the amr appears shallow , we find the mean metallicity of stars aged 36 gyr to be [ fe / h ] = @xmath00.46 @xmath50.02 , compared to @xmath00.72 @xmath50.03 for the stars aged 68 gyr , nearly a factor of two difference .
the low - metallicity component of our mdf is more metal - rich than the average of old star clusters or field rr lyrae variables , with a higher dispersion .
this indicates the continuous nature of field star formation , in that we have probed a much wider range of the lmc s history than just the oldest populations traced by the globular clusters and rr lyraes . if we just consider the 14 stars estimated to be older than 10 gyr and not suspected of differential reddening , the mean [ fe / h ] = @xmath01.31 , with a dispersion of @xmath50.51
this is consistent with the field rr lyrae stars of the bar , although the dispersion is larger than the value of @xmath50.29 dex in the rr lyrae sample of @xcite .
models for the chemical evolution of the lmc ( * ? ? ?
* ( pt98 ) ) , based on two different assumed star formation histories and with the yields adjusted to fit the ancient globular clusters and the numerous clusters aged 13 gyr , are overplotted on our binned age - metallicity relation .
the dashed line shows the chemical evolution derived from assuming that the star - formation rate had a broad peak @xmath710 gyr ago and has been very slowly declining since then .
the solid line marks instead the chemical history of a model lmc with roughly constant low level of star formation for most of its history , that then experienced a factor of six jump in star formation rate 3 gyr ago , with a subsequent rapid decline . both smooth and bursting classes of sfh can reproduce the star cluster age - metallicity relation , owing to the lack of clusters between 310 gyr old . as found in d1 and d2 ,
the field stars fill in the cluster age gap with a continuous distribution of ages and metallicities .
this raises the possibility that we can statistically distinguish between the two cases .
it can be seen from figure [ fig - amrb ] that stars aged from @xmath7210 gyr will have the strongest lever on the models , with little to differentiate between them at the oldest and youngest times . for each observed star , we start from the age derived in table [ tab - feh ] and calculate the probability that it was drawn from the age - metallicity relation appropriate to the bursting or the smooth model , taking the observational error on [ fe / h ] into account
. the relative likelihoods of the models are then computed by finding the joint probability of observing the entire ensemble of stars under each model .
the amount of cosmic scatter assumed in the model amr will influence the results , so we adopt @xmath75 = 0.15 as a realistic estimate .
we find that the observed amr is better matched by the smooth model from pt98 than their 3-gyr burst model at the 2@xmath2 level ( 95.8% confidence ) .
while the latter is a better match to the stars aged 510 gyr , these stars are greatly outnumbered by younger stars , which have higher abundances than predicted by the burst model .
however , many lines of evidence point to a bursting history of star fomation in the lmc . @xcite have shown that the epoch of increased star formation rate in our bar field is likely to have occurred earlier than 3 gyr ago .
figure [ fig - amrb ] makes plain that an earlier burst can be tuned to match the stars both older and younger than 5 gyr , and so can be made to be fit the data better than the smooth model .
based on the shape of the amr predicted by pt98 s bursting model , a burst would be expected to have occurred prior to @xmath75 gyr ago , but not much before @xmath77 gyr .
because of the low precision of our age estimates and the uncertainties in computing chemical evolution models , a maximum likelihood calculation of the time and amplitude of a starburst from these data is unlikely to produce meaningful astrophysical results .
we defer such an exercise to a future paper , in which we will simultaneously model the full wfpc2 cmds down to below the oldest main - sequence turnoff and the red giant mdf derived here .
we attempted to avoid star clusters and associations as much as possible , but three stars projected directly on two clusters ( hs 256 and hs 285 ) crept into the measured sample .
an additional thirteen stars ( identified in table [ tab - dat ] ) in the close vicinity of five star clusters were also measured .
considering the field density and the apparent blue colors of the clusters on our preimages , it is unlikely that the stars in the neighborhoods of the small clusters are bona fide members .
neither hs 256 nor hs 285 has a published age in the literature .
hs 256 partially overlaps with one of our wfpc2 fields , so it may be possible to examine its color - magnitude diagram separately from the field in a future paper .
two stars of our sample , 2mass 05223416 - 6944433 and 05223895 - 6945007 , are seen in projection against hs 256 .
their radial velocities are both very low compared to the sample mean , 222 and 227 km sec@xmath3 respectively , which may indicate their joint membership in a population with low velocity dispersion moving towards us at 30 km sec@xmath3 relative to the mean bar field . however
, the two stars have very different metallicities : [ fe / h ] = @xmath01.05 @xmath50.12 and @xmath00.42 @xmath50.14 .
there is no reason to suspect the quality of the abundance measurement in either star , but the widely discrepant values militate against common cluster membership .
the single star seen in projection against hs 285 is indistinguishable from the general field in both its radial velocity ( 261 km sec@xmath3 ) and metallicity ( @xmath00.37 ) .
we examined maps of the area , searching for spatial patterns in the radial velocity , metallicity , and age of the stars .
no strong evidence for structure in the populations was observed .
however , there did appear to be a slight concentration of the reddest stars into the southwest corner of the field , near the largest of the dark clouds identified by hodge , and some small , chainlike regions ( see figure 1 ) .
when comparing the 2mass j@xmath0k colors of the stars , we found the reddest stars to be far more evenly dispersed throughout the bar field . because the j@xmath0k color is less affected by reddening than is v@xmath0i , this is consistent with our interpretation that the concentration of stars with high v@xmath0i is not the effect of high metallicity or old age .
the high surface brightness and extreme crowding of the central regions of the large magellanic cloud have challenged observers for decades . in this paper , we present the first spectroscopic study of the abundances , kinematics , and age - metallicity relation for field red giants in the lmc bar . taking advantage of the superb image quality and efficiency of the fors2 spectrograph at vlt - ut4 ( yepun ) , we obtained spectra of 373 red giants in a 200 square arcminute region of the central bar that includes several fields singled out for detailed photometric study with wfpc2 @xcite .
we have derived abundances on a metallicity scale consistent with those of cg97 ( globular clusters ) and ( * ? ? ?
* open clusters ) ( paper iii ) , with internal accuracy of @xmath50.14 dex per star .
radial velocities accurate to @xmath57.5 km s@xmath3 were measured by fourier cross - correlation of our spectra with template stars of similar spectral type .
we used isochrones from @xcite and assumed non - solar elemental abundance ratios based on @xcite and hill et al.(in preparation ) to make age estimates with random errors of roughly 60% . 1 .
the mean metallicity of red giants in the central lmc is [ fe / h ] = @xmath00.45 , with a diserpsion about the mean of @xmath50.31 .
the distribution can be described by the sum of two normally distributed populations in the ratio of 8:1 , with the majority ( minority ) population having mean [ fe / h ] = @xmath00.37 ( @xmath01.08 ) and dispersion @xmath2 = 0.15 ( 0.46 ) .
half the stars have metallicities in the range @xmath00.51 @xmath1 [ fe / h ] @xmath1 @xmath00.28 ; only 10% are more metal - poor than [ fe / h ] = @xmath00.7 .
the mean heliocentric radial velocity of our sample is 257 km s@xmath3 the observed velocity dispersion of 24.7 @xmath50.4 km s@xmath3 is typical of intermediate - age lmc stars . the velocity dispersion increases with decreasing metallicity , from 16.7 @xmath51.6 km s@xmath3 for the most metal - rich 5% of stars , to 40.8 @xmath51.7 km s@xmath3 for the 5% of most metal - deficient stars ( [ fe / h ] @xmath61.15 ) .
over most of the intervening range , the velocity dispersion is roughly constant around 2227 km s@xmath3 .
the median age of the stars is roughly 2 gyr , with an interquartile range of 1.43.4 gyr .
90% of the rgb stars appear to be younger than 6 gyr .
this distribution does not linearly translate to the variation in star - formation rate over time because of strong rgb lifetime effects that bias the observed age distribution towards young stars .
the age - metallicity relation is in excellent agreement with measurements of the old and young star clusters and other tracer populations .
for the first time , we observe the evolution of metallicity over time through the cluster age gap from 310 gyr ago .
the amr combined with chemical evolution models appears to favor an increase in star - formation rate sometime prior to @xmath75 gyr ago .
the metal - rich component of the bar field mdf is similar in width to the mdf of solar neighborhood red giants as measured by @xcite , although the mean is shifted to lower metallicity by 0.2 dex .
the low - metallicity tail of the mdf appears not to be present in the solar neighborhood ; this is probably because our `` birds - eye view '' from above the lmc penetrates the disk at a steep angle , including populations regardless of the details of their vertical distribution .
the behavior of the velocity dispersion with metallicity is also reminiscent of the milky way disk , in which stars are born with low velocity dispersion that increases quickly with time for @xmath72 gyr , and then remains roughly constant with age until 10 gyr @xcite .
in the milky way , this is taken to be the signature of the thick disk at old times . in the lmc ,
the situation is less clear , and the possibility can not be eliminated that the most metal - poor , oldest stars are distributed in a spheroidal or halo distribution @xcite .
the suggestion of @xcite that the apparent bar may in reality a partially obscured , triaxial bulge seems disfavored by the observed velocity dispersion , which is much smaller than would be expected for a classical bulge .
the vast majority of red giants appear to be consistent with a thick disk type distribution .
a box- or peanut - shaped pseudobulge , with much lower velocity dispersion than an r@xmath76 bulge , some rotational support , and stellar populations similar to the surrounding disk is allowed ( although by no means required ) by the observations .
because of the very close association between bars and pseudobulges ( e.g. , * ? ? ?
* ) , it does nt seem tenable to invoke such a feature as an _ alternative _ to a bar ; rather if a boxy structure is present , it would almost certainly be _ additional _ to a bar .
the family resemblance to the milky way is less obvious when it comes to the age - metallicity relation .
there is no obvious amr in the milky way thin disk @xcite , although one does appear to exist in the thick disk @xcite .
our lmc bar sample shows a clear increase in mean metallicity over time , with most of the increase occurring since @xmath76 gyr ago
. however , there _ are _ a few metal - rich stars among the apparently old stars of the sample , as in the milky way disk . as in the milky way
, our data imply a cosmic abundance scatter of @xmath5 @xmath70.15 dex at given age ; the appearance of higher scatter at old ages is attributed to the rapid pace of enrichment in the youth of the galaxy .
a further point of comparison , the possible existence of a radial metallicity gradient in the lmc disk , will be addressed in a future paper . until radial velocity and detailed abundance analyses of sufficient sample size and precision are available
, it will remain uncertain how far parallels between the milky way and lmc disks can be taken .
attempts to understand the lmc s morphology and star formation history in terms of its status as a satellite of the milky way have a long history ( e.g. , * ? ? ?
it has long been appreciated that tidal interactions probably have a leading part in determining the star formation history of both galaxies @xcite , as well as the internal structure of the lmc @xcite .
it is instructive to compare our results to the predictions of recent gasdynamical n - bdoy simulations of the interaction between the milky way / lmc / smc triplet @xcite .
these models predict that the first era of strong interaction between the two clouds occurred @xmath767 gyr ago ; this resulted in the tidal capture of the smc by the lmc , produced the high surface brightness bar of the lmc , and initiated an epoch of enhanced star - formation .
this epoch of intense activity culminated in a violent collision between the clouds @xmath73.6 gyr ago , creating the generation of 13 gyr old star clusters and raising the mean metallicity of the lmc by a factor of @xmath83 during this time .
the @xcite simulations therefore predict that the field stars have a broader age distribution than the clusters , that the intermediate - age populations are centrally concentrated to the lmc bar , and that the metallicity began to increase rapidly between 36 gyr ago .
these predictions are borne out by the picture of the lmc s history that has been built up in @xcite and this paper . in this picture of a tidal origin for the lmc bar and intermediate - age clusters , a radial abundance gradient should exist , because the younger , more metal - rich populations are centrally concentrated . @xcite also predict that a small but non - negligible fraction of stars older than the first epoch of strong smc - lmc interaction will be scattered into a halo - like distribution with velocity dispersion @xmath7740 km s@xmath3 , and with higher metallicity and lower age than the milky way halo .
we will explore these issues , through direct comparison of the kinematics , mdf and amr in the bar to the inner disk ( paper ii ) and other outer disk fields ( cole et al .
, in preparation ) , in a future paper
. the next step towards a complete view of the history of the large magellanic cloud is to combine the abundance information gathered here with the deep cmd data we have already obtained with hst to self - consistently model the star formation history and chemical evolution of this galaxy .
this work is based on observations collected at the european southern observatory , chile , under proposal number 70.b0398 .
preimaging data were taken in service mode , thanks to the efforts of the paranal science operations staff .
aac would like to thank t. szeifert , e. mason , f. clarke , and p. gandhi for their support and assistance during the observing run .
thanks to i. prez - martn for helpful discussions about bars and pseudobulges .
aac is supported by a fellowship from the netherlands research school for astronomy ( nova ) .
et gratefully acknowledges support from a fellowship of the royal netherlands academy of arts and sciences ( knaw ) .
support for tsh s research was provided by the national science foundation through grant ast-0070895 .
this publication makes use of data products from the two micron all sky survey , which is a joint project of the university of massachusetts and ipac / caltech , funded by nasa and the nsf .
alcock , c. , et al .
2004 , , 127 , 334 armandroff , t.e .
, & da costa , g.s .
1991 , , 101 , 1329 armandroff , t.e .
, & zinn , r. 1988 , , 96 , 92 bekki , k. , couch , w.j . ,
beasley , m.a . ,
forbes , d.a . ,
chiba , m. , & da costa , g.s .
2004 , , 610 , l93 bensby , t. , feltzing , s. , & lundstrm , i. 2004 , , 421 , 969 bessell , m.s . , castelli , f. , & plez , b. 1998 , , 333 , 231 bica , e. , clara , j.j .
, & dottori , h. 1992 , , 103 , 1859 binney , j.j . ,
dehnen , w. , & bertelli , g. 2000 , , 318 , 658 blair , w.p . , et al .
2000 , , 537 , 667 bonatto , c. , bica , e. , & girardi , l. 2004 , , 415 , 571 bosler , t.l .
2004 , ph.d .
thesis , univ .
california , irvine bosler , t.l .
, smecker - hane , t.a . ,
cole , a.a . , & stetson , p.b .
2004 , in origin and evolution of the elements , carnegie obs .
astrophys . ser . , eds . a. mcwilliam & m. rauch , ( pasadena : carnegie obs . ) , 5 butcher , h. 1977 , , 216 , 372 carretta , e. , & gratton , r.g .
1997 , , 121 , 95 ( cg97 ) carretta , e. , cohen , j.g . ,
gratton , r.g .
, & behr , b.b .
2001 , , 122 , 1469 cenarro , a.j . , gorgas , j. , cardiel , n. , vazdekis , a. , & peletier , r.f .
2002 , , 329 , 863 cohen , r.s . , dame , t.m . , garay , g. , montani , j. , rubio , m. , & thaddeus , p. 1988
, , 331 , 95 cole , a.a .
, smecker - hane , t.a .
, & gallagher , j.s .
2000 , , 120 , 1808 ( paper i ) cole , a.a .
, smecker - hane , t.a .
, & mandushev , g. 2002 , in `` observed hr diagrams and stellar evolution '' , asp conf . ser . 274 , eds . t. lejeune & j. fernandes , ( asp : san francisco ) , p. 497
cole , a.a .
, smecker - hane , t.a . ,
tolstoy , e. , bosler , t.l . , & gallagher , j.s .
2004 , , 347 , 367 ( paper iii ) cutri , r.m .
2003 , `` explanatory supplement to the 2mass all sky data release '' , http://www.ipac.caltech.edu/2mass/releases/allsky/doc/explsup.html da costa , g.s .
1991 , in iau symp . 148 , the magellanic clouds , ed .
r. haynes & d. milne ( dordrecht : kluwer ) , p. 183 de vaucouleurs , g. , & freeman , k.c .
1972 , vistas astron .
, 14 , 163 dopita , m.a . , et al .
1997 , , 474 , 188 dottori , h. , bica , e. , clara , j.j . , & puerari , i. 1996 , , 461 , 742 fitzpatrick , e.l .
, ribas , i. , guinan , e.f . ,
maloney , f.p .
, & a. claret 2003 , , 587 , 685 freedman , w.l . , et al .
2001 , , 553 , 47 freeman , k. , & bland - hawthorn , j. 2002 , , 40 , 487 freeman , k.c . , illingworth , g. , & oemler , a. jr . 1983 , , 272 , 488 friel , e.d . , janes , k.a . ,
tavarez , m. , scott , j. , katsanis , r. , lotz , j. , hong , l. , & miller , n. 2002 , , 124 , 2693 gallagher , j.s .
iii , mould , j.r . , de feijter , e. , holtzman , j. , stappers , b. , watson , a. , trauger , j. , & the wfpc2 idt 1996 , , 466 , 732 gardiner , l.t . 1999 , in iau symp .
190 , new views of the magellanic clouds , eds .
chu , n.b .
suntzeff , j.e .
hesser , & d.a .
bohlender ( asp : san francisco ) , 480 gardiner , l.t . , sawa , t. , & fujimoto , m. 1994 , , 266 , 567 geha , m.c . , et al . 1998 , , 115 , 1045 geisler , d. , bica , e. , dottori , h. , clari , j.j . , piatti , a.e . , &
santos jr . , j.f.c .
1997 , , 114 , 1920 girardi , l. , bressan , a. , bertelli , g. , & chiosi , c. 2000 , , 141 , 371 gratton , r.g . ,
bragaglia , a. , clementini , g. , carretta , e. , di fabrizio , l. , maio , m. , & taribello , e. 2004 , , 421 , 937 hansen , b.m.s . , & liebert , j. 2003 , , 41 , 465 hayashi , c. , hshi , r. , & sugimoto , d. 1962 , prog .
, supp . , 22 hill , v. , franois , p. , spite , m. , primas , f. , & spite , f. 2000 , , 364 , l19 .
hodge , p.w .
1972 , , 84 , 365 hodge , p.w .
, & wright , f.w .
1967 , `` the large magellanic cloud '' , smithsonian pub . 4699 , ( smithsonian pr .
: washington , d.c . )
holtzman , j.a .
, gallagher , j.s . , cole , a.a .
, mould , j.r . , grillmair , c.j . ,
& the wfpc2 idt 1999 , , 118 , 2262 hoyle , f. , & schwarzschild , m. 1955 , , 2 , 1 hughes , s.m.g . ,
wood , p.r . , & reid , n. 1991 , , 101 , 1304 hurley - keller , d. , mateo , m. , & nemec , j. 1998 , , 115 , 1840 irwin , m. , & tolstoy , e. 2002 , , 336 , 643 johnson , j.a . , bolte , m. , hesser , j.e . , ivans , i.i . , & stetson , p.b .
2004 , in `` origin and evoltuion of the elements '' , carnegie observatories astrophysics series , eds .
a. mcwilliam & m. rauch , ( pasadena : carnegie obs . ) , p. 29
jrgensen , u.g . , carlsson , m. , & johnson , h.r .
1992 , , 254 , 258 kauffmann , g. , et al .
2003 , , 341 , 54 kim , s. , staveley - smith , l. , dopita , m.a . ,
freeman , k.c . , sault , r.j . ,
kesteven , m.j . , & mcconnell , d. 1998 , , 503 , 674 koorneef , j. 1982 , , 107 , 247 kormendy , j. , & kennicutt , r.c . ,
2004 , , 42 , 603 kotoneva , e. , flynn , c. , chiappini , c. , & matteuci , f. 2002 , , 336 , 879 lasker , b.m .
1978 , , 223 , 109 leggett , s.k . ,
ruiz , m.t .
, & bergeron , p. 1998
, , 497 , 294 luck , r.e . , moffett , t.j . ,
barnes iii , t.g . , & gieren , w.p .
1998 , , 115 , 605 luks , t. , & rohlfs , k. 1992 , , 263 , 41 mcwilliam , a. 1990 , , 74 , 1075 minniti , d. , borissova , j. , rejkuba , m. , alves , d.r . ,
cook , k.h . , & freeman , k.c .
2003 , science , 301 , 1508 morell , o. , kllender , d. , & butcher , h.r .
1992 , , 259 , 543 murai , t. , & fujimoto , m. 1980 , , 32 , 581 nikolaev , s. , & weinberg , m.d .
2000 , , 542 , 804 olszewski , e.w .
1993 , in `` the globular cluster - galaxy connection '' , asp conf .
48 , eds .
smith & j.p .
brodie , ( asp : san francisco ) , p. 351
olszewski , e.w . ,
schommer , r.a . ,
suntzeff , n.b . , & harris , h.c .
1991 , , 101 , 515 ( ossh ) osterbrock , d.e . , &
martel , a. 1992 , pasp , 104 , 76 osterbrock , d.e . ,
fulbright , j.p . ,
martel , a.r . ,
keane , m.j . ,
trager , s.c . , & basri , g. 1996 , , 108 , 277 pagel , b.e.j . , & tautvaiien , g. 1998 , , 299 , 535 ( pt98 ) peletier , r.f . , & balcells , m. 1996 , , 111 , 2238 reid , i.n .
1999 , , 37 , 191 rolleston , w.r.j . ,
trundle , c. , & dufton , p.l .
2002 , , 396 , 53 rutledge , g.a . ,
hesser , j.e . ,
stetson , p.b . , mateo , m. , simard , l. , bolte , m. , friel , e.d . , & copin , y. 1997a , , 109 , 883 rutledge , g.a . ,
hesser , j.e . , & stetson , p.b .
1997b , , 109 , 907 salaris , m. , chieffi , a. , & straniero , o. 1993 , , 414 , 580 salaris , m. , weiss , a. , & percival , s.m .
2004 , , 414 , 163 sanduleak , n. 1970 , contrib .
cerro tololo obs . , 89 scalo , j.m .
1987 , in `` starbursts and galaxy evolution '' , eds .
thuan , t. montmerle , & j.t .
thanh vn ( paris : editions frontires ) , 445 schommer , r.a .
, olszewski , e.w . ,
suntzeff , n.b .
, & harris , h.c .
1992 , , 103 , 447 searle , l. , wilkinson , a. , & bagnuolo , w. 1980 , , 239 , 803 shen , j. , & sellwood , j.a .
2004 , , 604 , 614 skillman , e.d . , &
gallart , c. 2002 , in `` observed hr diagrams and stellar evolution '' , asp conf .
274 , eds .
t. lejeune & j. fernandes , ( asp : san francisco ) , p. 535
smecker - hane , t.a .
, gallagher , j.s .
iii , cole , a.a . ,
tolstoy , e. , & stetson , p.b .
1999 , in iau symp . 190 , new views of the magellanic clouds , eds .
chu , n.b .
suntzeff , j.e .
hesser , & d.a .
bohlender ( san francisco : asp ) , 343 smecker - hane , t.a . ,
mandushev , g.i . ,
hesser , j.e . ,
stetson , p.b .
, da costa , g.s . , & hatzidimitriou , d. 1999 , in spectrophotometric dating of stars and galaxies , asp conf .
192 , eds .
i. hubeny , s.r .
heap , & r.h .
cornett , ( san francisco : asp ) , 159 smecker - hane , t.a . ,
cole , a.a . ,
gallagher , j.s .
iii , & stetson , p.b .
2002 , , 566 , 239 smecker - hane , t.a . , cole , a.a . ,
mandushev , g.i . ,
bosler , t. , & gallagher , j.s .
iii 2004 , , submitted ( paper ii ) smith , v.v .
, et al .
2002 , , 124 , 3241 sparke , l.s . , & sellwood , j.a .
1987 , , 225 , 653 spergel , d.n . , et al .
2003 , , 148 , 175 staveley - smith , l. , kim , s. , calabretta , m.r . ,
haynes , r.f . , & kesteven , m.j .
2003 , , 339 , 87 sweigart , a.v . , &
gross , p.g .
1978 , , 36 , 405 tolstoy , e. , irwin , m.j . ,
cole , a.a . , pasquini , l. , gilmozzi , r. , & gallagher , j.s .
2001 , , 327 , 918 tolstoy , e. , venn , k.a . , shetrone , m. , primas , f. , hill , v. , kaufer , a. , & szeifert , t. 2003 , , 125 , 707 tonry , j. , & davis , m. 1979 , , 84 , 1511 udalski , a. , szymaski , m. , kubiak , m. , pietrzyski , g. , soszyski , i. , woniak , p. , & ebru , k. 2000 , , 50 , 307 vandenberg , d.a . , richard , o. , michaud , g. , & richer , j. 2002 , , 571 , 487 van der marel , r.p .
2001 , , 122 , 1827 van der marel , r.p . ,
alves , d.r . , hardy , e. , & suntzeff , n.b .
2002 , , 124 , 2639 weinberg , m.d .
2000 , , 532 , 922 weisstein , e.w .
1999 , `` albers equal - area conic projection '' , from _
mathworld a wolfram web resource _ , http://mathworld.wolfram.com/albersequal-areaconicprojection.html westerlund , b.e .
1997 , `` the magellanic clouds '' , ( cambridge univ . pr . : cambridge ) worthey , g. 1999 , in spectrophotometric dating of stars and galaxies , asp conf .
192 , eds .
i. hubeny , s. heap , & r. cornett ( san francisco : asp ) , 283 zaritsky , d. 2004 , , 614 , l37 zhao , h.s . ,
ibata , r.a . ,
lewis , g.f . , & irwin , m , j .
2003 , , 339 , 701 lcccccl ngc 1904 & 05@xmath1724@xmath312 & @xmath024@xmath1931@xmath78 & 2002 - 12 - 25/01:13 & 0.7 & 17 & calibrator : [ fe / h ] = @xmath01.37 + ngc 104 & 00@xmath1726@xmath315 & @xmath071@xmath1951@xmath78 & 2002 - 12 - 25/01:25 & 0.7 & 8 & calibrator : [ fe / h ] = @xmath00.70 + lmc bar 1/a & 05@xmath1723@xmath310 & @xmath069@xmath1953@xmath78 & 2002 - 12 - 25/01:37 & 0.6 & 19 & + lmc bar 1/b & & & 2002 - 12 - 25/02:12 & 0.6 & 19 & + lmc bar 1/m & & & 2002 - 12 - 25/02:42 & 0.8 & 16 & + lmc bar 3/e & 05@xmath1724@xmath312 & @xmath069@xmath1945@xmath78 & 2002 - 12 - 25/03:16 & 0.9 & 19 & + lmc bar 3/f & & & 2002 - 12 - 25/03:46 & 1.0 & 19 & + lmc bar 3/n & & & 2002 - 12 - 25/04:17 & 0.8 & 19 & + melotte 66 & 07@xmath1726@xmath315 & @xmath047@xmath1941@xmath78 & 2002 - 12 - 25/04:48 & 0.7 & 14 & calibrator : [ fe / h ] = @xmath00.47 + berkeley 39 & 07@xmath1746@xmath317 & @xmath004@xmath1941@xmath78 & 2002 - 12 - 25/05:02 & 0.7 & 10 & calibrator : [ fe / h ] = @xmath00.26 + berkeley 20 & 05@xmath1732@xmath316 & @xmath3800@xmath1910@xmath78 & 2002 - 12 - 25/05:16 & 0.8 & 4 & calibrator : [ fe / h ] = @xmath00.61 + lmc bar 6/j & 05@xmath1725@xmath319 & @xmath069@xmath1940@xmath78 & 2002 - 12 - 25/05:40 & 0.8 & 18 & + lmc bar 6/k & & & 2002 - 12 - 25/06:06 & 0.9 & 18 & + lmc bar 6/l & & & 2002 - 12 - 25/06:31 & 0.7 & 18 & + ngc 1851 & 05@xmath1714@xmath312 & @xmath040@xmath1904@xmath78 & 2002 - 12 - 26/00:49 & 0.8 & 15 & calibrator : [ fe / h ] = @xmath00.98 + lmc bar 2/c & 05@xmath1723@xmath310 & @xmath069@xmath1945@xmath78 & 2002 - 12 - 26/01:04 & 0.8 & 17 & + lmc bar 2/d & & & 2002 - 12 - 26/01:28 & 0.8 & 19 & + lmc bar 2/aa & & & 2002 - 12 - 26/01:54 & 0.7 & 18 & + lmc bar 2/ee & & & 2002 - 12 - 26/02:18 & 0.7 & 16 & + ngc 2141 & 06@xmath1703@xmath310 & @xmath3810@xmath1930@xmath78 & 2002 - 12 - 26/04:15 & 1.2 & 15 & calibrator : [ fe / h ]
= @xmath00.33 + ngc 2682 & 08@xmath1751@xmath314 & @xmath3811@xmath1948@xmath78 & 2002 - 12 - 26/06:36 & 0.7 & 7 & calibrator : [ fe / h ] = @xmath00.15 + lmc bar 5/h & 05@xmath1722@xmath317 & @xmath069@xmath1938@xmath78 & 2002 - 12 - 26/06:51 & 0.9 & 18 & + lmc bar 5/i & & & 2002 - 12 - 26/07:15 & 1.0 & 18 & + lmc bar 5/bb & & & 2002 - 12 - 26/07:39 & 0.6 & 18 & + ngc 4590 & 12@xmath1739@xmath315 & @xmath026@xmath1945@xmath78 & 2002 - 12 - 26/08:20 & 0.6 & 9 & calibrator : [ fe / h ] = @xmath01.99 + lmc bar 7/o & 05@xmath1725@xmath313 & @xmath070@xmath1900@xmath78 & 2002 - 12 - 27/01:46 & 1.0 & 16 & + lmc bar 7/p & & & 2002 - 12 - 27/02:10 & 0.9 & 17 & + lmc bar 7/cc & & & 2002 - 12 - 27/02:34 & 1.0 & 18 & + lmc bar 7/dd & & & 2002 - 12 - 27/02:58 & 1.1 & 16 & + lmc bar 4/g & 05@xmath1724@xmath316 & @xmath069@xmath1953@xmath78 & 2002 - 12 - 27/03:27 & 1.1 & 17 & + ngc 2298 & 06@xmath1749@xmath310 & @xmath036@xmath1900@xmath78 & 2002 - 12 - 27/05:15 & 0.9 & 7 & calibrator : [ fe / h ] = @xmath01.74 + ccccccccccc 05230778 - 6950057 & 1/a & 17.648 & 0.003 & 16.063 & 0.007 & 264.8 & 7.6 & 8.55 & 0.12 & + 05231484 - 6950196 & 1/a & 17.486 & 0.004 & 16.120 & 0.004 & 257.0 & 7.4 & 8.60 & 0.26 & + 05225670 - 6950472 & 1/a & 17.350 & 0.009 & 15.860 & 0.005 & 282.7 & 7.5 & 8.60 & 0.13 & on n131 hii region + 05230606 - 6951113 & 1/a & 17.491 & 0.009 & 15.986 & 0.005 & 273.9 & 7.6 & 9.01 & 0.12 & + 05225436 - 6951262 & 1/a & 17.349 & 0.008 & 15.954 & 0.008 & 286.7 & 7.5 & 8.01 & 0.09 & on n131 hii region + cccccc 05230778 - 6950057 & -0.29 & 0.14 & -0.25 & 10.00 & 0.17 + 05231484 - 6950196 & -0.31 & 0.16 & -0.24 & 9.24 & 0.21 + 05225670 - 6950472 & -0.35 & 0.14 & -0.22 & 9.58 & 0.27 + 05230606 - 6951113 & -0.16 & 0.14 & -0.32 & 9.43 & 0.25 + rlccc @xmath00.14 & @xmath380.14 & 19 & 255.0 & 16.7 @xmath51.6 + @xmath00.26 & @xmath00.14 & 56 & 256.9 & 21.5 @xmath51.0 + @xmath00.44 & @xmath00.26 & 147 & 257.7 & 23.7 @xmath50.6 + @xmath00.55 & @xmath00.44 & 77 & 252.4 & 23.9 @xmath50.9 + @xmath00.70 & @xmath00.55 & 37 & 257.5 & 28.0 @xmath51.3 + @xmath01.15 & @xmath00.70 & 19 & 256.2 & 26.6 @xmath51.7 + @xmath02.13 & @xmath01.15 & 18 & 262.5 & 40.8 @xmath51.7 + lcccr rgb , bar , metal - rich & 10005000 & @xmath00.37 & 0.15 & this paper + rgb , bar , metal - poor & @xmath795000 & @xmath01.08 & 0.47 & this paper + b dwarfs & @xmath5820 & @xmath00.31 & 0.04 & @xcite + cepheid variables & 1060 & @xmath00.34 & 0.15 & @xcite + young red giants & 2001000 & @xmath00.45 & 0.10 & @xcite + int .
age clusters & 10003000 & @xmath00.40 & 0.22 & ossh , @xcite + planetary nebulae & 100010@xmath80 & @xmath00.5 & 0.2 & @xcite + rr lyr variables , bar & @xmath1410@xmath80 & @xmath01.23 & 0.29 & @xcite + old clusters & @xmath1410@xmath80 & @xmath01.74 & 0.36 & ossh , @xcite + | we report metallicities and radial velocities derived from spectra at the near - infrared calcium triplet for 373 red giants in a 200 square arcminute area at the optical center of the lmc bar .
these are the first spectroscopic abundance measurements of intermediate - age and old field stars in the high surface brightness heart of the lmc .
the metallicity distribution is sharply peaked at the median value [ fe / h ] = @xmath00.40 , with a small tail of stars extending down to [ fe / h ] @xmath1 @xmath02.1 ; 10% of the red giants are observed to have [ fe / h ] @xmath1 @xmath00.7 .
the relative lack of metal - poor stars indicates that the lmc has a `` g dwarf '' problem , similar to the milky way .
the abundance distribution can be closely approximated by two gaussians containing 89% and 11% of the stars , respectively : the first component is centered at [ fe / h ] = @xmath00.37 with @xmath2 = 0.15 , and the second at [ fe / h ] = @xmath01.08 with @xmath2 = 0.46 .
the dominant population has a similar metallicity distribution to the lmc s intermediate - age star clusters .
the mean heliocentric radial velocity of the sample is 257 km sec@xmath3 , corresponding to the same center of mass velocity as the disk ( measured at larger radii ) . because of the central location of our field , kinematic constraints are not strong , but there is no evidence that the bar deviates from the general motion of the lmc disk . the velocity dispersion of the whole sample is @xmath4 = 24.7 @xmath50.4 km sec@xmath3 . when cut by metallicity ,
the most metal - poor 5% of stars ( [ fe / h ] @xmath61.15 ) show @xmath4 = 40.8 @xmath51.7 km sec@xmath3 , more than twice the value for the most metal - rich 5% ; this suggests that an old , thicker disk , or halo population is present .
the age - metallicity relation ( amr ) is almost flat during the period from 510 gyr ago , with an apparent scatter of @xmath50.15 dex about the mean metallicity for a given age . comparing to chemical evolution models from the literature , we find that a burst of star formation 3 gyr ago does not reproduce the observed amr more closely than a steadily declining star - formation rate .
the amr suggests that the epoch of enhanced star formation , if any , must have commenced earlier , @xmath76 gyr ago the exact time is model - dependent .
we compare the properties of the lmc and the galaxy , and discuss our results in the context of models that attempt to use tidal interactions with the milky way and small magellanic cloud to explain the star and cluster formation histories of the lmc . |
the temperature - flow frg uses the physical temperature of the system as flow parameter . as in previous works using this method@xcite ,
the flow of the self - energy is neglected and the flow equations are truncated after the four - point vertex . furthermore , the interaction vertex is approximated to be frequency - independent , and the external frequencies are set to zero .
the wavevector dependence is treated in the @xmath78-patch discretization@xcite . the initial condition for the flow at the initial @xmath8 larger than the bandwidth of the system is the bare action , as all perturbative corrections vanish at high enough @xmath8 .
then the flow leads to lower @xmath8 and picks up the different one - loop corrections for the interactions , each in second order in the vertices at the given temperature .
more precisely , the flow equations read in this case@xcite @xmath79 where explicitly @xmath80 @xmath81 \end{aligned}\ ] ] @xmath82 the @xmath83 indices are patch numbers , corresponding to wavevectors in the conduction band . in the temperature flow without self - energy corrections , the one - loop integrals are given by @xmath84 in most situations , the usual momentum - shell cutoff schemes which have a cutoff - derivative in the loop diagrams give equivalent flows as the tfrg . in the case of particle - hole fluctuations with small wavevector transfer
, a momentum - shell cutoff would lead to underestimation of these processes at intermediate scales while the @xmath8-flow incorporates these contributions in the same way as fluctuations with larger wavevector transfers .
hence the @xmath8-flow can capture ferromagnetic instabilities and is hence a more unbiased frg - method .
note that also frequency cutoff can be used to circumvent this shortcoming of the momentum - shell schemes@xcite . | we investigate superconducting pairing driven by electron - electron interactions in a theoretical model for monolayer mos@xmath0 with the temperature - flow functional renormalization group(frg ) . at low doping
, the dominant instability is toward odd - parity pairing with @xmath1-wave mo - nearest - neighbor structure .
we compute the frg phase diagram versus electron doping below the van hove filling of the conduction band . in the superconducting regime
, the critical temperature grows with doping , comparable to the experiments . near van hove filling the system favors a ferromagnetic state .
we demonstrate that the triplet pairing is driven by ferromagnetic fluctuations and that the multiorbital nature of the conduction band as well as the hund s coupling appear crucial in making the physics of mos@xmath0 different from e.g. doped graphene .
superconductivity with critical temperatures around 10k has been reported in mos@xmath0 flakes by electron doping using combined ionic - liquid and solid - state gating@xcite .
the observed superconducting dome as function of doping suggests the possibility of unconventional pairing .
as mo is a transition metal and the electronic bands near the fermi level in mos@xmath0 have predominant @xmath2-orbital weight , electron - electron interactions are likely to play a role .
monolayer mos@xmath0 in the 2h - structure can be regarded as a sibling of of graphene .
compared to graphene with its two c sublattices , in mos@xmath0 , one sublattice site is occupied by mo , while the other sublattice is formed by two s atoms displaced above and below the mo plane .
the layered 2h - structure is a direct gap semiconductor with gap size @xmath3ev@xcite .
bulk mos@xmath0 has an indirect gap@xcite . upon electron doping
, the mo-@xmath2-dominated conduction band develops pockets around the @xmath4 and @xmath5-points . by continued doping
, these pockets merge with small extra pockets around the @xmath6-points . at the merging points , the density of states ( dos ) exhibits van - hove singularities .
the possible nature of superconductivity in mos@xmath0 is of high interest to theory@xcite .
the work by roldan et al.@xcite compares pairing due to electron - electron interactions and mediated by phonons using effective intra- and inter - pocket interactions between the pockets that form in the conductions band around the @xmath4 and @xmath5 points upon doping .
their conclusion is that phonon - mediated interactions have a smaller effect , and that the resulting pairing is driven by the short - ranged coulomb interactions , and comes out as odd parity .
topological @xmath7-wave states due to electronic interactions are also reported by a mean - field theory@xcite . a purely phonon - mediated mechanism based on first principles
is proposed in refs . .
the purpose of this paper is extend the study of ref . and to explore possible ground states of mos@xmath0 using a purely electronic model .
we apply a temperature - flow functional renormalization group ( frg ) technique@xcite to a three - orbital tight - binding model for mos@xmath0 proposed in ref . .
the @xmath8-flow frg is an unbiased approach and sufficiently flexible to treat the doping - dependent fermi surfaces with enough angular resolution such to disentangle different pairing types .
importantly in our context , it is not blind with respect to ferromagnetic fluctuations@xcite .
furthermore , we use as input effective interaction parameters obtained by constrained random - phase - approximation ( crpa ) calculations .
hence , in addition to confirming the qualitative picture for the paired state of ref . , we give an ab - initio - based estimate of the critical temperatures as function of the doping , unveil the origin of electronic pairing as mediated by ferromagnetic fluctuations and the importance of hund s rule coupling , and give an overview of the competing phases .
this suggests experimental searches for ferromagnetic correlations in mos@xmath0 and related systems . .
inset : schematic view of crystal structure of monolayer mos@xmath0 .
the purple sphere represents the mo , the yellow spheres sulfur atoms .
right : electronic density of states ( dos ) of the 3-band model .
the red line shows the total dos with the van - hove singularity at @xmath9ev .
the blue line is the band - projected dos for the @xmath10-orbital , while @xmath11- and @xmath12-contributions are shown by the coinciding solid purple and green dashed lines , respectively .
, width=302 ] for the low - energy physics in mos@xmath0 we use a three - band tight - binding model from ref . .
in the localized state basis @xmath13 the tight - binding model hamiltonian is given by a @xmath14 matrix .
for simplicity , we only consider nearest - neighboring hopping terms with the parameters of ref .
@xcite , resulting in bands as plotted in fig .
1(a ) .
spin - orbit coupling will be discussed more below .
doped electrons enter first into the middle band around @xmath15 , forming fermi pockets . in fig.1(b ) , we plot the density of states ( dos ) of this band .
it exhibits a van hove ( vh ) singularity around @xmath16ev , which is when the extending fermi pockets touch developing small pockets around @xmath6 . in this paper , we focus on smaller doping levels @xmath17 up to van - hove filling .
around @xmath4 , @xmath5 , the doped electrons have predominant @xmath18 character , while near the vh level , the @xmath11 and @xmath12 components are stronger , as can be seen from fig .
1(b ) . in the dft band structures and more elaborate tight - binding parametrizations@xcite
, additional band minima along the @xmath19-@xmath4 direction not captured in our simplified model can lead to additional fermi pockets at higher @xmath17 .
these however have smaller dos than e.g. the @xmath6-point regions and should therefore be of minor importance for the trends found in this work .
the interactions are given by @xmath20 with @xmath21ev , @xmath22ev , @xmath23ev , computed from first principles by crpa @xcite . in a broader view , besides the minor splitting of the van hove singularities near the @xmath6-points , the fermi surface evolution in our model for mos@xmath0 is reminiscent of that in doped graphene ( e.g. studied in @xcite ) .
the natural question arises if the phenomenology is similar .
we we argue that the multi - orbital nature due to the mixing of 3 @xmath2-orbitals at the fermi level in the conduction band causes qualitative differences , at least for model parameters close to the ab - initio predictions .
we in fact show that the hund s coupling is a tuning parameter that takes us from the doped - graphene scenario to mos@xmath0 .
we use the temperature - flow frg@xcite to analyze the possible instabilities due to interactions .
the main idea of the method is that at very high temperatures @xmath8 the interactions of the model are given by the bare ones , as all loop corrections are killed by @xmath8 . upon lowering @xmath8
, the @xmath8-flow tracks the build - up of the loop corrections .
the main object of study is an interaction function @xmath24 that depends in our case on three wavevectors @xmath25 , @xmath26 and @xmath27 for two incoming legs 1 and 2 and one outgoing leg 3 ( with the same spin projection as leg 1 ) in the conduction band .
the flow equations for @xmath28 for charge and spin - rotational invariance are taken from the general frg formalism for one - particle irreducible vertex functions ( for a review , see refs .
@xcite or also ref . ) and are given in the supplementary material .
we start the flow at an initial @xmath8 of roughly twenty times the bandwidth . lowering @xmath8 generally leads to a strong differentiation of the components of @xmath28 , where certain components become large at low @xmath8 and
actually diverge .
these generalized cooper instabilities can then be interpreted as ordering tendency in a particular channel , depending on which combination of @xmath29 , @xmath30 etc .
actually grows most strongly .
the flow is stopped when the maximal component of @xmath31 grows larger than thirty times larger than the bandwidth .
this defines a critical temperature @xmath32 which should be understood as upper estimate for any true ordering .
more information about the interpretation of the frg flow can e.g. be found in the reviews of refs . @xcite and @xcite . in a true two - dimensional system ,
collective effects should in fact destroy any long range order that breaks continuous symmetries .
these effects are however not included in the used approximation in the frg , so do not play a role in our results . also in view of the embedding of the experimental systems in three dimensions
we do not expect that such collective effects have to be considered at this stage .
further , we patch brillouin zone around the @xmath4 , @xmath5 points with patch center points numbered by @xmath33 , as seen in fig .
[ fig2 ] .
then we discretize the coupling function as @xmath34 , i.e. with @xmath35 components . at each corner
@xmath4 , @xmath5 , we take 5 patches ( schemes with more patches have also been checked , the result is stable ) .
we divide the radial direction of each patch into three sectors , providing some more resolution in the radial direction . in the loop calculations , we use 3 radial integration lines in each patch , but did also checks with more lines .
the fermi surface ( fs ) is around the @xmath4 , @xmath5 points . on the fs
we define 30 patch points ( grey filled circles ) .
the patch index @xmath36 runs from @xmath37 to @xmath38 .
in addition to these fs points , we define two other sets of patch points ( blue and pinks circles ) .
this way , the energy dispersion is monotonic in each sector in radial direction .
right : pair scattering at small @xmath17 .
the green symbols in the main plot show the frg data for the low - t pair scattering @xmath39 near the instability , where the two incoming wavevectors belonging to index @xmath36 and @xmath40 add up to total zero momentum .
the first outgoing wavevector is fixed at patch point 2 .
the red square symbols show a fit to the frg data using a nearest - neighbor-(i.e .
mo - mo-)bond pairing @xmath41 ) f_p^*(\mathbf k[2])$ ] .
inset : @xmath42 as function of patch indices @xmath43 and @xmath44 for @xmath45 .
clearly visible is that only interactions with a special relation between @xmath43 and @xmath44 ( implying zero total incoming wavevector ) become large . for the data in this plot , @xmath46ev .
, width=325 ] we run the @xmath8-flow frg for a wider range of electron dopings @xmath17 . for small @xmath17 , we find pairing instabilities with total - momentum - zero cooper pairs in the odd - parity channel , while at larger @xmath17 near van hove filling , a ferromagnetic ( fm ) instability is found .
the latter can be understood as a renormalized version of a stoner instability due to a large dos at the fermi level .
the data for a prototypical pairing instability at @xmath47 is shown in fig .
[ fig2 ] on the right side .
the inset shows the coupling function for incoming and outgoing legs at the fermi surface , as function of the two incoming patch indices , with the first outgoing leg fixed in patch 2 .
the two diagonal features are exactly on the lines where the two incoming wavevectors add up to zero .
these pair - scattering processes grow to very large absolute values and cause the instability .
we also see the sign structure with pieces of large positive and large negative values .
this is detailed more in the main part of fig .
[ fig2 ] , right side , which is basically a cut along the zero - total - momentum diagonal features in the inset . by looking up
the patch numbers fig . 1 one extracts that the pairing changes sign upon rotating the incoming pair by 60 degrees , from one fermi pocket around , e.g. around @xmath4 , to another pocket , e.g. around @xmath5 .
the same sign change is found when the incoming pair is fixed and the outgoing pair is varied .
the couplings do not vary much around @xmath15 points , i.e. is no nodal structure along the fermi pockets .
this odd - parity pairing can be identified to transform according to the one - dimensional @xmath48 irreducible representation of @xmath49 point group of mos@xmath0 . in real space
it can be understood as having its main contributions on the bonds from one mo site to its 6 nearest neighbor mo sites , with the sign alternating upon 60 degrees rotation , implying the odd parity .
the form factor in wave vector space can also be interpreted as @xmath1-wave and reads @xmath50 .
the fit to this form factor shows good agreement ( red squares and green circles in fig.3 ) .
we also run a momentum - cutoff frg flow@xcite which disfavors fm fluctuations with small wave - vector transfer across the fermi surface down to very low scales@xcite .
these flows do not give any instabilities but show some tendency to even - parity pairing .
therefore , we can conclude that @xmath1-wave pairing is driven by @xmath51 ferromagnetic fluctuations .
a similar conclusion is reached by turning off specific channel spin fluctuation channels in the flow equation directly . in fig .
[ fig4 ] we show the frg phase diagram of our mos@xmath0 model with doping . in the low doping regime , the discussed @xmath1-wave superconductivity is the dominant instability , with a relatively low critical temperature @xmath32 ( of order @xmath52mev ) .
we can also see that the critical temperature grows with increasing electron doping .
the highest scale for pairing for our parameters is around @xmath53mev , considerably higher than the experimental data around @xmath54mev .
we comment on this further below .
the optimal doping for the @xmath1-wave pairing is reached around @xmath55 .
this shows good agreement with experiment@xcite .
the increasing dos with growing @xmath17 also enhances the fm fluctuations . in simple mean - field or stoner pictures the critical interaction strength required for fm order is inversely proportional to the dos at the fermi level , and as the dos diverges at the van hove filling , one will definitely find ordering in these approaches . in the frg calculations we also obtain a regime where the leading instability is of fm type , however in reduced extent compared to mean - field theory due to the competition with other fluctuations .
hence , the prediction form the frg approach would be that there are predominant fm fluctuations around the van hove filling . whether the ordering actually occurs or not can not be answered by our calculations . for the hubbard model on the square lattice at large second - nearest - neighbor hopping
, the frg also predicts fm ordering , in agreement with variational montecarlo results@xcite .
however , the van hove filling may be hard to realize precisely in the experiment as disorder and spatial fluctuations will tend to smear out the diverging dos .
nevertheless , the fm correlations should still be visibly enhanced . versus electron doping @xmath17 .
the blue dots give the @xmath32s according to frg for the @xmath1-wave superconducting phase ( * sc * ) .
the purple dots represent the @xmath32s ferromagnetic state ( * fm * ) , now scaled down by a factor of 100 .
the frg @xmath32s should be interpreted as upper bounds for true ordering temperatures .
right : frg critical temperatures as function of hund s coupling @xmath56 for both the triplet pairing ( blue circles , ) and fm instability ( pink circles , again multiplied by 0.01 ) .
notably , for @xmath57 the dominant pairing instability appears in the @xmath2-wave singlet pairing channel .
, width=302 ] at small @xmath17 the states near the fermi level are mainly of @xmath18-type .
thus the local intraorbital repulsion , @xmath58 , should be important .
we hence run the flow with only @xmath58 being nonzero .
as expected , the flow still ends up with the @xmath1-wave instability .
however , @xmath32 is lowered from about 4.8mev with full interactions down to @xmath59mev .
this shows that interorbital processes contribute significantly to the pairing .
next we vary the hund s coupling @xmath56 in the interactions as given in eq .
( 3 ) , with @xmath58 and @xmath60 fixed , and the doping fixed at @xmath61ev . from fig . 4
, we can read off an enhancement of @xmath32 in the @xmath1-wave regime by a larger @xmath56 , from @xmath62mev to around @xmath53mev .
thus , @xmath56 enhances the @xmath1-wave instability as well .
even more strikingly , when we set the @xmath63ev , the system gives way to an even - parity pairing structure , with most attractive eigenvalues in the @xmath2-wave symmetry channel@xcite , associated with the two - dimensional @xmath64 irreducible representation of @xmath49 point group .
complex combinations of @xmath12 and @xmath11 break time - reversal symmetry , leading to chiral edge modes as known from strongly doped graphene models@xcite .
we also study the importance of the hund s coupling @xmath56 in the fm regime .
the data is shown as pink dots in fig .
5 .
the critical temperature is again enhanced dramatically when @xmath56 is increased .
the hund s coupling intensifies the fm fluctuations just as it strengthens the @xmath1-wave pairing .
such an effect of @xmath56 also reported in models for naco@xmath65o@xcite .
we also observe a kink in the critical temperature as a function of @xmath56 at @xmath66ev where the system undergoes a phase transition from even - parity pairing at small @xmath56 to odd - parity pairing .
hence @xmath56 has a key role as a switch between two different situations . on the one hand
, we have the single - orbital scenario of graphene , where @xmath56 is ineffective as the bands near the fermi level are just made from c @xmath67-states .
then one finds strong antiferromagnetic fluctuations and chiral @xmath2-wave singlet pairing . for mos@xmath0 ,
the multi - orbital character of the conduction band makes @xmath56 important and , for realistic values of @xmath56 according to crpa , one finds predominant fm fluctuations and triplet pairing .
these considerations indicate that the superconducting ground state is quite susceptible to parameter changes .
this is one reason why the rather large pairing scales @xmath68mev exceeding the experimental scales by one order of magnitude should not be taken as a failure of this theoretical picture .
some reduction of the @xmath32s should be expected if we would be able to include self - energy corrections or more collective fluctuations into the frg , which is however not yet technically feasible for multiband systems . in view of the importance of the hund s rule coupling
, we refer to recent works by georges et al.@xcite , who pointed out that @xmath56 typically degrades the quasiparticle weight resulting in a bad _ hund s metal_. this may be another important factor that reduces the superconducting @xmath32s compared to our study .
then there are uncertainties about the actual experimental system .
disorder is expected to have an effect on the @xmath32 of any unconventional superconductor , although in the @xmath1-wave case only large - momentum impurity scattering will have an impact of the phase structure of the pairing . on the positive side , the high scales from the theory
might actually show that there may be some headroom for reaching higher @xmath32s in mo@xmath69 and related systems . in mos@xmath0 and ws@xmath0 ,
the spin - orbit coupling splitting can reach the order @xmath70ev @xcite . in the conduction band near @xmath4 , @xmath5 , the actual values may be smaller @xmath71mev@xcite .
the @xmath72-component is conserved around the small fermi pockets shown in fig .
2 which then split into two pockets around each @xmath4 , @xmath5 . by time - reversal symmetry , the pockets with the same radius at different pockets will have opposite spin projections .
the @xmath1-wave pairing does not violate time - reversal symmetry@xcite and should not be changed strongly by the small splitting near @xmath4 , @xmath5 .
in such a case , the @xmath73 component of the gap function @xmath74 can be expected to be dominant .
this corresponds to @xmath75 pairing . in conclusion
, we investigate the possible pairing structure in monolayer - mos@xmath0 by the unbiased @xmath8-flow frg . for interaction parameters obtained by crpa ,
the leading instability at low electron doping lies in the @xmath1-wave pairing channel .
we argued that ferromagnetic fluctuations provide the pairing glue .
the phase diagram exhibits dominant @xmath1-wave pairing up to @xmath76 . for doping close to the van - hove filling
, sufficiently strong interactions result in fm order .
the pairing @xmath32 is maximal around an optimal doping of @xmath77 , in quite a good agreement with experiment .
varying the hund coupling strength @xmath56 showed strong effects on both the @xmath1-wave pairing scale and the fm tendencies .
notably , when @xmath56 is relatively small or set to zero , the leading pairing instability is in the singlet @xmath2-wave channel .
this emphasizes the importance of the hund s coupling and the multi - orbital nature of the mo-@xmath2-dominated conduction band .
our study extends earlier work by roldan et al.@xcite who already found @xmath1-wave pairing .
it clarifies the nature of the pairing mechanism , the influence of the model parameters , the importance of fm fluctuations and the hund s coupling .
future studies should include the phonon - mediated electronic interactions .
this should allow one to compare the relative importances of these two distinct pictures for the superconductivity in mos@xmath0 .
we acknowledge useful discussions with ersoy aiolu , tim wehling , stefan haas and fu - chun zhang and support by dfg - for 912 and spp1459 , as well as dfg - ho2422/10 - 1 .
j. t. ye , y. j. zhang , r. akashi , m. s. bahramy , r. arita , y. iwasa , science * 338 * , 1193(2012 ) k. taniguchi , a. matsumoto , h. shimotani , h. takagi , appl .
phys .
lett * 101 * , 042603(2012 ) r. roldn , e. cappelluti , f. guinea , phys .
rev .
b * 88 * , 054515(2013 ) y. ge , a. y. liu , phys . rev .
b * 87 * , 241408(r)(2013 ) m. r " osner , s. haas , t. o. wehling , arxiv : 1404.4295v1 t. das , k. dolui , phys . rev .
b * 91 * , 094510(2015 ) n. f. q. yuan , k. f. mak , k. t. law , phys . rev . lett * 113 * , 097001(2014 ) z. y. zhu , y. c. cheng , u. schwingenschlgl , phys .
rev .
b * 84 * , 153402(2012 ) g. liu , w. shan , y. yao , w. yao , d. xiao , phys . rev .
b * 88 * , 085433(2013 ) l. f. mattheiss , phys . rev .
lett * 30 * , 784(1973 ) s. lebegue and o. eriksson , phys
. rev .
b * 79 * , 115409 ( 2009 ) .
i. v. solovyev , m. imada , phys .
rev .
b * 71 * , 045103(2005 ) e. aiolu , c. friedrich , s. blgel , phys . rev .
b * 83 * , 121101(r)(2011 ) private communication with e. aiolu c. honerkamp , m. salmhofer , phys . rev .
b * 64 * , 184516 w. metzner , m. salmhofer , c. honerkamp , v. meden , k. schnhammer , rev . mod .
phys * 84 * , 299(2012 ) christian platt , werner hanke , ronny thomale , adv .
phys .
62 , 453 - 562 ( 2013 ) n. fukukawa , t. m. rice , m. salmhofer , phys .
rev .
lett * 81 * , 3195(1998 ) r. nandkishore , l. s. levitov , a. v. chubukov , nature physics * 8 * 158(2012 ) maximilian l. kiesel , christian platt , werner hanke , dmitry a. abanin , and ronny thomale , phys .
rev .
b * 86 * , 020507(r ) ( 2012 ) wan - sheng wang , yuan - yuan xiang , qiang - hua wang , fa wang , fan yang , and dung - hai lee , phys . rev .
b * 85 * , 035414 ( 2012 ) .
m. mochizuki , y. yanase , m. ogata , phys .
rev .
lett * 94 * , 147005(2005 ) a. m. black - schaffer , c. honerkamp , j. phys . : condens .
matter * 26 * , 423201 ( 2014 ) m. sigrist , m. ueda , rev .
mod .
phys * 63 * , 239(1991 ) j. yuan , c. honerkamp , unpublished r. hlubina , s. sorella , and f. guinea , phys .
rev .
lett . * 78 * , 1343 ( 1997 ) .
antoine georges , luca de medici , jernej mravlje , ann .
rev .
cond . mat .
phys .
* 4 * , 137 ( 2013 ) .
andor kormanyos , viktor zolyomi , neil d. drummond , and guido burkard , phys .
rev .
x * 4 * , 011034 ( 2014 ) . c. husemann and m. salmhofer , phys .
rev .
b * 79 * 195125 ( 2009 ) .
k .-
u . giering and m. salmhofer , phys .
rev .
b * 86 * 245122 ( 2012 ) . |
facility location problems have been widely studied in the operations research and computer science communities ( see , e.g. , @xcite and the survey @xcite ) , and have a wide range of applications . in its simplest version , _ uncapacitated facility location _ ( ) , we are given a set of facilities or service - providers with opening costs , and a set of clients that require service , and we want to open some facilities and assign clients to open facilities so as to minimize the sum of the facility - opening and client - assignment costs .
an oft - cited prototypical example is that of a company wanting to decide where to locate its warehouses / distribution centers so as to serve its customers in a cost - effective manner .
we consider facility - location problems that abstract settings where facilities are mobile and may be relocated to destinations near the clients in order to serve them more efficiently by reducing the client - assignment costs .
more precisely , we consider the _ mobile facility location _
( ) problem introduced by @xcite , which generalizes the classical @xmath4-median problem ( see below ) .
we are given a complete graph @xmath8 with costs @xmath9 on the edges , a set @xmath10 of clients with each client @xmath11 having @xmath12 units of demand , and a set @xmath13 of @xmath4 initial facility locations .
we use the term facility @xmath6 to denote the facility whose initial location is @xmath14 .
a solution @xmath15 to moves each facility @xmath6 to a final location @xmath16 ( which could be the same as @xmath6 ) , incurring a _ movement cost _ @xmath17 , and assigns each client @xmath11 to a final location @xmath18 , incurring _ assignment cost _ @xmath19 .
the total cost of @xmath15 is the sum of all the movement costs and assignment costs .
more formally , noting that each client will be assigned to the location nearest to it in @xmath15 , we can express the cost of @xmath15 as @xmath20 where @xmath21 ( for any node @xmath22 ) gives the location in @xmath15 nearest to @xmath22 ( breaking ties arbitrarily ) .
we assume throughout that the edge costs form a metric .
we use the terms nodes and locations interchangeably .
mobile facility location falls into the genre of _ movement problems _ introduced by demaine et al .
@xcite . in these problems , we are given an initial configuration in a weighted graph specified by placing `` pebbles '' on the nodes and/or edges ; the goal is to move the pebbles so as to obtain a desired final configuration while minimizing the maximum , or total , pebble movement .
was introduced by demaine et al . as the movement problem where facility- and client- pebbles are placed respectively at the initial locations of the facilities and clients , and in the final configuration
every client - pebble should be co - located [ [ our - results . ] ] our results .
+ + + + + + + + + + + + we give the first _ local - search based _ approximation algorithm for this problem and achieve the best - known approximation guarantee .
our main result is a @xmath2-approximation for this problem for any constant @xmath3 using a simple local - search algorithm .
this improves upon the previous best 8-approximation guarantee for due to friggstad and salavatipour @xcite , which is based on lp - rounding and is not combinatorial .
the local - search algorithm we consider is quite natural and simple .
observe that given the final locations of the facilities , we can find the minimum - cost way of moving facilities from their initial locations to the final locations by solving a minimum - cost perfect - matching problem ( and the client assignments are determined by the function @xmath23 defined above ) .
thus , we concentrate on determining a good set of final locations . in our local - search algorithm , at each step , we are allowed to swap in and swap out a fixed number ( say @xmath24 ) of locations .
clearly , for any fixed @xmath24 , we can find the best local move efficiently ( since the cost of a set of final locations can be computed in polytime ) .
note that we do not impose any constraints on how the matching between the initial and final locations may change due to a local move , and a local move might entail moving all facilities .
it is important to allow this flexibility , as it is known @xcite that the local - search procedure that moves , at each step , a constant number of facilities to chosen destinations has an unbounded approximation ratio .
our main contribution is a tight analysis of this local - search algorithm ( section [ 3apx ] ) . our guarantee _ matches _ ( up to @xmath5 terms ) the best - known approximation guarantee for the @xmath4-median problem .
since there is an approximation - preserving reduction from the @xmath4-median problem to @xcite choose arbitrary initial facility locations and give each client a huge demand @xmath25any improvement of our result would imply an analogous improvement for the @xmath4-median problem .
( in this respect , our result is a noteworthy exception to the prevalent state of affairs for various other generalizations of and @xmath4-median e.g .
, the data placement problem @xcite , \{matroid- , red - blue- } median @xcite , @xmath4-facility - location @xcite where the best approximation ratio for the problem is worse by a noticeable factor ( compared to or @xmath4-median ) ; @xcite is another exception . )
furthermore , _ our analysis is tight _ ( up to @xmath5 factors ) because by suitably setting @xmath25 in the reduction of @xcite , we can ensure that our local - search algorithm for coincides with the local - search algorithm for @xmath4-median in @xcite which has a tight approximation ratio of 3 . we also consider a weighted generalization of the problem ( section [ extn ] ) , wherein each facility @xmath6 has a weight @xmath7 indicating the cost incurred per - unit distance moved and the cost for moving @xmath6 to @xmath26 is @xmath27 .
( this can be used to model , for example , the setting where different facilities move at different speeds . )
our analysis is versatile and extends to this weighted generalization to yield the same performance guarantee . for the further generalization of the problem , where the facility - movement costs may be arbitrary and unrelated to the client - assignment costs ( for which a 9-approximation can be obtained via lp - rounding ; see `` related work '' ) , we show that local search based on multiple swaps has a bad approximation ratio ( section [ locgap ] ) .
the analysis leading to the approximation ratio of 3 ( as also the simpler analysis in section [ 5apx ] yielding a 5-approximation ) crucially exploits the fact that we may swap multiple locations in a local - search move .
it is natural to wonder then if one can prove any performance guarantees for the local - search algorithm where we may only swap in and swap out a single location in a local move .
( naturally , the single - swap algorithm is easier to implement and thus may be more practical ) . in section [ oneswap ]
, we analyze this single - swap algorithm and prove that it also has a constant approximation ratio .
[ [ our - techniques . ] ] our techniques .
+ + + + + + + + + + + + + + + the analysis of our local - search procedure requires various novel ideas . as is common in the analysis of local - search algorithms
, we identify a set of test swaps and use local optimality to generate suitable inequalities from these test swaps , which when combined yield the stated performance guarantee .
one of the difficulties involved in adapting standard local - search ideas to is the following artifact : in , the cost of `` opening '' a set @xmath15 of locations is the cost of the min - cost perfect matching of @xmath28 to @xmath15 , which , unlike other facility - location problems , is a highly non - additive function of @xmath15 ( and as mentioned above , we need to allow for the matching from @xmath28 to @xmath15 to change in non - local ways ) . in most facility - location problems with opening costs
for which local search is known to work , we may always swap in a facility used by the global optimum ( by possibly swapping out another facility ) and easily bound the resulting change in _ facility cost _ , and the main consideration is to decide how to reassign clients following the swap in a cost - effective way ; in we do not have this flexibility and need to carefully choose how to swap facilities so as to ensure that there is a good matching of the facilities to their new destinations after a swap _ and _ there is a frugal reassignment of clients .
this leads us to consider long relocation paths to re - match facilities to their new destinations after a swap , which are of the form @xmath29 , where @xmath26 and @xmath30 are the locations that facility @xmath6 is moved to in the local and global optimum , @xmath15 and @xmath31 , respectively , and @xmath32 is the @xmath15-location closest to @xmath30 . by considering a swap move involving the start and end locations of such a path @xmath33 ,
we can obtain a bound on the movement cost of all facilities @xmath34 where @xmath26 is the start of the path or @xmath30 serves a large number of clients . to account for the remaining facilities , we break up @xmath33 into suitable intervals , each containing a constant number of unaccounted locations which then participate in a multi - location swap .
this _ interval - swap _ move does not at first appear to be useful since we can only bound the cost - change due to this move in terms of a significant multiple of ( a portion of ) the cost of the local optimum !
one of the novelties of our analysis is to show how we can _ amortize _ the cost of such expensive terms and make their contribution negligible by considering multiple different ways of covering @xmath33 with intervals and averaging the inequalities obtained for these interval swaps .
these ideas lead to the proof of an approximation ratio of 5 for the local - search algorithm ( section [ 5apx ] ) . the tighter analysis leading to the 3-approximation guarantee ( section [ 3apx ] ) features another noteworthy idea , namely that of using `` recursion '' ( up to bounded depth ) to identify a suitable collection of test swaps .
we consider the tree - like structure created by the paths used in the 5-approximation analysis , and ( loosely speaking ) view this as a recursion tree , using which we spawn off interval - swap moves by exploring this tree to a constant depth . to our knowledge
, we do not know of any analysis of a local - search algorithm that employs the idea of recursion to generate the set of test local moves ( used to generate the inequalities that yield the desired performance guarantee ) .
we believe that this technique is a notable contribution to the analysis of local - search algorithms that is of independent interest and will find further application . [ [ related - work . ] ] related work .
+ + + + + + + + + + + + + as mentioned earlier , was introduced by demaine et al .
@xcite in the context of movement problems .
friggstad and salavatipour @xcite designed the first approximation algorithm for .
they gave an 8-approximation algorithm based on lp rounding by building upon the lp - rounding algorithm of charikar et al .
@xcite for the @xmath4-median problem ; this algorithm works only however for the unweighted case .
they also observed that there is an approximation - preserving reduction from @xmath4-median to .
we recently learned that halper @xcite proposed the same local - search algorithm that we analyze .
his work focuses on experimental results and leaves open the question of obtaining theoretical guarantees about the performance of local search .
chakrabarty and swamy @xcite observed that , even with arbitrary movement costs is a special case of the matroid median problem @xcite .
thus , the approximation algorithms devised for matroid median independently by @xcite and @xcite yield an 8-approximation algorithm for with arbitrary movement costs .
there is a wealth of literature on approximation algorithms for ( metric ) uncapacitated and capacitated facility location ( and ) , the @xmath4-median problem , and their variants ; see @xcite for a survey on . whereas constant - factor approximation algorithms for and @xmath4-median can be obtained via a variety of techniques such as lp - rounding @xcite , primal - dual methods @xcite , local search @xcite , all known @xmath35-approximation algorithms for ( in its full generality ) are based on local search @xcite .
we now briefly survey the work on local - search algorithms for facility - location problems . starting with the work of @xcite ,
local - search techniques have been utilized to devise @xmath35-approximation algorithms for various facility - location problems .
korupolu , plaxton , and rajaraman @xcite devised @xmath35-approximation for , and with uniform capacities , and @xmath4-median ( with a blow - up in @xmath4 ) .
charikar and guha @xcite , and arya et al .
@xcite both obtained a @xmath36-approximation for .
the first constant - factor approximation for was obtained by pl , tardos , and wexler @xcite , and after some improvements , the current - best approximation ratio now stands at @xmath37 @xcite . for the special case of uniform capacities ,
the analysis in @xcite was refined by @xcite , and aggarwal et al .
@xcite obtain the current - best 3-approximation .
arya et al .
@xcite devised a @xmath38-approximation algorithm for @xmath4-median , which was also the first constant - factor approximation algorithm for this problem based on local search .
gupta and tangwongsan @xcite ( among other results ) simplified the analysis in @xcite .
we build upon some of their ideas in our analysis .
local - search algorithms with constant approximation ratios have also been devised for various variants of the above three canonical problems .
mahdian and pl @xcite , and svitkina and tardos @xcite consider settings where the opening cost of a facility is a function of the set of clients served by it . in @xcite
, this cost is a non - decreasing function of the number of clients , and in @xcite this cost arises from a certain tree defined on the client set .
devanur et al .
@xcite and @xcite consider @xmath4-facility location , which is similar to @xmath4-median except that facilities also have opening costs .
hajiaghayi et al .
@xcite consider a special case of the matroid median problem that they call the red - blue median problem .
most recently , @xcite considered a problem that they call the @xmath4-median forest problem , which generalizes @xmath4-median , and obtained a @xmath38-approximation algorithm .
as mentioned earlier , to compute a solution to , we only need to determine the set of final locations of the facilities , since we can then efficiently compute the best movement of facilities from their initial to final locations , and the client assignments .
this motivates the following local - search operation .
given a current set @xmath15 of @xmath39 locations , we can move to any other set @xmath40 of @xmath4 locations such that @xmath41 , where @xmath24 is some fixed value .
we denote this move by @xmath42 .
the local - search algorithm starts with an arbitrary set of @xmath4 final locations . at each iteration
, we choose the local - search move that yields the largest reduction in total cost and update our final - location set accordingly ; if no cost - improving move exists , then we terminate .
( to obtain polynomial running time , as is standard , we modify the above procedure so that we choose a local - search move only if the cost - reduction is at least @xmath43 . )
[ sec:5appx ] we now analyze the above local - search algorithm and show that it is a @xmath44-approximation algorithm . for notational simplicity , we assume that the local - search algorithm terminates at a local optimum ; the modification to ensure polynomial running time degrades the approximation by at most a @xmath45-factor ( see also remark [ polyremk ] ) .
[ 5apxthm ] let @xmath46 and @xmath47 denote respectively the movement and assignment cost of an optimal solution .
the total cost of any local optimum using at most @xmath24 swaps is at most @xmath48 .
although this is not the tightest guarantee that we obtain , we present this analysis first since it introduces many of the ideas that we build upon in section [ 3apx ] to prove a tight approximation guarantee of @xmath49 for the local - search algorithm . for notational simplicity , we assume that all @xmath12s are 1 .
all our analyses carry over trivially to the case of non - unit ( integer ) demands since we can think of a client @xmath11 having @xmath12 demand as @xmath12 co - located unit - demand clients . [ [ notation - and - preliminaries . ] ] notation and preliminaries .
+ + + + + + + + + + + + + + + + + + + + + + + + + + + we use @xmath50 to denote the local optimum , where facility @xmath6 is moved to final location @xmath51 .
we use @xmath52 to denote the ( globally ) optimal solution , where again facility @xmath6 is moved to @xmath30 . throughout
, we use @xmath53 to index locations in @xmath15 , and @xmath54 to index locations in @xmath31 . recall that , for a node @xmath22
, @xmath55 is the location in @xmath15 nearest to @xmath22 .
similarly , we define @xmath56 to be the location in @xmath31 nearest to @xmath22 . for notational similarity with facility location problems , we denote @xmath17 by @xmath57 , and @xmath58 by @xmath59 .
( thus , @xmath57 and @xmath59 are the movement costs of @xmath6 in @xmath15 and @xmath31 respectively . ) also , we abbreviate @xmath60 to @xmath61 , and @xmath62 to @xmath63 .
thus , @xmath61 and @xmath63 are the assignment costs of @xmath11 in the local and global optimum respectively .
( so @xmath64 . )
let @xmath65 be the set of clients assigned to the location @xmath18 , and @xmath66 .
for a set @xmath67 , we define @xmath68 ; we define @xmath69 for @xmath70 similarly .
define @xmath71 .
we say that @xmath53 _ captures _ all the locations in @xmath72 .
the following basic lemma will be used repeatedly .
[ reasgn ] for any client @xmath11 , we have @xmath73 .
let @xmath74 .
the lemma clearly holds if @xmath75 .
otherwise , @xmath76 where the second inequality follows since @xmath77 is the closest location to @xmath54 in @xmath15 . to prove the approximation ratio
, we will specify a set of local - search moves for the local optimum , and use the fact that none of these moves improve the cost to obtain some inequalities , which will together yield a bound on the cost of the local optimum .
we describe these moves by using the following digraph .
consider the digraph @xmath78 .
we decompose @xmath79 into a collection of node - disjoint ( simple ) paths @xmath80 and cycles @xmath81 as follows .
repeatedly , while there is a cycle @xmath82 in our current digraph , we add @xmath82 to @xmath81 , remove all the nodes of @xmath82 and recurse on the remaining digraph . after this step , a node @xmath22 in the remaining digraph , which is acyclic , has : exactly one outgoing arc if @xmath83 ; exactly one incoming and one outgoing arc if @xmath84 ; and exactly one incoming , and at most one outgoing arc if @xmath85 .
now we repeatedly choose a node @xmath83 with no incoming arcs , include the maximal path @xmath86 starting at @xmath22 in @xmath80 , remove all nodes of @xmath86 and recurse on the remaining digraph .
thus , each triple @xmath87 is on a unique path or cycle in @xmath88 .
define @xmath89 to be @xmath90 such that @xmath91 is an arc in @xmath88 ; if @xmath53 has no incoming arc in @xmath88 , then let @xmath92 . we will use @xmath80 and @xmath81 to define our swaps . for a path @xmath93 ,
define @xmath94 to be @xmath95 and @xmath96 to be @xmath97 .
notice that @xmath98 . for
each @xmath18 , let @xmath99 , @xmath100 , and @xmath101 .
note that latexmath:[$| { \ensuremath{\mathcal p}}_c(s ) | = @xmath18 with @xmath103 . for a set @xmath67 ,
define @xmath104 .
a basic building block in our analysis , involves a _ shift _ along an @xmath105 sub - path @xmath33 of some path or cycle in @xmath88 .
this means that we swap out @xmath53 and swap in @xmath54 .
we bound the cost of the matching between @xmath28 and @xmath106 by moving each initial location @xmath107 to @xmath108 and moving @xmath109 to @xmath110 .
thus , we obtain the following simple bound on the increase in movement cost due to this operation : @xmath111 the last inequality uses the fact that @xmath112 for all @xmath6 . for a path @xmath113
, we use @xmath114 as a shorthand for @xmath115 .
we now describe the local moves used in the analysis .
we define a set of swaps such that each @xmath90 is swapped in to an extent of at least one , and at most two .
we classify each location in @xmath15 as one of three types .
define @xmath116 .
we assume that @xmath117 .
@xmath118=0.5ex = 0ex @xmath119 : locations @xmath120 with @xmath121 .
@xmath122 : locations @xmath123 with @xmath124 or @xmath125 .
@xmath126 : locations @xmath120 with @xmath127 and @xmath128 . also define @xmath129 ( so @xmath130 iff @xmath131 and @xmath132 ) . to gain some intuition , notice that it is easy to generate a suitable inequality for a location @xmath133
: we can `` delete '' @xmath53 ( i.e. , if @xmath134 , then do @xmath135 ) and reassign each @xmath136 to @xmath137 ( i.e. , the location in @xmath15 closest to the location serving @xmath11 in @xmath31 ) .
the cost increase due to this reassignment is at most @xmath138 , and so this yields the inequality @xmath139 .
( we do not actually do this since we take care of the @xmath119-locations along with the @xmath122-locations . )
we can also generate a suitable inequality for a location @xmath140 ( see lemma [ s2lem ] ) since we can swap in @xmath72 and swap out @xmath141 .
the cost increase by this move can be bounded by @xmath142 and @xmath143 , and the latter quantity can be charged to @xmath144 ; our definition of @xmath126 is tailored precisely so as to enable this latter charging argument . generating inequalities for the @xmath122-locations is more involved , and requires another building block that we call an interval swap ( this will also take care of the @xmath119-locations ) , which we define after proving lemma [ s2lem ] .
we start out by proving a simple bound that one can obtain using a cycle in @xmath81 .
[ cycleswap ] for any cycle @xmath145 , we have @xmath146 .
consider the following matching of @xmath147 to @xmath148 : we match @xmath6 to @xmath149 .
the cost of the resulting new matching is @xmath150 which should at least @xmath151 since the latter is the min - cost way of matching @xmath28 to @xmath15 .
so we obtain @xmath152 .
[ s2lem ] let @xmath140 and @xmath153 , and consider @xmath154 .
we have @xmath155 we can view this multi - location swap as doing @xmath156 for each @xmath157 and @xmath158 simultaneously .
( notice that no path @xmath157 contains @xmath53 , since @xmath159 . ) for each @xmath156 the movement - cost increase is bounded by @xmath160 .
for @xmath158 we move the facility @xmath6 , where @xmath134 , to @xmath54 , so the increase in movement cost is at most @xmath161 for every @xmath162 .
so since @xmath163 , we have @xmath164 .
thus , the increase in total movement cost is at most we upper bound the change in assignment cost by reassigning the clients in @xmath165 as follows .
we reassign each @xmath166 to @xmath54 .
each @xmath167 is assigned to @xmath168 , if @xmath169 , and otherwise to @xmath170 .
note that @xmath171 : @xmath172 since @xmath173 , and @xmath174 since @xmath175 .
the change in assignment cost for each such client @xmath11 is at most @xmath176 by lemma [ reasgn ] .
thus the change in total assignment cost is at most @xmath177 .
combining this with the bound on the movement - cost change proves the lemma .
we now define a key ingredient of our analysis , called an _ interval - swap _ operation , that is used to bound the movement cost of the @xmath122- and @xmath119-locations and the assignment cost of the clients they serve .
( we build upon this in section [ 3apx ] to give a tighter analysis proving a 3-approximation . )
let @xmath178 be a subset of at most @xmath179 locations on a path or cycle @xmath33 in @xmath180 , where @xmath181 is the next location in @xmath182 after @xmath183 .
let @xmath184 where @xmath185 for @xmath186 and @xmath187 is an arbitrary location that appears after @xmath188 ( and before @xmath189 ) on the corresponding path or cycle .
consider each @xmath183 .
if @xmath190 , choose a _ random _ path @xmath191 with probability @xmath192 , and set @xmath193 and @xmath194 .
if @xmath195 , set @xmath196 , and @xmath197 . set @xmath198 and @xmath199 .
note that @xmath200 since @xmath201 for every @xmath202 .
notice that @xmath203 is a random set , but @xmath204 is deterministic . to avoid cumbersome notation
, we use @xmath205 to refer to the distribution of swap - moves that results by the random choices above , and call this the _ interval swap corresponding to @xmath40 and @xmath206_. we bound the expected change in cost due to this move below . let @xmath207 be the indicator function that is 1 if @xmath130 and 0 otherwise .
[ s1lem ] let @xmath208 and @xmath206 be as given above .
let @xmath209 , where @xmath210 and @xmath211 if @xmath212 .
consider the interval swap @xmath213 corresponding to @xmath40 and @xmath206 , as defined above .
we have @xmath214}}\ \leq\ \sum_{q=1}^{r } { \ensuremath{\mathsf{shift}}}(s'_q , o'_q)+\sum_{{p\in{\ensuremath{\mathcal p}}_c(s ' ) , i\in p}}2f^*_i + \sum_{j\in d^*(o ' ) } ( c^*_j - c_j ) \hspace*{-4ex } \\
& + \sum_{j\in d\left(t(s'\cap s_3)\cup ( s'\cap s_3)\right ) } 2c^*_j + \sum_{j\in d\left(t(s'{\ensuremath{\setminus}}s_3)\right ) } \tfrac{2c^*_j}{t } + { \ensuremath{\mathbf{1}}}(s'_1)\sum_{j\in d^*(o'_0 ) } ( f^*_{{\ensuremath{\hat i}}}+f_{{\ensuremath{\hat i}}}+c^*_j ) .
\end{split } \label{t3}\ ] ] let @xmath33 be the path in @xmath80 or cycle in @xmath81 such that @xmath215 .
we first bound the increase in movement cost .
the interval swap can be viewed as a collection of simultaneous @xmath216 moves .
if @xmath193 for a random path @xmath217 , the movement - cost increase can be broken into two parts .
we do a shift along @xmath86 , but move the last initial location on @xmath86 to @xmath183 , and then do shift on @xmath33 from @xmath183 to @xmath218 .
so the expected movement - cost change is at most @xmath219 which is at most @xmath220 .
similarly , if @xmath195 , we can break the movement - cost increase into @xmath160 for all @xmath191 and @xmath221 .
thus , the total increase in movement cost is at most @xmath222 next , we bound the change in assignment cost by reassigning clients in @xmath223 as follows .
we assign each client @xmath224 to @xmath168 . if @xmath225 , then @xmath226 . for every client @xmath227 , observe that either @xmath169 or @xmath228 . to see this ,
let @xmath229 and @xmath230 .
if @xmath231 then @xmath232 ; also @xmath233 , and so @xmath234 . so we assign @xmath11 to @xmath168 if @xmath169 and to @xmath137 otherwise ; the change in assignment cost of @xmath11 is at most @xmath176 ( lemma [ reasgn ] ) .
now suppose @xmath235 , so @xmath236 .
for each @xmath237 ,
we again have @xmath169 or @xmath228 , and we assign @xmath11 to @xmath168 if @xmath169 and to @xmath137 otherwise . we assign every @xmath238 to @xmath239 ( recall that @xmath240 ) , and overestimate the resulting change in assignment cost by @xmath241 . finally , note that we reassign a client @xmath242 with probability at most @xmath243 ( since @xmath244 with probability at most @xmath243 ) .
so taking into account all cases , we can bound the change in total assignment cost by @xmath245 in , we are double - counting clients in @xmath246 .
we are also overestimating the change in assignment cost of a client @xmath247 since we include both the @xmath248 term , and the @xmath176 or @xmath249 terms . adding and yields the lemma . notice that lemma [ s2lem ] immediately translates to a bound on the assignment cost of the clients in @xmath250 for @xmath140 .
in contrast , it is quite unclear how lemma [ s1lem ] may be useful , since the expression @xmath251 in the rhs of may be as large as @xmath252 ( but no more since @xmath253 if @xmath254 ) and it is unclear how to cancel the contribution of @xmath255 on the rhs .
one of the novelties of our analysis is that we show how to _ amortize _ such expensive terms and make their contribution negligible by considering multiple interval swaps .
we cover each path or cycle @xmath33 in @xmath179 different ways using intervals comprising consecutive locations from @xmath256 .
we then argue that averaging , over these @xmath179 covering ways , the inequalities obtained from the corresponding interval swaps yields ( among other things ) a good bound on the movement - cost of the @xmath257-locations on @xmath33 and the assignment cost of the clients they serve .
[ corbigi ] let @xmath258 , @xmath259 , where @xmath181 is the next @xmath122-location on @xmath33 after @xmath183 , and @xmath260 .
let @xmath261 if @xmath262 and @xmath263 otherwise . for @xmath264 , @xmath265
we first define formally an interval of ( at most ) @xmath179 consecutive @xmath257 locations along @xmath33 . as before , let @xmath266 for @xmath202 . for a path @xmath33 ,
define @xmath267 for @xmath268 and @xmath269 for @xmath270 . also define @xmath271 for @xmath268 and @xmath272 for @xmath273 .
if @xmath33 is a cycle , we let our indices wrap around and be @xmath274 , i.e. , @xmath275 for all @xmath276 ( so @xmath277 ) . for @xmath278 ,
define @xmath279 to be an interval of length at most @xmath179 on @xmath33 .
define @xmath280 .
note that we have @xmath281 if @xmath33 is a path , and @xmath282 if @xmath33 is a cycle .
consider the collection of intervals , @xmath283 .
for each @xmath284 , where @xmath285 , we consider the interval swap @xmath286 corresponding to @xmath284 .
we add the inequalities @xmath287 for all such @xmath288 .
since each @xmath289 participates in exactly @xmath179 such inequalities , and each @xmath290 is the start of only the interval @xmath291 , we obtain the following .
@xmath292 notice that the @xmath15-locations other than @xmath183 on the @xmath293 sub - paths of @xmath33 lie in @xmath126 , and for each @xmath6 such that @xmath294 , we have @xmath295 .
thus , using , we have @xmath296 since @xmath297 means that @xmath298 , and so @xmath299 , we have @xmath300 incorporating and in , and simplifying yields the desired inequality . for a path or cycle @xmath33 where @xmath301
, we obtain an inequality similar to .
since we can now cover @xmath33 with a single interval , we never have a client @xmath11 such that none of @xmath302 are in our new set of final locations .
so the resulting inequality does not have any @xmath303 terms .
[ corsmalli ] let @xmath258 , @xmath259 , where @xmath181 is the next @xmath122-location on @xmath33 after @xmath183 , and @xmath260 .
let @xmath261 if @xmath262 and @xmath263 otherwise . for @xmath304 , @xmath305
the proof is similar to that of lemma [ corbigi ] , except that since we can cover @xmath33 with a single interval , we only need to consider a single ( multi - location ) swap .
we consider two cases for clarity . .
= 0.5ex = 0.5ex * @xmath33 is a path , or @xmath306 . * as before ,
let @xmath266 for @xmath202 .
if @xmath33 is a path , define @xmath307 .
if @xmath33 is a cycle , we again set @xmath275 for all @xmath276 .
consider the interval swap @xmath308 corresponding to @xmath309 .
the inequality generated by this is similar to except that we do not have any @xmath310 terms since for client @xmath311 , we always have that either @xmath169 or @xmath228 .
thus , translates to the following .
@xmath312 substituting @xmath313 as in yields the stated inequality .
* @xmath33 is a cycle with @xmath314 . * here , lemma [ cycleswap ] yields @xmath315 ( which is the special case of the earlier inequality with @xmath316 ) .
theorem [ 5apxthm ] we consider the following set of swaps . a. = 0ex = 0.5ex for every @xmath140 , the move @xmath317 . for every path or cycle @xmath33 with @xmath318 , the @xmath319-weighted interval swaps as defined in lemma [ corbigi ] .
for every path or cycle @xmath33 with @xmath320 , the interval swap defined in lemma [ corsmalli ] .
notice that every location @xmath90 is swapped in to an extent of at least 1 and at most 2 .
( by `` extent '' we mean the total weight of the inequalities involving @xmath54 . ) to see this , suppose first @xmath321 for some path @xmath33 , then @xmath54 is involved to an extent of 1 in the interval swaps for @xmath33 in a2 or a3 . in this case
, we say that the interval swap for @xmath33 is _ responsible _ for @xmath54 . additionally , if @xmath322 , then @xmath54 belongs to the multi - swap for @xmath53 in a1 , else if @xmath130 then @xmath54 is part of the interval swap for the path / cycle containing @xmath53 in a2 or a3 .
now suppose @xmath153 .
if @xmath140 , then @xmath54 is included in the multi - swap for @xmath53 in a1 .
we say that this multi - swap is responsible for @xmath54 .
if @xmath323 , then @xmath54 is included in the interval swap for the path / cycle containing @xmath53 in a2 or a3 , and we say that this interval swap is responsible for @xmath54 .
consider the compound inequality obtained by summing , , and corresponding to the moves considered in a1 , a2 , and a3 respectively .
the lhs of this inequality is 0 .
we now need to do some bookkeeping to bound the coefficients of the @xmath324 terms on the rhs .
we ignore @xmath5 coefficients like @xmath325 in this bookkeeping , since for a given @xmath326 term , such coefficients appear in only a constant number of inequalities , so they have an @xmath5 effect overall .
let @xmath327 and @xmath82 denote respectively the movement- and assignment- cost of the local optimum .
* contribution from the @xmath63 and @xmath61 terms .
* first , observe that for each @xmath90 , we have labeled exactly one move involving @xmath54 as being responsible for it .
consider a client @xmath328 .
observe that @xmath63 or @xmath61 terms appear ( with a @xmath329-coefficient ) in an inequality generated by a move if ( i ) @xmath11 is reassigned because the move is responsible for @xmath54 ; or ( ii ) @xmath53 is swapped out ( to an extent of 1 ) by the move ( so this excludes the case where @xmath330 and the move is the interval swap for the path containing @xmath77 ) .
if ( i ) applies , then the inequality generates the term @xmath331 .
if ( ii ) applies then the term @xmath176 appears in the inequality .
finally , note that there are at most two inequalities for which ( ii ) applies :
= 0ex = 0ex if @xmath332 , then ( ii ) applies for the interval - swap move for @xmath33 .
if @xmath333 , then ( ii ) again applies , for the multi - swap move for @xmath77 if @xmath334 , or for the interval swap for the path containing @xmath77 if @xmath335 .
if @xmath336 , then ( ii ) applies for the interval swap for the path containing @xmath53 . if @xmath140 , then ( ii ) applies for the multi - swap move for @xmath53 .
so overall , we get a @xmath337 contribution to the rhs , the bottleneck being the two inequalities for which ( ii ) applies when @xmath338 and @xmath339 .
* contribution from the @xmath59 and @xmath57 terms . * for every @xmath14 , the expression @xmath340 is counted once in the rhs of the inequality or for the unique path or cycle @xmath33 containing @xmath6 .
the total contribution of all these terms is therefore , @xmath341 .
the remaining contribution comes from expressions of the form @xmath342 on the rhs of , , and .
the paths @xmath86 involved in these expressions come from @xmath343 .
therefore , the total contribution of these terms is at most @xmath344 .
thus , we obtain the compound inequality @xmath345 where the @xmath5 terms are @xmath346 .
this shows that @xmath347 .
[ polyremk ] a subtle point to note is that in the above analysis : ( 1 ) we consider only a polynomial number of swap moves ( since there are at most @xmath348 swap moves available at any point ) , and ( 2 ) we place a constant weight ( of at most 1 ) on the inequality obtained from any given swap move ( when we take the weighted sum of the inequalities obtained from the various swap moves ) .
this is relevant because we can only ensure in polynomial time that we terminate at an approximate local optimum .
more precisely , for any polynomial @xmath349 and an @xmath350 , we obtain in polynomial time a solution with cost @xmath351 such that the change in cost due to any local move is at least @xmath352 ( instead of 0 ) . but
this means that the lhs of is now @xmath353 , where @xmath354 is the total weight placed on the inequalities generated from the various swap moves whose suitable linear combination yields .
therefore , since @xmath355 , this only results in a @xmath45-loss in approximation factor .
in this section , we improve the analysis from section [ sec:5appx ] . specifically , we prove the following theorem .
[ thm:3appx ] the cost of a locally - optimal solution using @xmath24 swaps is at most @xmath356 times the optimum solution cost . to gain some intuition behind this tighter analysis , note
that the _ only _ reason we lost a factor of 5 in the previous analysis was because there could be locations @xmath332 that are swapped out to an extent of 2 ; hence , there could be clients @xmath136 for which we `` pay '' @xmath176 each time @xmath53 is swapped out , and also pay an additional @xmath357 term when @xmath168 is swapped in . to improve the analysis , we will consider a set of test swaps that swap out each location in @xmath15 to an extent of @xmath358 .
the aforementioned bad case happens only when @xmath333 , because when we close ( i.e. , swap out ) @xmath77 as part of an interval swap or a multi - swap , we open ( i.e. , swap in ) all the locations in @xmath359 , and we achieve this via path swaps ( i.e. , @xmath360 moves ) along paths in @xmath361 that swap out locations in @xmath362 ( for a second time ) . the main idea behind our refined analysis is to not perform such path swaps , but instead to `` recursively '' start an interval swap on each path in @xmath361 . of course , we can not carry out this recursion to arbitrary depth ( since we can only swap a bounded number of locations ) , so we terminate the recursion at a depth of @xmath179 .
so , whereas an interval swap included at most @xmath179 @xmath122-locations on the main path or cycle @xmath33 , we now consider a `` subtree '' swap obtained by aggregating interval swaps on the paths in @xmath363 .
a subtree swap can be viewed as a bounded - depth recursion tree where each leaf to root path encounters at most @xmath179 locations in @xmath122 . because we no longer initiate path swaps for @xmath364-locations , a leaf location @xmath365 in this recursion tree will not have any locations in @xmath366 opened .
but we will slightly redefine the @xmath367 sets to ensure that @xmath368 , and use the same trick that we did with interval swaps in section [ 5apx ] : we _ average _ over different sets of subtree swaps ( like we did with interval swaps in section [ 5apx ] ) to ensure that @xmath369 is a leaf location with probability at most @xmath319 .
this ensures that we incur , to an extent of at most @xmath243 , the cost @xmath370 , where @xmath371 , for moving @xmath11 with @xmath372 from @xmath369 to @xmath26 .
[ [ notation . ] ] notation .
+ + + + + + + + + let @xmath373 be an integer such that @xmath374 .
we prove that the local - search algorithm has approximation ratio @xmath375 .
we redefine @xmath376 and @xmath364 as follows .
@xmath118=0.5ex = 0ex @xmath377 .
@xmath378 .
@xmath379 .
@xmath380 .
clearly , @xmath381 .
we also redefine @xmath89 to be the location in @xmath72 _ closest to @xmath53_. [ s2dist ] let @xmath140 and @xmath153 . then @xmath382 .
we have @xmath383 for any @xmath384 , and @xmath385 for any @xmath386 .
therefore , @xmath387 for any @xmath388 , and the claim follows since @xmath389 as @xmath140 .
[ fig : hgraph ] it will be more convenient to work with the digraph @xmath390 obtained from @xmath79 by contracting each triple @xmath391 of nodes associated with a facility @xmath6 into a single node that we also denote by @xmath6 . thus , @xmath392 is an arc in @xmath393 if @xmath394 ( it may be that @xmath395 ) . note that @xmath396 may have self loops , and each node in @xmath396 has outdegree exactly 1 ( counting self - loops ) so each component of @xmath396 looks like a tree with all edges oriented toward the root , except the root is in fact a directed cycle ( possibly a self - loop ) .
the figure above illustrates this graph and some of the subgraphs and structures discussed below . for brevity
, we say that an edge @xmath392 in @xmath396 is a _ center _
edge if @xmath397 . in the arc set @xmath398 ,
each node has indegree and outdegree at most 1 , so @xmath399 consists of a collection node - disjoint paths @xmath80 and cycles @xmath81 . for a facility @xmath400 ,
let @xmath401 denote the unique path in @xmath80 containing @xmath6 .
let @xmath402 and @xmath403 denote the start and end facilities of @xmath401 respectively
. for distinct facilities @xmath404 with @xmath405 , the paths @xmath401 and @xmath406 are clearly vertex disjoint .
now define @xmath407 ; that is , @xmath408 is the subgraph of @xmath396 with node - set @xmath28 and edges @xmath392 of @xmath393 where @xmath392 is a center edge or @xmath409 . call a node @xmath6 of @xmath408 a _ root _
if @xmath6 has no outgoing arc or @xmath6 lies on a directed cycle in @xmath408 .
we consider an integer @xmath410 and describe a set of swaps for each index @xmath411 .
the inequalities for the swaps for different @xmath411 will be averaged in the final analysis .
we obtain @xmath412 by deleting the edges @xmath392 of @xmath408 where : @xmath6 is not on a cycle , @xmath413 , and the number of facilities @xmath414 with @xmath415 on the path between @xmath109 and the root of its component in @xmath408 ( both included ) is @xmath416 .
we define a _ subtree _ of @xmath412 to be an acyclic component of @xmath412 , or a component that results by deleting the edges of the cycle contained in a component of @xmath412 . for a facility @xmath6 ,
define @xmath417 .
note that @xmath418 .
the reason we define @xmath419 is that we will sometimes perform a shift along some path @xmath420 to reassign the facilities on @xmath33 but we will not want this to interfere with the operations in the subtree of @xmath412 containing @xmath6 . for a facility @xmath6 with @xmath421 ,
let @xmath422 be the facility obtained as follows .
follow the unique walk from @xmath6 in @xmath396 using only center edges until the walk reaches a node @xmath109 with either no outgoing center edge , or the unique @xmath423 center edge satisfies @xmath415 ; we set @xmath424 .
[ subtreesize ] the number of facilities @xmath6 with @xmath425 in any subtree of @xmath412 is at most @xmath426 .
the facilities @xmath6 in such a subtree that are in @xmath126 have indegree and outdegree at most 1 . shortcutting past these facilities
yields a tree with depth at most @xmath179 and branching factor at most @xmath373 .
[ [ the - test - swaps . ] ] the test swaps .
+ + + + + + + + + + + + + + + for a subtree @xmath427 of @xmath412 , we describe a set of nodes @xmath428 to be swapped out and a set of nodes @xmath429 to be swapped in with @xmath430 .
we do not actually perform these swaps yet to generate the inequalities since we will have to combine some of these swaps for various components .
for each @xmath431 with @xmath432 , we add the following location in @xmath15 to @xmath428 : if @xmath433 we add @xmath26 to @xmath428 ; otherwise ( so @xmath434 ) , we choose any single @xmath435 uniformly at random and add @xmath436 to @xmath428 .
we also add @xmath437 to @xmath429 .
when we say perform @xmath438 , we specifically mean the following reassignment of facilities .
for @xmath439 with @xmath433 , we perform @xmath440 . for @xmath439 with @xmath441 ,
say @xmath109 is the facility in @xmath419 for which @xmath436 was added to @xmath428 .
then we perform @xmath442 , move facility @xmath109 from @xmath110 to @xmath26 , and finally perform @xmath440 . as always ,
each client is then assigned to its nearest final location .
lemma [ lem : disjoint ] implies that these shift operations charge different portions of the local and global optimum .
[ lem : disjoint ] for a subtree @xmath427 , all of the shift operations described for @xmath443 have their associated paths being vertex disjoint . for any subtree @xmath427 ,
the paths between @xmath26 and @xmath437 for the facilities @xmath444 with @xmath425 are vertex - disjoint by definition of @xmath422 . also , for any two distinct @xmath445 , and any @xmath446 , we have @xmath447 , and so their associated paths @xmath401 and @xmath406 are also vertex - disjoint
. finally , consider any @xmath444 with @xmath432 , and @xmath448 ( @xmath414 could be @xmath6 ) with @xmath449 .
consider the paths involved in @xmath450 and @xmath451 , where @xmath452 .
note that both of these paths consist of only center edges .
therefore , since each facility has at most one incoming and one outgoing center edge , and @xmath453 , if these paths are not vertex - disjoint , then it must be that the path involved in @xmath450 is a subpath of the path involved in @xmath454 .
this means that @xmath6 and @xmath109 , and hence , @xmath455 , are all in the same component of @xmath408 . also , the edge @xmath423 is not in @xmath408 so @xmath109 is the root of its component in @xmath408 .
but then there is a path from @xmath414 to @xmath109 , which contradicts that @xmath456 .
we need to coordinate the swaps for the various subtrees of @xmath412 .
consider a component @xmath33 in @xmath408 .
let @xmath457 if @xmath33 is rooted at a node , otherwise let @xmath82 be its cycle of root nodes .
let @xmath458 be the facilities on @xmath82 with @xmath459 , indexed by order of appearance on @xmath82 starting from an arbitrary facility @xmath460 on @xmath82 ( @xmath461 if @xmath457 ) .
we consider four kinds of swaps . * type . *
= 0.5ex = 0ex = 0ex = if @xmath462 , simultaneously do @xmath443 for all subtrees @xmath427 rooted at some @xmath463 with @xmath464 .
otherwise , if @xmath465 , define @xmath466 for all @xmath467 ( where the indices are @xmath468 ) . simultaneously perform @xmath469 for each subtree @xmath427 rooted at a facility in @xmath470 . reasoning similarly as in lemma [ lem : disjoint ] and noting that the subtrees involved in a single type-1 or type-2 swap are all disjoint , we can see that all shift paths involved in a single type-1 or type-2 swap are vertex - disjoint . for each @xmath6 with @xmath471 , simultaneously perform @xmath443 for all subtrees @xmath427 rooted at some @xmath109 with @xmath472 .
at the same time , we also swap out @xmath26 and swap in @xmath473 for a total of at most @xmath474 swaps
. it may be that some ( at most one ) shift path in this swap includes @xmath26 , but then we just move @xmath414 to @xmath475 instead of @xmath26 , and then move @xmath6 according to the shift operation . finally , for every other subtree @xmath427 of @xmath412 that was not swapped in the previous cases , perform @xmath443 on its own .
[ [ analysis . ] ] analysis .
+ + + + + + + + + we first bound the net client - assignment cost increase for any single one of these test swaps .
so , fix one such swap , let @xmath476 be the set of subtrees involved in the swap , and let @xmath477 denote the set of facilities @xmath6 such that @xmath478 and @xmath479 is closed during the swap while @xmath30 is not opened .
so @xmath477 consists of facilities with a center edge to some leaf of some subtree @xmath480 or , if the swap is of type 2 , to the start of an interval @xmath470 . for this swap ,
let @xmath481 , @xmath482 , and @xmath483 .
[ lem : clientbound ] the expected change in client - assignment cost for a test swap is at most @xmath484 . here
, the expectation is over the random choices involved in selecting facilities from the appropriate @xmath485 sets .
after the swap , we move every @xmath486 from @xmath487 to @xmath488 for a cost change of @xmath489 .
every client in @xmath490 for which @xmath487 is closed is moved initially to @xmath491 for a cost increase of at most @xmath176 .
suppose @xmath6 is such that @xmath492 and @xmath493 .
it may be that @xmath479 is still not open which means that @xmath494 .
note that either @xmath26 or @xmath30 is opened after the shift and we move every client that was moved to @xmath479 to @xmath26 or @xmath30 ( whichever is open ) .
this extra distance moved is at most @xmath495 .
note that @xmath496 implies that @xmath497 , otherwise @xmath149 would not have been closed down in the swap .
so @xmath498 , by definition of @xmath364 , and at most @xmath373 clients will be moved to either @xmath26 or @xmath30 in this manner .
finally , we note that while @xmath499 may have @xmath487 being closed , this only happens with probability at most @xmath500 .
now , we consider the following weightings of the swaps .
first , for a particular index @xmath410 we perform all type 1 , 3 , and 4 swaps . for a component of @xmath412 containing a cycle @xmath82
, we perform all type-2 swaps for the various intervals @xmath470 for @xmath82 and weight the client and facility cost change by @xmath319 .
finally , these weighted bounds on the client and facility cost change are averaged over all @xmath410 .
[ lem : client ] the expected change in client - assignment cost under the weighting described above , is at most @xmath501 . for a fixed @xmath411 , every client @xmath11 is in @xmath502 as in lemma [ lem : clientbound ] to an extent of 1 ; either once in a type 1 , 3 , or 4 swap or exactly @xmath179 times among the type-2 swaps , each of which is counted with weight @xmath319 .
similarly , every client @xmath11 is in @xmath503 to an extent of at most 1 and is in @xmath504 to an extent of at most 1 over all swaps for this fixed @xmath411 .
finally , we note each facility @xmath6 on a cycle in @xmath408 lies in the set @xmath477 for at most one offset @xmath505 for that cycle , so its contribution @xmath506 to the bound is only counted with weight @xmath319 for this fixed @xmath411 .
lastly , every facility @xmath6 not on a cycle in @xmath408 lies in @xmath477 for at most one index @xmath507 and , then , in only one swap for that particular @xmath411 .
since we average the bound over all indices @xmath411 between 1 and @xmath179 , the contribution @xmath508 of such @xmath6 is counted with weight only @xmath319 .
next we bound the expected facility movement cost change .
let @xmath509 be the set of facilities @xmath6 that do not lie on a cycle in @xmath408 consisting solely of facilities @xmath109 with @xmath510 .
[ lem : fac ] the expected change in movement cost ( under the weighting described above ) is at most @xmath511 .
we consider two cases for a facility @xmath6 .
first , suppose @xmath425 .
then when @xmath26 is swapped out in a subtree during the shift from @xmath26 to @xmath437 , @xmath6 is moved to either @xmath30 , if @xmath512 , or to @xmath479 , if @xmath513 .
the latter case implies that @xmath514 .
the total movement change is at most @xmath515 if @xmath6 is moved to @xmath30 and is at most @xmath516 if @xmath6 is moved to @xmath479 . since @xmath517 and @xmath493
, by claim [ s2dist ] we have that @xmath518 .
the only other time @xmath6 is moved is when @xmath403 is randomly chosen from @xmath519 for some facility @xmath109 . but
this happens with probability at most @xmath500 . in this case
, @xmath6 is shifted from @xmath26 to @xmath479 .
we do not necessarily have @xmath520 in this case , but we can use the bound @xmath521 to bound the expected movement - cost change for @xmath6 in this case to be at most @xmath522 .
overall , the expected movement - cost change for @xmath6 is at most @xmath523 next , we consider the case @xmath471 .
let @xmath524 .
when the swap consisting of @xmath6 and all subtrees rooted at @xmath525 is performed , @xmath6 is moved from @xmath26 to @xmath110 unless @xmath6 lies on a shift path during that swap , in which case it is moved like in the shift . since @xmath471
, we have @xmath526 .
unless @xmath6 lies on a cycle with no @xmath122-locations , that is , @xmath527 , @xmath6 lies between @xmath414 and @xmath528 for exactly one @xmath414 , and @xmath529 is performed to an extent of 1 ; this holds even if @xmath26 lies on a shift path during the corresponding type-3 swap involving @xmath6 .
all other times @xmath6 when is moved , it is due to the same reasons as in the previous case , so the total change in movement cost for facility @xmath6 is at most @xmath530 adding up the appropriate expression for each facility accounts for the expected change in total movement cost .
theorem [ thm:3appx ] by local optimality , the change in total cost for every test swap ( counting every random choice ) is nonnegative . by averaging over the various swaps ,
the expected change in total cost is nonnegative , so the sum of the expressions in lemmas [ lem : client ] and [ lem : fac ] is nonnegative . to generate an inequality involving a @xmath531 term for facilities @xmath527
, we sum the bound given by lemma [ cycleswap ] here over all cycles of @xmath408 involving only facilities @xmath6 with @xmath532 .
this yields @xmath533 , and we can bound @xmath534 by @xmath535 . adding this to the inequality that the expected change in total cost is nonnegative gives @xmath536 .
the analysis in section [ 3apx ] ( as also the proof of the 5 approximation ) extends easily to the weighted generalization , wherein each facility @xmath6 has a weight @xmath537 and the cost of moving @xmath6 to @xmath53 is given by @xmath538 , to yield the same @xmath49-approximation guarantee . with the exception of one small difference in the analysis , this requires only minor changes in the arguments .
we discuss these briefly in this section .
unless otherwise stated , the same notation from section [ 3apx ] is used in this section .
we emphasize that @xmath59 and @xmath57 now refer to the weighted movement cost of facility @xmath6 in the global or local optimum , respectively .
so , @xmath539 and @xmath540 .
one difference in notation is that the definition of @xmath122 is slightly revised to this weighted setting : @xmath464 if @xmath541 , or @xmath542 and @xmath543 , where @xmath109 is such that @xmath544 ( equivalently , @xmath545 is a center edge in @xmath396 ) .
if all facility weights are 1 , then this definition of @xmath122 agrees with the definition in section [ 3apx ] .
similarly , we say @xmath471 if @xmath546 and @xmath547 . under these definitions ,
similar to claim [ s2dist ] , we now have that @xmath548 ( since @xmath549 for any @xmath550 as before , and @xmath551 ) .
we consider the same set of test swaps and the same averaging of the inequalities generated by these swaps .
when a test swap is performed , we consider the same shift and reassignment of facilities to generate the inequalities . in most cases
, we also move the clients in the same way as before with the exception that if a client @xmath11 has all of @xmath552 and @xmath491 being closed , then we do not necessarily send it to @xmath553 where @xmath6 is such that @xmath554 .
this is discussed in lemma [ lem : w_clientbound ] . as in the discussion before lemma [ lem
: clientbound ] , we consider a swap involving subtrees @xmath555 .
let @xmath477 be as before , and let @xmath556 be the set of facilities @xmath6 such that @xmath6 is a leaf in some @xmath480 or , if the swap is a type-2 swap , that @xmath6 is the first facility in @xmath470 .
note that @xmath496 if and only if the unique @xmath392 arc in @xmath408 is a center arc with @xmath557 .
we let @xmath558 , and @xmath504 also be defined as in section [ 3apx ] .
[ lem : w_clientbound ] the expected change in client assignment cost for a test swap is at most @xmath559 consider one particular swap .
as in the proof of lemma [ lem : clientbound ] , we move @xmath560 to @xmath488 and @xmath561 to @xmath491 and bound their assignment cost change in the same way . as before , it may be that for some of these clients @xmath561 we have that @xmath491 was closed in the swap .
for such clients , we do the following slight variant of the reassignment that was done in the proof of lemma [ lem : clientbound ] .
suppose @xmath392 is the center edge such that @xmath562 for a client @xmath561 , and @xmath32 is not open .
if @xmath563 , then we send @xmath11 to either @xmath553 or @xmath564 . as in the proof of lemma [ lem : clientbound ] , one of these must be open and the total cost of moving @xmath11 from @xmath32 to either @xmath553 or @xmath564 is at most @xmath565 .
otherwise , if @xmath566 then we send @xmath11 to either @xmath110 or @xmath567 ( one of them must be open ) .
the distance from @xmath32 to either @xmath110 or @xmath567 is bounded by @xmath568 .
we conclude by noting that each facility @xmath569 has at most @xmath570 clients sent to either @xmath239 or @xmath571 from @xmath572 in the manner just described , because @xmath573 must be in @xmath364 . similarly , each @xmath574 has at most @xmath570 clients sent to either @xmath571 or @xmath572 from @xmath239 in the manner described above , since @xmath575 .
so , the total client movement charged to @xmath576 this way is at most @xmath577 . using the same weighting of the swaps as in section [ 3apx ] we get the following bound on the contribution of
the client movement cost changes over these swaps .
the proof is nearly identical , except we notice that a facility @xmath109 lies in the @xmath556-set for various swaps to an extent of at most @xmath319 ( under this weighting ) , since the facility @xmath6 such that @xmath392 is a center edge lies in some @xmath477-set to an extent of at most @xmath319 .
[ lem : w_client ] the expected total client assignment cost change , weighted in the described manner , is at most @xmath501 .
the contribution of the facility movement costs is bounded in essentially the same way as in lemma [ lem : fac ] .
we just provide the details on how to account for the weights of the facilities .
as before , let @xmath509 be the set of facilities @xmath6 that do not lie on a cycle in @xmath408 consisting solely of facilities @xmath109 with @xmath510 .
[ lem : w_fac ] when @xmath578 is performed , we move facilities @xmath6 from @xmath26 to @xmath579 . if this shift was performed during a path swap , then the movement - cost change for a facility @xmath6 moved in this shift is at most @xmath580 so the same bound used before applies .
if such a shift was performed along a path in a subtree , then we did not want to bound @xmath534 by @xmath581 because we do not want to cancel the contribution of @xmath582 to the bound .
however , this only happened when @xmath583 so we can use the fact that @xmath584 is large and that @xmath549 for any @xmath585 . in our setting , as noted earlier , the movement cost @xmath586 can be bounded by @xmath587 , which is the same upper bound we used in the unweighted case .
finally , the only other time we moved a facility was from some @xmath471 to @xmath588 .
the cost of this move is now @xmath589 which can also be bounded by @xmath590 using the same argument as in the previous paragraph .
so , all bounds for the unweighted facility movement cost increase averaged over the swaps also hold in the weighted case .
finally , we remark that the same bound for the facility movement cost for facilities on a cycle with only @xmath126 facilities holds for the weighted case , again using arguments like in the proof of lemma [ lem : w_fac ] to bound @xmath586 .
thus , the proof of theorem [ thm:3appx ] is adapted to prove the following result for the weighted case .
the cost of a locally - optimal solution using @xmath24 swaps is at most @xmath591 times the optimum solution cost in weighted instances of mobile facility location .
we show in this section that the local - search algorithm has a constant approximation guarantee also when @xmath592 ( which corresponds to @xmath593 in section [ 5apx ] ) , even in the weighted setting .
this requires a different analysis than in section [ 5apx ] since we now no longer have the luxury of amortizing the `` expensive '' terms in an interval swap via multi - location swaps .
the approximation factor we obtain in the analysis below is large , but we emphasize that we have not sought to optimize this constant . also , we remark that the analysis can be significantly simplified and improved in the unweighted setting .
recall that @xmath594 and @xmath595 .
we use much of the same notation as in section [ 5apx ] . the digraph @xmath79 , its decomposition into paths @xmath80 and cycles @xmath81 , and the definition of @xmath89 are as in section [ 5apx ] .
thus , for a path @xmath113 , we have @xmath596 . define @xmath597 .
let @xmath598 with @xmath599 .
we place @xmath26 in @xmath122 if @xmath600 or @xmath601 ; otherwise we place @xmath26 in @xmath126 . also define @xmath602 .
let @xmath603 for @xmath90 .
[ 1sws2lem ] let @xmath532 and @xmath604 , and consider @xmath605 .
we have @xmath606 we reassign clients in @xmath607 to @xmath54 , and each client @xmath11 in @xmath608 to @xmath137 incurring a total assignment - cost change of @xmath609 .
the change in movement cost is at most @xmath610 where the last inequality follows since @xmath532 . adding this to the expression for the change in assignment cost yields the lemma . in lemmas [ pclem][cyclem ] ,
we generate inequalities that will allow us to bound the total movement cost , and the total assignment cost of clients in @xmath256 .
we use the following notation for this sequence of lemmas .
given a path or cycle @xmath258 , let @xmath611 , where @xmath181 is the next @xmath122-location on @xmath33 after @xmath183 .
let @xmath185 for @xmath202 , and @xmath612 .
let @xmath613 and @xmath614 for @xmath615 ( see fig . [ labeling ] ) .
let @xmath616 denote the @xmath293 subpath of @xmath33 .
consider an @xmath617 subpath of @xmath33 .
let @xmath618 consist of all such @xmath32 where @xmath619 and @xmath620 .
let @xmath621 consist of all such @xmath32 where @xmath619 and @xmath622 .
note that if @xmath623 then @xmath624 and @xmath625 .
let @xmath626 consist of all such @xmath32 where @xmath620 .
clearly , @xmath627 and @xmath628 .
when @xmath33 is clear from the context ( as in lemmas [ pclem][cyclem ] ) , we drop the subscript @xmath33 from @xmath629 .
[ pclem ] let @xmath258 be such that @xmath630 .
then , @xmath631{0.88in}{if $ z\in{\ensuremath{\mathcal p}}$ with \\ $ s'_0={\ensuremath{\mathsf{start}}}(z)$ \\ $ o'_r={\ensuremath{\mathsf{end}}}(z)$ } } & \ & 0\ & \leq\ \sum_{i\in z}(f^*_i - f_i ) + \sum_{j\in d^*((z\cap o){\ensuremath{\setminus}}\{o'_r\})}\negthickspace\negthickspace\negthickspace \tfrac{2}{3}(c_j+c^*_j ) + \sum_{j\in d^*(o'_r)}\negthickspace ( c^*_j - c_j ) + \sum_{j\in d^*(s'_0)}\negthickspace 2c^*_j . \label{pcineq1 } \\[1ex ] \text{\parbox{0.88in}{if $ z\in{\ensuremath{\mathcal{c}}}$ } } & \ & 0\ & \leq\ \sum_{i\in z}(f^*_i - f_i ) + \sum_{j\in d^*(z\cap o)}\tfrac{2}{3}(c_j+c^*_j ) .
\label{pcineq2}\end{aligned}\ ] ] first , suppose @xmath33 is a path . for every @xmath632 ,
we have that @xmath517 , and so we have @xmath633 .
thus , the move @xmath634 yields the following inequality , which implies .
@xmath635 for a cycle @xmath33 , analogous to lemma [ cycleswap ] , we have @xmath636 , and this coupled with the above bound on @xmath637 for @xmath34 yields .
[ pathlem1 ] let @xmath262 ( with @xmath638 as defined above ) .
let @xmath307 , and @xmath639 .
suppose there is no @xmath276 such that @xmath640 .
then , @xmath641 we derive by taking a weighted combination of inequalities generated from various swap moves .
we assume @xmath306 ( otherwise implies ) . for a predicate @xmath642 ,
let @xmath643 be the indicator function that is 1 if @xmath642 is true and 0 otherwise . to keep notation simple
, we follow the convention that a non - existent object ( like @xmath644 ) is @xmath645 , and @xmath646 for every set @xmath427 ( e.g. , @xmath647 for every @xmath427 . ) also , @xmath648 , and @xmath649 . * ) * = the first swap move is @xmath634 , but we bound the change in cost slightly differently .
we only reassign clients in @xmath650 , obtaining the inequality @xmath651 define @xmath652 .
note that @xmath653 $ ] if @xmath654 .
consider each edge @xmath655 . if @xmath656 and @xmath657 , we simply bound @xmath637 by @xmath658 . otherwise ,
if @xmath659 , we use the bound @xmath660 which is valid since @xmath661 .
if @xmath662 and @xmath663 , then @xmath622 and we bound @xmath637 by @xmath664 .
incorporating these bounds in , we obtain the following inequality .
@xmath665 for every @xmath615 such that @xmath666 ,
we consider the move @xmath667 .
we move each facility @xmath668 to @xmath149 if @xmath669 and to @xmath218 otherwise .
note that for every facility @xmath6 such that @xmath670 , we have @xmath517 , and so @xmath633 . we reassign all clients in @xmath671 to @xmath218 , and reassign each client @xmath11 in @xmath672 to @xmath137 .
the resulting change in assignment cost is at most @xmath673 . if @xmath674 , we reassign all clients in @xmath675 to @xmath676 and bound the resulting change in assignment cost by @xmath677 . therefore , if @xmath666 , we obtain the inequality @xmath678 for every @xmath276 such that @xmath679 , we again consider @xmath667 .
we move facilities on @xmath616 and reassign clients in @xmath680 as in case 2 ) .
we reassign clients in @xmath681 to @xmath682 if @xmath683 , and to @xmath684 otherwise .
the assignment - cost change due to this latter reassignment is at most @xmath685 if @xmath683 , and @xmath686 otherwise , where to obtain the latter inequality we use the fact that @xmath687 since @xmath688 and @xmath689 for all @xmath690 .
we obtain the inequality @xmath691 for every @xmath276 such that @xmath692 , we pick some arbitrary @xmath191 and consider @xmath693 . we analyze this by viewing this as a combination of : ( i ) a shift along @xmath86 , ( ii ) moving @xmath6 from @xmath30 to @xmath183 , where @xmath694 , and ( iii ) a shift along the appropriate subpath of @xmath33 . the resulting inequality we obtain
is therefore closely related to , .
we incur an additional @xmath695 term for the change in movement cost due to ( i ) and
( ii ) , and @xmath696 for reassigning clients in @xmath697 .
also , since @xmath183 is no longer swapped out , we do not incur any terms that correspond to reassigning clients in @xmath698 .
thus , we obtain the following inequality : @xmath699 we are finally ready to derive . an inequality subscripted with an index , like @xmath700 , denotes that inequality for that particular index
. it will be useful to define [ segineq ]
@xmath701 we take the following linear combination .
@xmath702 the lhs of is 0 .
we analyze the contribution from the @xmath324 terms to the rhs .
facilities @xmath703 contribute at most @xmath704 .
consider @xmath34 .
if @xmath705 , we pick up @xmath706 from parts 1 and 2 , and we may pick up an additional @xmath707 from part 2 if @xmath708 .
so overall , we obtain a contribution of at most @xmath709 .
next suppose @xmath108 .
if @xmath710 and @xmath657 , we gather @xmath711 from part 1 and @xmath712 from part 2 , so the total contribution is @xmath713 .
suppose @xmath714 .
notice then that @xmath657 , otherwise if @xmath715 , then @xmath716 , which contradicts our assumptions .
recall that @xmath652 .
we gather @xmath717 from part 1 , and @xmath718\leq 6.25f^*_i-2.5f_i\bigl(1-\frac{n^*_{o_i}}{w_i}\bigr)$ ] from part 2 . thus , we gather at most @xmath719 . if @xmath710 and @xmath663 , then note that actually @xmath720 .
we gather @xmath721 from part 1 , and @xmath722 $ ] from part 2 , so the total contribution is at most @xmath709 .
we now bound the @xmath723-contribution .
part of this is @xmath724 .
we proceed to analyze the remaining contribution .
the clients whose remaining contribution is non - zero are of two types : ( i ) clients in @xmath725 , which are reassigned when a location in @xmath726 is swapped in or a location in @xmath727 is swapped out ; and ( ii ) clients in @xmath728 , where @xmath729 , which are charged when we bound the cost @xmath637 or when @xmath663 is swapped out .
consider a client @xmath730 .
let @xmath731 .
its ( remaining ) part-1 contribution is @xmath732 if @xmath714 or @xmath663 , and 0 otherwise .
the part-2 contribution is at most @xmath733 ( this happens when @xmath714 ) ; so the total ( remaining ) contribution is at most @xmath734 . for @xmath735 , where @xmath736 , we know that @xmath737 , and so we have already accounted for its contribution of at most @xmath738 above . finally , consider @xmath739 , where @xmath740 .
note that @xmath517 .
we gather @xmath732 from part 1 if @xmath663 and 0 otherwise , and @xmath741 from part 2 if @xmath663 and @xmath742 otherwise ; so in total we gather at most @xmath743 . putting everything together ,
leads to inequality .
[ pathlem2 ] let @xmath262 , @xmath307 and @xmath261 .
we have @xmath744 we focus on the case where there exists an index @xmath276 such that @xmath745 , @xmath679 , and @xmath746 as otherwise , immediately implies .
this case is significantly more involved , in part because when @xmath747 and @xmath683 , we accrue both the term @xmath748 in when @xmath183 is swapped out , and the term @xmath749 in when @xmath181 is swapped out .
hence , there is no way of combining to get a compound inequality having both @xmath750 and @xmath751 on the rhs . in order to achieve this
, we define a structure called a _ block _ , comprising multiple @xmath616 segments , using which we define additional moves that swap in @xmath218 but swap out neither @xmath183 nor @xmath181 , so that the extent to which @xmath218 is swapped in exceeds the extent to which @xmath183 or @xmath181 are swapped out .
we call a set @xmath752 of consecutive @xmath40-locations a _ block _ , denoted by @xmath753 , if ( i ) @xmath754 and @xmath755 for all @xmath756 , ( ii ) @xmath757 ( recall that if @xmath758 is non - existent , then this condition is not satisfied ) , and ( iii ) @xmath759 or @xmath760 .
note that by definition , any two blocks correspond to disjoint subpaths of @xmath33 .
we say that @xmath761 are _ siblings _ , denoted by @xmath762 , if @xmath763 and they belong to a common block ; note that this means that neither @xmath53 nor @xmath77 is at the start of a block .
we use @xmath764 to denote that @xmath53 and @xmath77 are not siblings .
let @xmath765 . before defining the additional swap moves for each block ,
we first reconsider @xmath634 and account for the change in cost due to this move differently to come up with a slightly different inequality than .
we again start with , and bound @xmath637 in different ways for an @xmath766 edge of @xmath33 .
we use @xmath767 incorporating this in yields the following inequality .
@xmath768 for every block @xmath753 , we consider the swap moves @xmath769 for all @xmath770 .
for each such @xmath771 , we obtain the inequality @xmath772 where we bound @xmath773 by @xmath774 . also consider @xmath775 if @xmath776 or @xmath777 if @xmath692 , where @xmath86 is some arbitrary path in @xmath778 .
note that if @xmath679 , then @xmath779 .
this yields the inequality @xmath780 we now derive by taking the following linear combination of : @xmath781 as before , the lhs of is 0 , and we analyze the contribution from the @xmath324 terms to the rhs .
many of the terms are similar to those that appear in , so we state these without much elaboration .
facilities @xmath703 contribute at most @xmath782 .
consider @xmath34 .
if @xmath783 , then note that @xmath188 does not belong to a block , and we gather at most @xmath784 from parts 1 and 2 .
suppose @xmath108 .
if @xmath785 , we gather @xmath786 from part 1 . let @xmath787 .
if @xmath183 does not belong to any block then we pick up @xmath788 from part 2 .
otherwise , note that @xmath183 must be the start of a block , therefore , we pick up @xmath789 from parts 3 and 4 .
so the overall contribution is at most @xmath790 .
next , suppose @xmath791 .
let @xmath792 .
we pick up @xmath786 from part 1 , @xmath793 $ ] from part 3 , and @xmath794 $ ] from part 4 .
this amounts to at most @xmath795 total contribution .
suppose @xmath714 .
we gather @xmath796 from part 1 . if @xmath657 , we gather @xmath797 $ ] from parts 2 , 3 , and 4 . if @xmath663 , then @xmath798 is the end of some block , and we gather @xmath799 + 9(f^*_i - f_i)$ ] from parts 2 , 3 , and 4 .
so the total contribution in both cases is at most @xmath800 . if @xmath801 and @xmath663 , then @xmath798 is the end of some block .
we gather @xmath802 from part 1 , and @xmath803 + 9(f^*_i - f_i)$ ] from parts 3 and 4 , which amounts to at most @xmath804 total contribution .
finally , if @xmath720 , @xmath663 and @xmath805 , we gather @xmath806 from part 1 . if @xmath26 does not belong to a block , we gather @xmath807 $ ] from part 2 . if @xmath26 belongs to a block , then notice that it can only be the start @xmath183 of the block .
so @xmath517 and we gather @xmath808 from parts 3 and 4 . accounting for both cases , we gather at most @xmath809 .
now consider the @xmath723-contribution .
this includes the terms @xmath810 and @xmath811 .
we bound the remaining contribution . consider a client @xmath730 .
let @xmath731 .
its ( remaining ) part-1 contribution is @xmath812 if @xmath714 or @xmath813 , and 0 otherwise .
if @xmath183 is not in any block , we pick up at most @xmath814 from part 2 .
if @xmath183 is the start of a block , we pick up @xmath815 from part 3 , and note that @xmath816 . if @xmath183 is an intermediate @xmath40-location of a block , then @xmath715 and @xmath791 .
we pick up @xmath815 from part 3 , and @xmath817 from part 4 .
so the total ( remaining ) contribution is at most @xmath818 in all cases . now consider
@xmath739 , where @xmath740 .
we have @xmath517 .
let @xmath819 .
we accrue @xmath820 from part 1 if @xmath663 and 0 otherwise .
if @xmath183 does not belong to a block , we gather at most @xmath821 from part 2 ; if @xmath183 belongs to a block , it must be the start of the block , and we gather at most @xmath822 from part 3 .
so in total , we accrue at most @xmath823 .
thus , leads to inequality .
[ cyclem ] let @xmath145 and @xmath824 .
define @xmath275 for all @xmath276 .
then , @xmath825 the arguments are almost identical to those in the proofs of lemmas [ pathlem1 ] and [ pathlem2 ] .
the only change is that we no longer have inequality or .
instead , we start with the inequality @xmath636 , and we bound @xmath637 suitably , as in lemma [ pathlem1 ] or lemma [ pathlem2 ] , to obtain and that are analogous to and respectively .
@xmath826 & \begin{split } 0\ \leq\ & \sum_{\substack{i\in z : s_i{\ensuremath{\approx}}{\ensuremath{\sigma}}(o_i ) \text { or } \\ { \ensuremath{\sigma}}(o_i)\in z{\ensuremath{\setminus}}a , s_i\notin b } } \negthickspace\negthickspace 2f^*_i + \sum_{\substack{i\in z : { \ensuremath{\sigma}}(o_i)\in a\text { or } \\ { \ensuremath{\sigma}}(o_i)\in c{\ensuremath{\setminus}}a , s_i\in b } } \negthickspace\negthickspace\bigl(2f^*_i-(1-{\ensuremath{\alpha}}_i)f_i\bigr ) + \sum_{\substack{i\in z : { \ensuremath{\sigma}}(o_i)\in z{\ensuremath{\setminus}}c \\ s_i\in b , s_i{\ensuremath{\not\approx}}{\ensuremath{\sigma}}(o_i ) } } \negthickspace\negthickspace(f^*_i - f_i ) \\ & + \sum_{\substack{i\in z : { \ensuremath{\sigma}}(o_i)\in a\text { or } \\
s_i\in b , s_i{\ensuremath{\not\approx}}{\ensuremath{\sigma}}(o_i ) } } \sum_{j\in d^*(o_i)}\tfrac{2}{3}(c_j+c^*_j ) .
\end{split } \label{cineq2}\end{aligned}\ ] ] the rest of the proof proceeds as in lemmas [ pathlem1 ] and [ pathlem2 ] : we substitute for in the proof of lemma [ pathlem1 ] , and substitute for in the proof of lemma [ pathlem2 ] .
it is not hard to see then that we obtain inequality .
[ 1swthm ] the cost of a local optimum using 1-swaps is at most @xmath35 times the optimum solution cost .
let @xmath327 and @xmath82 denote respectively the total movement- and assignment- cost of the local optimum .
for a set @xmath67 , let @xmath827 denote @xmath828 .
summing for all @xmath140 and simplifying , we obtain that @xmath829 summing for every path @xmath262 , and or for every cycle @xmath145 , yields the following .
@xmath830 adding @xmath831 to , we get that @xmath832
in this section , we present an example that shows that if the facility - movement costs and the client - assignment costs come from different ( unrelated ) metrics then the @xmath24-swap local - search algorithm has an unbounded locality gap ; that is , the cost of a local optimum may be arbitrarily large compared to optimal cost .
we first show a simple example for a single swap case , which we will later generalize for @xmath24 swaps .
suppose we have two clients @xmath833 and two facilities @xmath834 .
some distances between these clients and facilities are shown in the fig .
[ fig : locgap](a ) ; all other distances are obtained by taking the metric completion . note that in this example , in order to have a bounded movement cost for facilities , the only option is to have one of @xmath835 as a final location of facility @xmath836 and one of @xmath837 as a final location of facility @xmath460 . as can be seen from the figure ,
the solution @xmath838 has total cost @xmath839 ( the movement cost is @xmath840 and the client - assignment cost is @xmath839 ) . now consider the solution @xmath841 which has a total cost of @xmath842 ( the movement cost is @xmath842 and the client - assignment cost is @xmath840 ) .
this is a local optimum since if we swap out @xmath843 , then we have to swap in @xmath460 to have a bounded movement cost , which leads @xmath843 having assignment cost of @xmath844 . by symmetry , there is no improving move for solution @xmath15 , and the locality gap is @xmath25 .
now consider the example shown in fig .
[ fig : locgap](b ) for local - search with @xmath24 simultaneous swaps .
suppose we have facility set @xmath845 and client set @xmath846 .
the global optimum @xmath847 has total cost @xmath848 ( facility movement cost is @xmath840 and client - assignment cost is @xmath849 ) while @xmath850 is a local optimum whose total cost is @xmath851 ( facility movement cost is @xmath851 and client - assignment cost is @xmath840 ) .
consider any move @xmath852 .
note that @xmath853 iff @xmath854 ( where indices are @xmath855 ) to ensure bounded movement cost .
let @xmath4 be such that @xmath853 and @xmath856 .
then , @xmath857 has an assignment cost of @xmath844 in the solution @xmath858 .
hence , @xmath15 is a local optimum . | we consider the _ mobile facility location _
( ) problem .
we are given a set of facilities and clients located in a common metric space @xmath0 .
the goal is to move each facility from its initial location to a destination ( in @xmath1 ) and assign each client to the destination of some facility so as to minimize the sum of the movement - costs of the facilities and the client - assignment costs .
this abstracts facility - location settings where one has the flexibility of moving facilities from their current locations to other destinations so as to serve clients more efficiently by reducing their assignment costs .
we give the first _ local - search based _ approximation algorithm for this problem and achieve the best - known approximation guarantee .
our main result is @xmath2-approximation for this problem for any constant @xmath3 using local search .
the previous best guarantee for was an 8-approximation algorithm due to @xcite based on lp - rounding .
our guarantee _ matches _ the best - known approximation guarantee for the @xmath4-median problem .
since there is an approximation - preserving reduction from the @xmath4-median problem to , any improvement of our result would imply an analogous improvement for the @xmath4-median problem .
furthermore , _ our analysis is tight _ ( up to @xmath5 factors ) since the tight example for the local - search based 3-approximation algorithm for @xmath4-median can be easily adapted to show that our local - search algorithm has a tight approximation ratio of 3 .
one of the chief novelties of the analysis is that in order to generate a suitable collection of local - search moves whose resulting inequalities yield the desired bound on the cost of a local - optimum , we define a tree - like structure that ( loosely speaking ) functions as a `` recursion tree '' , using which we spawn off local - search moves by exploring this tree to a constant depth .
our results extend to the weighted generalization wherein each facility @xmath6 has a non - negative weight @xmath7 and the movement cost for @xmath6 is @xmath7 times the distance traveled by @xmath6 . |
the purpose of this note is to investigate the relation between rectifiability and * radial projections * @xmath5 , @xmath6 , defined by the expression @xmath7 .
motivated by the famous result of besicovitch and federer concerning orthogonal projections ( see ( * ? ? ?
* theorem 18.1 ) ) , one has reason to anticipate that the radial projections @xmath8 of @xmath0-rectifiable sets @xmath1 with @xmath9 ought to have positive @xmath10 measure for almost all parameters @xmath6 , whereas the opposite behaviour should be manifest in the purely unrectifiable case .
the problem of verifying this intuition in the unrectifiable case appears to be rather involved , and only partial results are available ( see the discussion and references in section [ unrectifiable ] ) ; this note complements those results by settling the rectifiable case .
we show that if no @xmath0-plane contains almost all of an @xmath0-rectifiable set @xmath1 with @xmath9 , then @xmath10 almost all radial projections of @xmath4 have positive @xmath0-dimensional measure .
we also provide an accurate worst case description of the geometry of the * exceptional set * @xmath11 .
our studies were initiated by a question on directions of rectifiable sets , raised by iosevich , mourgoglou and senger in the recent preprint @xcite : a positive answer to this question is acquired in remark [ ims ] .
an @xmath10 measurable set @xmath12 with @xmath13 is called @xmath0-*rectifiable * if there are countably many @xmath14-embeddings @xmath15 such that @xmath16 this definition is equivalent to the classical definition of rectifiability , where we consider coverings by lipschitz - images instead of @xmath14-manifolds , see ( * ? ? ?
* theorem 15.21 ) .
we include the requirement @xmath9 in the definition of rectifiability merely to avoid repetition .
we say that a set @xmath12 is @xmath0-*flat * , if there exists an @xmath0-plane @xmath17 such that @xmath18 . relaxing this condition slightly
, we say that @xmath4 is * essentially * @xmath0-*flat * , if @xmath19 .
our main result is the following .
[ thm1 ] let @xmath1 be an @xmath0-rectifiable set , which is not essentially @xmath0-flat .
then @xmath20 for @xmath21 almost every @xmath22 . is not , strictly speaking , well defined , if @xmath23 . here , and
in the rest of the paper , the notation @xmath8 should be interpreted as @xmath24 .
] moreover , the exceptional set is always @xmath25-flat .
a @xmath26-plane is , by definition , a singleton in @xmath27 .
note that the result fails for any essentially flat set @xmath1 .
indeed , if @xmath4 essentially @xmath0-flat , then we may find an @xmath0-plane @xmath17 such that @xmath28 .
write @xmath29 the first set on the right is @xmath21 null , since @xmath30 .
for every @xmath31 , the second one is a subset of an @xmath25-sphere .
hence @xmath32 for every @xmath31 .
furthermore , the exceptional set can be an @xmath25-plane .
for this just take any two @xmath0-planes @xmath33 and @xmath34 such that @xmath35 but @xmath36 is non - empty .
now @xmath3 is an @xmath25-plane .
define @xmath37 .
then , for every @xmath38 , the radial projection @xmath39 consists of two @xmath25-spheres .
hence , @xmath32 for every @xmath38 .
the main result is a consequence of the following lemma , which concerns the local bilipschitz - properties of the radial projection .
[ lemma1 ] let @xmath40 be a @xmath41-embedding , and let @xmath42 be a point such that @xmath43 . then ,
if @xmath44 , there exists @xmath45 such that @xmath46 is bilipschitz .
we may and will assume that @xmath47 .
write @xmath48 , and fix @xmath49 . since @xmath50 is differentiable at @xmath51 , say , we have @xmath52({\mathbf{y}}- { \mathbf{z } } ) + { \varepsilon}({\mathbf{y}}- { \mathbf{z}})|{\mathbf{y}}- { \mathbf{z}}| , \end{aligned}\ ] ] where @xmath53 is , as usual , some function with the property that @xmath54 as @xmath55 in @xmath56 .
denote the three vectors on the previous line by @xmath57 and @xmath58 .
since @xmath43 , we have the inequality @xmath59 for some @xmath60 . on the other hand , choosing @xmath61 and @xmath62 close to @xmath47 ensures that @xmath63 and @xmath51 are close to @xmath64 , since @xmath50 is an embedding .
it follows that @xmath65 as soon as @xmath66 . above @xmath67
can be made arbitrarily small by taking @xmath45 small .
here we also needed the fact that @xmath50 is continuously differentiable .
denote by @xmath68 the orthogonal projection onto @xmath69 . since @xmath70
, we may estimate @xmath71 the factor of @xmath72 tends to one as @xmath73 , so this may be re - written more neatly as @xmath74 next fix @xmath75 . by making @xmath76 yet smaller if necessary , we may assume that @xmath77 for any @xmath78 . hence there exists @xmath79 such that @xmath80 for @xmath81 .
now we claim that there exists @xmath60 ( not necessarily the same as before ) such that @xmath82 where @xmath83 is the angle formed by @xmath84 and @xmath85
. assume the contrary and locate @xmath66 with @xmath86 .
denote by @xmath87 the orthogonal projection of @xmath84 onto the line spanned by @xmath85 .
then , as the picture shows , @xmath88 hence @xmath89 .
thus we have shown that @xmath90 for some @xmath91 and @xmath92 .
this yields @xmath93 taking @xmath68 on both sides and estimating , @xmath94|y - z| .
\end{aligned}\ ] ] since @xmath67 could be chosen arbitrarily small and @xmath79 , we have reached a contradiction . thus ( [ form1 ] ) holds .
the angle formed by any points @xmath95 and @xmath96 on @xmath97 is precisely @xmath83 , and this angle is of the same order of magnitude as @xmath98 .
thus @xmath99 hence @xmath100 restricted to @xmath101 is bilipschitz .
choose a @xmath14-embedding @xmath102 so that @xmath103 intersects @xmath4 in a set of positive @xmath10 measure .
next choose a point @xmath104 such that @xmath105 and @xmath106 is a density point of @xmath107 .
all this is possible by sard s theorem and the fact that @xmath10 almost every point of @xmath107 is a density point .
write @xmath108 . by our assumption , @xmath109 .
now repeat the previous procedure : choose a @xmath14-embedding @xmath110 such that @xmath111 intersects @xmath112 in a set of positive @xmath10 measure .
also , pick a point @xmath113 such that , again , @xmath114 and @xmath115 is a density point of @xmath116 .
. then @xmath35 .
let @xmath22 . if @xmath118 .
then , by lemma [ lemma1 ] , there exists @xmath45 such that @xmath119 is bilipschitz . since @xmath120 has positive @xmath21 measure , so does @xmath121 .
similarly , if @xmath122 , we find @xmath45 such that @xmath123 \cap m_2 \cap b(z_2,\delta))$ ] has positive @xmath21 measure .
this implies that the set of all @xmath22 with @xmath32 is necessarily a subset of @xmath124 , which is always contained in an @xmath2-plane .
[ ims ] in ( * ? ? ? * conjecture 1.12 ) , iosevich _ et al .
_ asked the following : _
if @xmath4 is @xmath125-rectifiable and not @xmath125-flat , then does the set of directions @xmath126 always have positive @xmath127 measure ? _
lemma [ lemma1 ] implies a positive answer : one may even replace @xmath128 by any integer @xmath129 .
in fact , slightly more is true : if @xmath1 is an @xmath0-rectifiable set , which is not @xmath0-flat , then there exists @xmath23 such that @xmath20
. the proof of this is almost the same as that of theorem [ thm1 ] only simpler . choose a @xmath14-embedding @xmath40 , such that @xmath48 intersects @xmath4 in a set of positive @xmath10 measure .
then choose a point @xmath130 such that @xmath43 and @xmath131 is a density point of @xmath132 .
write @xmath133 . by our assumption @xmath134 , so we may choose a point @xmath135 .
now lemma [ lemma1 ] yields @xmath45 such that @xmath136 is bilipschitz .
since @xmath137 has positive @xmath21 measure , so does @xmath138 .
in analogue with the besicovitch - federer projection theorem , it seems reasonable to conjecture that the following converse for the result in this note is also true : the radial projections @xmath39 of every purely @xmath0-unrectifiable set @xmath12 are @xmath10 null for @xmath21 almost every @xmath23 .
this was recently verified for self - similar sets in @xmath139 satisfying the open set condition by simon and solomyak in @xcite . in @xcite marstrand and mattila also proved that @xmath140 can only happen for @xmath141 in a set of dimension at most @xmath0 . in the plane ,
at least , this upper bound can not be improved : marstrand exhibited an example of a purely @xmath142-unrectifiable set @xmath143 , which projects radially onto a set of positive length in a set of hausdorff - dimension exactly @xmath142 .
a related phenomenon was discovered by csrnyei and preiss in @xcite : given a purely @xmath142-unrectifiable set @xmath143 with @xmath144 , there exists a @xmath142-rectifiable set @xmath145 with @xmath146 such that for @xmath147 almost every @xmath23 almost every line passing through @xmath141 intersects @xmath148 in an infinite set . | we show that if no @xmath0-plane contains almost all of an @xmath0-rectifiable set @xmath1 , then there exists a single @xmath2-plane @xmath3 such that the radial projection of @xmath4 has positive @xmath0-dimensional measure from every point outside @xmath3 . |
massive star formation is a phenomenon of fundamental importance in astrophysics yet a detailed understanding of it has been elusive @xcite .
the present state of theory and numerical simulation research fall into two major categories of processes : `` core accretion '' and `` competitive accretion '' , as recently reviewed by @xcite . in the core accretion scenario , massive stars ( like low - mass stars )
are formed via the collapse of self - gravitating , centrally condensed cores .
these cores are discrete structures within the larger - scale molecular cloud , and each core constitutes the mass reservoir for a single star or small multiple system . as a result ,
core mass maps directly to stellar mass , and the core mass function ( cmf ) maps to the stellar initial mass function ( imf ) ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* and references therein ) .
in contrast , competitive accretion models are intrinsically models of star cluster formation . fragmentation in a cluster - scale gas clump produces many low - mass protostars , which then accrete ( competitively ) from the large - scale gas reservoir . in this picture , massive stars _
must _ form in a cluster environment , and massive stars and their surrounding clusters must form simultaneously ( e.g. * ? ? ?
* ; * ? ? ?
* and references therein ) .
there are several significant observational constraints that these theories must face .
first , there is mass segregation the fact that the most massive members of young clusters are concentrated in the center ( e.g. * ? ? ?
it remains unclear whether this property is primordial or a result of dynamical evolution .
@xmath14-body simulations of star clusters have shown that it is common to form compact groupings of massive stars ( i.e. trapezium - like systems ) near the center of the cluster in as little as one free - fall time , particularly when there is initial substructure @xcite .
observational studies of young clusters caught early in the act of forming are critical to address the origin of mass segregation .
second , there is the correlation between the mass of the most massive star and the number of cluster members @xcite .
an ensuing question is whether the high - mass and low - mass members of clusters form at the same time .
perhaps the only way to answer this question is to search for actively forming low - mass protostars amidst their high - mass counterparts . toward this end , the past decade of advances in
( sub)millimeter interferometry has enabled the identification of proto - trapezium - like systems , termed `` protoclusters '' ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , in which four or more compact millimeter continuum sources exist within a projected diameter of 10000 au . in many cases ,
the protocluster members present a striking diversity in star formation indicators including the presence or absence of h@xmath1o masers , hot core line emission , and free - free emission providing strong evidence that closely neighboring objects can exist in different evolutionary stages of formation @xcite . at last count of the literature ,
half of these massive cores studied at @xmath15 au resolution were resolved into four or more millimeter sources @xcite .
these results emphasize the possibility of interaction between massive protostars , as well as their probable impact on low - mass protostars forming in their midst .
furthermore , given the deeply embedded nature of protoclusters , sensitive high dynamic range millimeter and centimeter imaging will be needed to obtain an accurate census of their low - mass membership .
a third observational constraint on theory is the tentative yet growing evidence for massive keplerian accretion disks around massive protostars
. discovered through subarcsecond angular resolution imaging , there is a steadily increasing number of massive disk candidates around central stars of varying mass @xmath16 .
examples of disk candidates with an enclosed mass of @xmath17 @xmath11 include iras 20126 + 4104 , iras 18360 - 0537 , and orion kl source i @xcite .
larger scale structures ( toroids ) of diameter several thousand au have also been reported which may encompass either an o - type protostar or a cluster of massive protostars ( e.g. * ? ? ?
* ; * ? ? ?
there are also cases of apparent sub - keplerian rotation ( afgl2591-vla3 : * ? ? ?
* ) as well as of a lack of keplerian rotation signatures on 500 au scales ( ngc7538 irs1 : * ? ? ?
most of the candidate disks show a strong bipolar molecular outflow perpendicular to the disk plane , analogous to low - mass protostellar disk / outflow systems .
it is important to note that the presence of such disks does not immediately favor either the core accretion or competitive accretion model , as they are expected to exist under both scenarios .
however , as our knowledge of the massive disk population grows , both theories are certain to face new challenges . to further explore the protostellar population of massive protoclusters ,
we have been pursuing detailed observations of the nearby examples in ngc 6334 , a region containing multiple sites of high mass star formation @xcite . a recent deep near- and
mid - infrared survey revealed over 2200 young stellar object ( yso ) candidates , and subsequent estimates of the star formation rate suggest that it may be undergoing a `` mini - starburst '' event @xcite . at the northeastern end of the region ,
the deeply - embedded source `` i(n ) '' was first identified at 1 mm by @xcite and later detected at 400 by @xcite , who estimated a size of 50 .
single - dish observations of high velocity sio emission indicated significant outflow activity at this location @xcite .
our initial sma observations of ngc 6334 i(n ) at @xmath18 resolution resolved a trapezium - like protocluster of seven compact millimeter continuum sources within a projected diameter of 0.1 pc @xcite .
as revealed by the millimeter spectral line data @xcite , the brightest three continuum sources ( sma 1 , sma 2 , and sma 4 ) are the origin of the hot core line emission seen in the many single - dish molecular line observations of this source @xcite .
hot nh@xmath5 was resolved by 1.5 resolution australia telescope compact array ( atca ) observations of the ( 5,5 ) and ( 6,6 ) lines , which peak toward sma 1 @xcite .
the profile of the ( 6,6 ) line showed a double peak separated by @xmath194 km s@xmath9 , which was interpreted as possibly tracing a rotating circum - protostellar disk @xcite . with similar angular resolution
, @xcite detected a comparable velocity gradient toward sma 1 in a few other hot core molecules .
however , karl g. jansky very large array ( vla ) 7 mm continuum observations with a 0.5 beam resolved sma 1 into multiple components @xcite , making the larger - scale velocity gradient difficult to interpret .
the ngc 6334 i(n ) millimeter protocluster is embedded in a region that is remarkably dim in the mid - infrared _ spitzer _ images , and has the characteristics of an infrared dark cloud ( irdc ) . to provide an overview , figure [ fig0 ] reprises the millimeter and infrared continuum imaging results from @xcite .
despite the millimeter multiplicity , we identified only some extended 4.5 emission associated with the bipolar outflows from sma 1 , 4 and 6 , and a single 24 point source near sma 4 @xcite .
in contrast , our vla detection of copious amounts of 44 ghz class i ch@xmath5oh maser emission to the southeast of the compact millimeter continuum sources is indicative of outflow activity @xcite , specifically in the area surrounding the single - dish ( sub)millimeter continuum source sm2 @xcite .
similarly , our vla h@xmath1o maser observations revealed 11 locations of emission , only 8 of which were associated with the known compact millimeter continuum sources , suggesting the presence of additional young stellar objects @xcite . in this paper , we present new sub - arcsecond sma 1.3 mm , 0.87 mm and vla 6 cm imaging that indeed reveals significant further multiplicity in this protocluster , as well as the detailed kinematics of the hot cores and complex spectral energy distributions .
the details of the observations are summarized in
[ obs ] , while 3 and 4 present our key results and discussion , respectively . for
the distance to ngc 6334 i(n ) , we adopt 1.3 kpc based on recent h@xmath1o maser parallax studies : @xmath20 kpc @xcite and @xmath21 kpc @xcite . in the past ,
the most commonly used value was 1.7 kpc from photometric estimates for the ngc 6334 region @xcite , implying a reduction by a factor of 1.7 for derived quantities based on the distance squared , such as mass and luminosity .
for example , the total luminosity of ngc 6334 i(n ) as measured by @xcite is now @xmath22 @xmath23 with this revised distance .
the details of the sma 1.3 mm and 0.87 mm observations and the nrao vla 6 cm observations are summarized in table [ smaobs ] .
the very extended ( henceforth , vex ) 1.3 mm sma data were calibrated in miriad @xcite , then exported to casa @xcite where the continuum was subtracted to create a line - only vex dataset .
the continuum is then composed of the line - free portions of the dataset .
self - calibration was performed on the continuum data , and solutions were transferred to the line data .
line cubes were generated with a channel spacing of 1.1 km s@xmath9 and robust weighting of 1.0 , with the exception of the 218.44 and 229.76 ghz class i methanol maser transitions which were imaged with robust weighting of 0.0 .
the flux calibration is based on titan , ceres , and sma flux monitoring of the observed quasars and is estimated to be accurate to within 20% .
the same procedure was also followed for the vex 0.87 mm data .
unfortunately , the observing conditions for the 0.87 mm data were significantly worse than for the 1.3 mm data in terms of higher winds and more variable opacity , leading to greater phase instability .
as a result , the 0.87 mm spectral line cubes are too noisy to be useful .
two 1.3 mm continuum images were constructed : ( 1 ) to maximize the continuum sensitivity and minimize artifacts from resolved out structure , the vex 1.3 mm continuum data were combined with the extended configuration ( henceforth , ext ) continuum data presented in @xcite and @xcite and imaged with robust weighting of 0.5 . note that the ext 1.3 mm data only cover a portion of the spectral coverage of the vex data so no such combination was possible for the line data . the relative weight of the individual visibilities between the two configurations is such that the angular resolution of the combination is similar to that of the vex data alone .
this `` ext+vex '' continuum image was used to identify and characterize the 1.3 mm dust continuum sources ; it is not sensitive to smooth structures larger than about @xmath24 .
( 2 ) in an effort to better match the uv - coverage of the vex 0.87 mm data , the line - free portions of the vex 1.3 mm data were imaged with uv - spacings @xmath25k@xmath26 and robust=0 weighting .
this image is henceforth termed the `` vex - uv '' 1.3 mm continuum image .
the 0.87 mm continuum image was also created two ways , both using robust weighting of 1.0 : ( 1 ) a version using all of the data to obtain good angular resolution and sensitivity ; and ( 2 ) a version with a 300k@xmath26 uv - taper to better match the vex 1.3 mm data , which was then subsequently convolved to the same resolution as the vex - uv 1.3 mm continuum image .
this uv - tapered and convolved image was used to determine the 0.87 mm dust continuum properties .
the 0.87 mm images are not sensitive to smooth structures larger than about @xmath27 .
all measurements from all images were obtained from versions that had been corrected for the primary beam response .
the vla data were calibrated in casa with scripts based on the vla pipeline .
imaging and self - calibration were performed manually in casa .
the flux calibration accuracy is estimated to be 10% . due to the large primary beam ( nearly @xmath28 )
, the vla 6 cm data include the bright ultracompact hii region ngc 6334 f as well as the 45-diameter hii region ngc 6334 e @xcite , which are both located south of the field of interest .
we thus imaged a large field ( @xmath29 pixels ) containing these objects in order to minimize the confusion toward ngc 6334 i(n ) . comparing the new 6 cm jvla data to the 1.3 cm and 7 mm vla data presented in ( *
* see also rodriguez et al .
2007 ) , we found a @xmath30 disagreement in the astrometry between the old and new data .
the most likely culprit was the phase calibrator used in the older vla observations j1720 - 358 ( b1717 - 358 ) .
we alerted both vla and alma to our suspicion , which led to an alma calibration observation of j1720 - 358 using the nearby vlba calibrator j1717 - 3342 @xcite at band 3 ( 92 ghz ) as the reference source with 24 antennas ( oct .
29 , 2013 , execution blocks : uid://a002/x70c186/x12 , x25 , and x4a ) . as a result of these observations ,
the alma calibrator database has been updated with a revised position of 17:20:[email protected] , -35:52:[email protected] ( a correction of 0.43 , as of this writing the vla calibrator database has not yet been updated ) .
we have used this new information to correct the astrometry of the 1.3 cm and 7 mm images used in this paper .
we note that our h@xmath1o maser observations used j1717 - 3342 as the phase calibrator @xcite and thus do not require correction .
the 1.3 mm ext+vex image of the continuum emission is shown in figure [ fig1 ] .
all of the seven previously - identified millimeter sources @xcite are detected , with the exception of one of the fainter , more extended sources ( sma 7 ) , whose emission is apparently resolved out by the longer baselines employed in these observations .
we detect many new compact sources in the surrounding field , primarily because the rms noise achieved is nearly 4 times better than @xcite , and the higher angular resolution leads to a lower confusion limit . due to the limited size of the sma primary beam ,
the rms noise in the image corrected for primary beam response increases as a function of angular radius from the phase center .
therefore , we have established our detection threshold corresponding to @xmath31 where @xmath32 is the rms noise measured in an annulus @xmath33 .
we defined four annuli corresponding to the following levels in the casa sensitivity image ( which for a single pointing observation is simply the image of the primary beam response ) : 80%100% , 60%80% , 40%60% , and 20%40% , for which we measured @xmath32 values of 2.2 , 2.5 , 3.4 and 5.4 mjy beam@xmath9 .
the area of the image above the 20% sensitivity level corresponds to 5500 times the area of the synthesized beam .
therefore , we chose @xmath34 because the expected number of false @xmath35 `` peaks '' is only 0.019 .
( for @xmath36 , the value is 0.17 which we considered to be too close to 1 false positive . ) with our @xmath37 criterion , we identify 24 sources , 16 of which are new ( table [ contsources ] ) , and two of which ( sma 20 and 18 ) correspond to two of the brightest h@xmath1o maser components ( c1 and c2 , respectively ; * ? ? ? * ) .
all of the new sources , starting with sma 8 , are numbered in order of increasing right ascension .
the majority of new sources are found in the southern half of the sma primary beam , consistent with the area of extended dust emission seen in the single dish 0.45 mm image ( see figs 1 and 2 ) .
sma 18 coincides with the single - dish source sm2 to within the single - dish position uncertainty @xcite .
we used the casa imfit task to fit each source with a two - dimensional gaussian to find the peak position and integrated flux density and attempt to find the deconvolved size . for 9 of the 24 sources ,
the fitted size was well constrained in both the major and minor axes , and we can compute the brightness temperature ( a lower limit to the physical temperature ) . for 11 sources ,
the minor axis was not well constrained , and the geometric mean of the major and minor axes of the fitted gaussian was less than that of the synthesized beam ( 0.5 ) .
four sources were consistent with a point source in both axes . for the latter two cases , we assign 0.5 as an upper limit to the size and use this to compute a lower limit to the brightness temperature . with the exception of the 24 source near sma 4 , which likely traces hot dust in the walls of the cavity formed by the outflow from that source ( e.g. * ? ? ?
* ) , none of the millimeter continuum sources have counterparts in the mid - infrared _ spitzer _ images @xcite .
the vla 6 cm image ( figure [ fig1 ] ) shows two previously known 6 cm point sources , associated with sma 1 and sma 4 @xcite , plus two new sources that are not associated with 1.3 mm emission .
all four sources are consistent with unresolved compact sources with sizes @xmath38 . their positions and flux densities are listed in table [ cmcontsources ] .
one of the new sources coincides within 0.054 ( 70 au ) with the centroid of h@xmath1o maser component c4 @xcite , hence we call it h@xmath1o - c4 .
the other new source is located @xmath3930 ( 0.2 pc ) south of sma 1 . because it is not coincident with any maser or 1.3 mm source
, we call this source vla 2 in order to distinguish it from both the 1.3 cm source vla - k1 reported by @xcite ( outside the sma field of view ) and the 7 mm source vla 3 reported by @xcite ( which was named for its proximity to sma 3 and yet independent nature )
. the central portion of the ngc 6334 i(n ) protocluster is shown in figure [ sma1zoom ] .
the 1.3 mm emission from sma 1 has been clearly resolved into three components ( a , b+d , and c ) , which are composed of the four sources previously identified at 7 mm . in addition , the east / west extension of sma 1b at 1.3 mm is consistent with the presence of the fourth 7 mm component ( sma 1d ) .
both sma 1b and sma 1d are detected at 1.3 cm and 7 mm while only sma 1b is detected at 6 cm . revised positions of the 1.3 cm and 7 mm counterparts based on the corrected astrometry of those images are given in table [ cmpositions ] . with the new astrometry
, there is now very good agreement between the vla and sma continuum contours for the primary millimeter sources ( sma 1b+d , 4 , and 6 ) .
the angular separation between sma 1b and 1d in the 7 mm image is 0.45 ( 590 au ) at a position angle of -67 ( east of north ) .
the area surrounding sma 6 is shown in figure [ sma6zoom ] .
the peaks of the 7 mm , 1.3 mm and 0.87 mm emission are in good agreement .
there is an extended ridge of 1.3 mm emission to the southwest , part of which may arise from a distinct source .
however , because it is not clearly separated from sma 6 and there is no compact counterpart at any other wavelength , we have not identified this ridge as a separate object . a further zoomed view of the sma 1b+d region
is shown in figure [ sma1bd]a , b in 1.3 mm and 0.87 mm continuum , along with the h@xmath1o maser positions observed with a beam of @xmath40 at p.a.=+7 @xcite and the first moment of the ch@xmath5cn @xmath6=1211 , @xmath41=7 transition . with higher resolution
than the 1.3 mm vex image , the 0.87 mm image clearly indicates that sma 1d is a source of submillimeter emission separate from sma 1b .
further evidence that these are distinct sources comes from the h@xmath1o maser positions : the vast majority are coincident with sma 1b while none are seen toward sma 1d ( see also * ? ? ? * ) .
the position angle of the spatial extent of the h@xmath1o masers is somewhat inclined to that of the large - scale bipolar outflow axis measured in sio 54 @xcite .
also , the kinematics are complicated , and reminiscent of the h@xmath1o maser system of radio source i in orion bn / kl ( see e.g. * ? ? ?
* ) in terms of spatial extent ( @xmath19500 au ) and the apparent overlap of widely disparate velocities . similar to the h@xmath1o emission , the ch@xmath5cn emission is centered on sma 1b ( see [ hotcore ] ) , with no measurable emission coming from sma 1d . in an attempt to apportion the 1.3 mm continuum flux density between sma 1b and 1d ,
we fit a single two - dimensional gaussian to sma 1b in the 1.3 mm vex - uv continuum image .
the residual image showed a weak source of unresolved emission coincident with the 7 mm and 1.3 cm source sma 1d . to estimate the flux density of this source
, we fit the residual image with a single gaussian , which yielded a point source of @xmath42 mjy at the j2000 position : 17:20:[email protected] , -35:45:04.14 @xmath10 0.07 .
the angular separation of this position from sma 1b is [email protected] at position angle -69@xmath1010 , in good agreement with the 7 mm separation between sma 1b and 1d .
finally , we estimate the flux density of sma 1b alone to be the joint flux density of sma 1b+d from table [ contsources ] minus the value for sma 1d , or @xmath43 jy . a similar procedure
was then performed on the 0.87 mm image , yielding a flux density of @xmath44 mjy for sma 1d .
eight of the 1.3 mm sources are detected in the uv - tapered 0.87 mm image .
note that 10 of the 16 non - detected sources ( including vla 2 ) lie beyond the one - third sensitivity radius of the primary beam , where the @xmath37 limit is @xmath45 jy beam@xmath9 . using this image , the fitted positions , flux densities , and sizes
are listed in table [ contsourcesband7 ] .
we caution that the uv - tapered 0.87 mm image is still not sensitive to the largest angular scales that the 1.3 mm ext+vex image contains .
therefore , the 0.87 mm flux densities should be taken as lower limits for those sources having 1.3 mm fitted sizes significantly larger than @xmath46 , particularly those located in the complicated central cluster . for the 1.3 mm sources smaller than @xmath46 and located away from the central cluster ( including sma 13 and sma 15 ) , the 0.87 mm measurements are unlikely to suffer from missing flux and may be considered to be accurate .
in our sma 1.3 mm data cubes , sma 1b , sma 2 and sma 4 all show copious hot core line emission , as do sma 6 , sma 15 , and sma 18 to a lesser extent . at the current sensitivity ,
1.3 mm spectral line emission is not detected toward any of the other cores .
the spectral line emission is dominated by spatially compact emission from complex molecules .
indeed , most of the more abundant species that exhibited outflow emission ( e.g. co , @xmath47co , sio , dcn , etc . ) in the lower resolution sma data presented in @xcite are mostly resolved out in the very extended configuration data . as a result ,
the channels with these molecules have significantly higher noise ( by factors of 3 - 10 ) due to imaging artifacts caused by the lack of short spacing information .
from the 8 ghz of available 1.3 mm spectral bandwidth we identified for detailed analysis 12 transitions from 9 different chemical species that are representative of the range of line emission morphologies , kinematics , and line excitation temperatures detected in these data , and furthermore do not suffer significant line blending ( see table [ linelist ] for more details ) .
first moment images of these transitions toward the central region of the cluster are shown in figure [ moments ] , where the field of view is the same as figure [ sma1zoom ] .
the parent cube was made with briggs weighting and a robust parameter of 1.0 .
details of the transitions shown are provided in table [ linelist ] , including their rest frequencies , excitation energies , and integrated intensities .
the first striking feature in the images is the velocity gradient across the central 0.8 ( 1000 au ) of sma 1b , which is seen consistently in all transitions .
the orientation and magnitude of the velocity gradient is comparable to that seen at @xmath192 angular resolution @xcite . in the lower energy lines such as hc@xmath5n and ocs , the emission and its accompanying gradient also extend somewhat to the northeast toward sma 1c . a zoomed view of the ch@xmath5cn @xmath41=7 transition is shown in figure [ sma1bd]b .
the magnitude of the gradient ranges from 3 to 5 km s@xmath9/arcsec , with the largest values seen in the lines of ch@xmath5oh and ch@xmath5ocho .
the position angle of the gradient was measured in each transition by viewing the data cube in casa , marking the ra / dec centroid of the emission in the outer velocity channels , and computing the corresponding slope .
the median and standard deviation of the position angle taken over all the transitions is -52@xmath105 .
sma 2 and sma 4 also show compact emission , but in only 9 or 10 of the 12 transitions , respectively .
the differences in detections indicate differences in both chemistry ( hc@xmath5n being detected in sma 4 but not sma 2 ) and excitation conditions ( sma 2 being more extended in the lower energy transitions ) .
also , the central velocity of the emission in sma 1b differs by several km s@xmath9 from the emission seen toward sma 2 and sma 4 , indicating a significant velocity dispersion in this protocluster .
to better quantify the characteristics of the line emission and estimate the gas physical conditions , we extracted the @xmath6=1211 ch@xmath5cn spectra at the peak positions of all the continuum sources in table [ contsources ] .
only 6 sources show detectable emission ; in particular , there is no sign of emission from sma 1c or 1d .
we used the cassis package to perform a local thermodynamic equilibrium ( lte ) model fit to the ch@xmath5cn and ch@xmath48cn ladders , which provide a good measurement of gas temperature in hot cores ( e.g. * ? ? ?
* ; * ? ? ?
there were five free parameters in the model : temperature , velocity , line width , column density , and diameter .
based on the galactocentric distance of ngc 6334 , we assume a @xmath49c:@xmath47c ratio of 58 @xcite .
we used the markov chain monte carlo ( mcmc ) @xmath50 minimization option , which is far more cpu time - efficient and tractable than a uniform grid search . unlike a grid search ,
the mcmc search begins with an initial guess and a range for each parameter , and makes random steps in each parameter in order to explore the range .
the amplitude of the random step size is initially large but decreases during the first stage of the computation as the space is explored . in the latter stage ,
smaller steps are used to refine the best chi - squared potential . in order to establish the parameter space for each source and the initial guess , we ran initial executions with very broad ranges in order to encompass plausible values .
based on these results , we narrowed the ranges somewhat .
the typical search parameters were ranges of @xmath51 k for cool sources and @xmath52 k for warm sources in temperature , @xmath53 km s@xmath9 in linewidth , @xmath54 km s@xmath9 in velocity , and a factor of 10 - 50 in column density ( all centered on the initial guess ) , and a range of 0.1 - 0.5 for angular diameter . with these ranges
, we ran several executions of mcmc each with a different value of the automatic step size reduction factor ( called reducephysicalparam ) , thereby achieving a variety of acceptance rates in the range of 0.2 to 0.4 after 2000 iterations .
the cutoff parameter , which determines when the step size reduction factor becomes fixed , was set to half the number of iterations .
the best fit values reported in table [ temperatures ] are taken from the execution that achieved an optimal acceptance rate of @xmath390.25 . to obtain realistic uncertainties for the parameters ,
we take the standard deviation from this single execution and add ( in quadrature ) half the total spread in the individual best fit model values .
the best - fit model spectra are overlaid on the observed spectra in fig .
[ linefits ] . in the case of sma 4 ,
the line shapes are clearly non - gaussian and a single component fit yields an abnormally high line width of @xmath55 km s@xmath9 . for this source
, we also performed a two component fit , which resulted in smaller , more reasonable line widths from two spatially unresolved components that have rather different temperatures and velocities ( by 6 km s@xmath9 ) .
this result may indicate the presence of an unresolved binary hot core system . for sma 1b
, we also found that two temperature components produced a better fit to the low and high-@xmath41 lines of the spectrum , as is typical of many hot cores ( e.g. * ? ? ?
* ; * ? ? ?
* ) . in this case
, the search range for the cooler component was 60 - 150 k , and for the hot component was 150 - 600 k. the best - fit temperature of the hot component of sma 1b being in excess of 300 k is consistent with the detection of the @xmath41=10 component with @xmath56 k. the final column in table [ temperatures ] provides an order - of - magnitude estimate of the volume density of h@xmath1 that was calculated from the fitted column density assuming spherical geometry and an assumed ch@xmath5cn abundance of @xmath57 .
this value is intermediate between the extrema of recent measurements of the ch@xmath5cn abundance in hot cores , which range from @xmath58 to @xmath59 ( see e.g. * ? ? ?
* ; * ? ? ?
a comparison of the ch@xmath5cn column density to the dust column density is discussed further in [ dustcalc ] .
probable maser emission in the class i 229.75881 ghz ch@xmath60oh ( 8@xmath617@xmath62 ) e transition is a conspicuous feature of recent sma observations of massive protostellar outflows ( e.g. * ? ? ?
* ; * ? ? ?
* and references therein ) . in these outflows
, the 229 ghz emission is generally co - located ( spatially and spectrally ) with lower frequency class i masers . in ngc
6334 i(n ) , we observe strong features in both the 229 ghz transition and the class i 218.44005 ghz ( 4@xmath633@xmath64 ) e transition at the position of the brightest 44 ghz maser ( spot 55 of * ? ? ?
* ) , which also coincides in velocity and position with the @xmath65 km s@xmath9 24.9 ghz ch@xmath5oh maser @xcite .
although interferometric observations of the 229 ghz line have been reported previously ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , our observations are the first with sufficient angular resolution to test unequivocally whether the brightness temperature ( @xmath66 ) exceeds the upper energy level of the transition ( @xmath67 ) .
the peak values of @xmath66 in the 229 ghz and 218 ghz lines ( 3100 k and 270 k , respectively ) are both significantly higher than the @xmath67 of these transitions ( 89 k and 45 k , respectively ) , indicative of maser emission .
strong , point - like emission also appears at several other positions in these lines , but primarily in the 229 ghz line ( figure [ methanolmasers ] ) .
we note that the 229 ghz line is part of the 36 ghz maser series , while the 218 ghz line is part of the 25 ghz maser series @xcite .
all positions with peak intensity greater than 0.55 jy beam@xmath9 ( @xmath68 k ) are listed in table [ masertable ] ; this threshold was chosen because it excludes all of the emission seen toward the hot cores .
seven of the nine components are within 0.2 of a 44 ghz maser @xcite and/or within 0.33 of a 24.9 ghz ch@xmath5oh maser @xcite .
all nine have brightness temperatures greater than the energy level of the transition . in both transitions ,
fainter emission appears toward the continuum sources sma 1 , sma 2 and sma 4 .
this emission is likely to be thermal because the brightness temperatures are below 50 k. finally , there are two candidate maser positions in the 218 ghz line which do not meet our peak intensity threshold , but do have brightness temperatures higher than 50 k and are not associated with a continuum source .
one corresponds in position and velocity to component 4 of the 229 ghz line , and the other lies at 17:20:53.781 , -35:45:13.58 ( j2000 ) with an lsr velocity of @xmath691.3 km s@xmath9 .
nine of the continuum sources are detected at more than one wavelength . to explore the nature of these objects ,
we have constructed the spectral energy distributions ( seds ) in figure [ seds ] .
two of the objects , sma 1b and sma 4 , are detected longward of 1.3 cm , and their seds are well - described by a combination of free - free emission plus dust emission .
the other seven sources all show a steeply rising spectral index in the submillimeter .
note , because the existing 3 mm observations @xcite have more than four times poorer angular resolution than the other wavelengths presented here , it is not possible to separate the contributions to the 3 mm flux density for the individual components of sma 1 .
additionally , the 3 mm flux densities for sma 4 and sma 6 are regarded as upper limits for the size scales considered here . further details on individual sources are described below along with descriptions of the modelling procedure . with the addition of the very sensitive 6 cm data and the improved astrometry at 1.3 cm and 7 mm compared to @xcite , it is now possible and appropriate to do somewhat more sophisticated modeling of the free - free emission from sma 1b and sma 4 , since the millimeter and centimeter emission have been shown to coincide .
the long - wavelength spectral indices are + 0.4 and + 0.7 , respectively .
the possible interpretations of a moderately positive spectral index include jets and winds ( e.g. * ? ? ?
* ; * ? ? ?
* ) and hypercompact hii ( hchii ) regions with steep radial density profiles @xcite . in this case
, we do not see any sign of spatial extension in the centimeter wavelength emission for either of these sources they are both unresolved with sizes @xmath38(@xmath70 au ) .
thermal radio jets from massive protostars have been seen to extend to scales as large as 0.1 pc ( * ? ? ?
* e.g. iras 16547 - 4247 ) .
however , the small angular extent of the radio jet in the low - to - intermediate mass young stellar object hh 111 vla 1 @xcite implies a size of only 116 au , so the jet interpretation can not be ruled out entirely on the basis of size alone .
nevertheless , the prior evidence that these are high - mass protostars @xcite leads us to favor the hchii interpretation . to model hchii region emission ,
we have used the bremsstrahlung emission model v of @xcite in which the electron density profile follows a power - law distribution ( @xmath71 ) transitioning to a sphere of constant density embedded in the center , which avoids a non - physical singularity .
this formulation yields a spectral index of + 0.6 at frequencies below the turnover point , which provides a good match to the seds of known hypercompact hii regions @xcite . to determine the optimal model
, we used the `` lmfit '' python package , which performs non - linear least square minimization using the levenberg - marqardt algorithm .
because the free - free emission and dust emission are comparable at frequencies just above the free - free turnover point , it is best to model both emission mechanisms simultaneously . to model the dust emission
, we use a single - temperature modified greybody function @xcite . with six flux density measurements , we attempted to model five free parameters , including three for the central sphere of the hchii region : electron temperature ( @xmath72 ) , electron density ( @xmath73 ) , and diameter ( @xmath74 ) ; and two for the dust emission : the dust grain opacity index ( @xmath75 ) and the opacity at the reference wavelength of 1.3 mm ( @xmath76 ) . because all of our measurements are on the rayleigh - jeans portion of the dust emission , we are unable to fit for the dust temperature ; therefore , we fixed the dust temperature to the value determined for the gas ( see [ hotcore ] ) , which is a valid assumption at the high densities of these sources .
unfortunately , simultaneous fits to all five parameters are not very constraining due to degeneracies between the parameters .
instead , we defined a 16 by 16 point grid of electron density ( @xmath77 to @xmath78 cm@xmath3 ) and temperature ( 4000 to 14000 k ) , and fit the remaining three parameters at each grid point .
the grid points were uniformly distributed
logarithmically for density and linearly for temperature . for sma 1b , 15% of the results reproduced all the flux density measurements well and had rather similar @xmath50 values . for sma 4 ,
23% of the results were in this category .
the rest of the combinations of parameters were notably deviant at one or more data points .
in figure [ seds ] , we plot the curve corresponding to the median parameter values and report the standard deviation of the five parameters across the good fits . for sma 1b
, the dotted curve corresponds to @xmath80 cm@xmath3 and a diameter of @xmath81 au .
the expected full width at half maximum ( fwhm ) of the total hchii emission ( i.e. including the portion outside the central sphere ) is then 54 au .
the corresponding values for sma 4 are @xmath82 cm@xmath3 and @xmath83 au , yielding a fwhm of 9 au .
both of these sizes are consistent with being unresolved by our beam . therefore
, one explanation for these objects is that they are currently in the gravitationally - trapped phase of hchii evolution @xcite .
if so , then in the spherical accretion flow model , the measured radius of an hchii must be less than the bondi - parker transonic radius , which in turn places a lower limit to the central stellar mass ( see equation 3 of * ? ? ?
using a sound speed of 11 km s@xmath9 for 10000 k gas , the corresponding lower limits for the central stars in sma 1b and sma 4 are 4.4 and 1.4 @xmath11 , respectively .
of course , the presence of ionizing radiation required to form the hchii implies a central stellar mass of @xmath198 @xmath11 or more . evidently , the accretion flow is still sufficiently strong to maintain the ionization radius significantly inward of the point where pressure - driven expansion can begin , and these stars can potentially continue to gain mass .
in addition to a dust component , this source shows excess emission at 1.3 cm , the nature of which remains unclear .
there is no significant compact spectral line emission coincident with this source ; the moment images of hc@xmath5n , ocs and @xmath47cs give the impression that emission avoids it and wraps around it ( figure [ moments ] ) .
more sensitive observations at wavelengths longward of 2 cm are needed to understand the nature of the centimeter continuum emission . to model the dust emission
, we use the 7 mm and 1.3 mm flux densities and the assumed temperature range for the gas ( 20 - 50 k , [ hotcore ] ) . because only two flux densities are available ( as is the case for sma13 and 15 ) , we can not fit for @xmath75 and @xmath76 simultaneously .
instead , we first established a @xmath84 range for @xmath75 by a monte - carlo treatment of the flux densities and their uncertainties .
we then derive the corresponding range of @xmath76 by solving the greybody model at each combination of the @xmath85 and @xmath75 extrema to find the value of @xmath86 that matches the 1.3 mm flux density measurement .
we report the minimum and maximum of the resulting values as the @xmath76 range in the text labels of figure [ seds ] .
sma 1d is perhaps the most intriguing source as it shows a consistent spectral index from 1.3 cm to 0.87 mm . with measurements at four wavelengths available , we used `` lmfit '' to fit the spectral index of + [email protected] .
this slope is only marginally steeper than an optically - thick blackbody , and is similar to the values recently reported in the low - mass class 0 protostar l1527 @xcite and in source vla2 in the high mass star - forming core afgl2591 @xcite . if this emission arises from dust ,
then , as summarized by @xcite , it could be explained by large dust grains ( compared to the wavelength ) , non - isothermal conditions , or optically - thick structure , in which case it must be very compact .
a similar spectral index is seen in the source ngc 6334 i - sma4 @xcite , suggesting that it arises from a not uncommon phase of evolution of massive star formation .
an alternative to dust is the core of an optically thick jet @xcite , in which the outer parts of the jet are too faint to detect at the current sensitivity level .
again , more sensitive and higher resolution observations are needed to explore these possibilities .
the rest of the continuum sources that are detected at 1.3 mm and 0.87 mm but not at longer wavelengths almost certainly arise from dust emission . if they were free - free emission , the allowed range of spectral index ( -0.1 for optically - thin emission to @xmath87 for optically - thick emission )
would make them easily detectable in the vla images . as described in
[ missingflux ] , our 0.87 mm measurements of sma 13 and sma 15 do not suffer from missing flux , and their steep spectral indices from 1.3 mm to 0.87 mm are consistent with thermal dust .
however , we must consider the chances of these sources being extragalactic background objects dominated by dust emission . the south pole telescope ( spt - sz ) survey of 771 square degrees in three frequency bands provides a good reference point for comparison @xcite . from their cumulative distribution plot ,
the number of spt detections with dust - like spectral energy distributions that are brighter than 25 mjy at 220 ghz is only 0.25 - 0.40 per square degree .
the probability of encountering such a source in the field of view of figure [ fig1 ] is only @xmath88 .
thus , it is safe to conclude that these are not background objects , but are members of the protocluster . in the case of sma 2 and sma 5 , although the spectral indices are consistent with dust emission , we can only place lower limits on @xmath75 : 1.0 and 0.5 , respectively ( based on the upper limits at 1.3 cm and the lower limits at 0.87 mm ) . regarding the two new 6 cm point sources ( vla 2 and h@xmath1o - c4 ) that are not detected in millimeter continuum , the fact that one of them ( h@xmath1o - c4 ) coincides with a h@xmath1o maser is compelling evidence that it is tracing a protostellar object , due to the strong association of molecular outflows with h@xmath1o masers @xcite .
in fact , it is the fourth brightest of the 11 h@xmath1o masers in the region , with an isotropic luminosity of @xmath89 @xmath23 , which is characteristic of massive young stellar objects ( mysos ) ( e.g. * ? ? ? * ) and the high end of intermediate - mass ysos ( e.g. * ? ? ? * ) . in the @xcite red msx source ( rms )
survey , which includes over 300 h@xmath1o maser measurements , the bolometric luminosities of mysos with this maser luminosity range from @xmath90 @xmath23 , suggesting that h@xmath1o - c4 is a _ massive _ protostellar object . the other new source ( vla2 ) has no maser counterpart ,
so we must assess the possibility that it is a background source .
using the formula of @xcite for the expected number of background sources at 6 cm as a function of field size , we calculate a 4.4% chance of finding a background object of @xmath91 mjy in the field of view in figure [ fig1 ] . alternatively , using the source counts from more recent evla 10 cm observations and extrapolating their flux densities to 6 cm with the typical spectral index of @xmath92 @xcite , we find an expected source density of @xmath93 sr@xmath9 .
multiplying by the solid angle of figure [ fig1 ] , we calculate a 6.3% chance of finding a source in the range of 0.05 - 0.15 mjy .
thus , the most likely scenario is that vla2 resides in the protocluster and may trace some form of young stellar object .
given the lack of millimeter continuum and maser emission , the centimeter emission could also arise from a low - mass , pre - main sequence ( pms ) star in which any associated circumstellar dust emission is too faint to be detected .
such objects have been seen , for example in the cepheus a east star - forming region @xcite at a distance of 700 pc @xcite .
the variable sources ( hw 3a , 8 , 9 ) emit 0.153.0 mjy at 6 cm and show no point source counterparts in sma 0.87 mm images @xcite . at the distance of ngc 6334
i(n ) , the 6 cm flux densities of these objects would be 0.040.9 mjy , a range that includes both vla2 and h@xmath1o - c4 .
it is interesting to note that hw 8 and hw 9 are both x - ray sources with hw 8 postulated to be a pre - main sequence star ( due to its high median photon energy ) , while the softer , brighter , and more variable hw 9 is consistent with activity on a b - type star @xcite . in ngc
6334 i(n ) , a recent _ chandra _ x - ray survey finds 7 sources in the field of figure [ fig1 ] , but none of the detections correspond in position to within 1.5 of any of the millimeter or centimeter sources , and their 1@xmath94 x - ray position uncertainties are @xmath95@xcite .
the unprecedented multiplicity of sources in this field allows us to quantitatively analyze the structure of a young massive protocluster for the first time .
first , we define the membership of the protocluster by requiring either the detection of compact millimeter continuum emission or the positional coincidence of at least two star formation indicators . starting with the source list of table [ contsources ] , we exclude sma 7 as there is no compact component detected . of the two new 6 cm sources
, we can include only the h@xmath1o maser source h@xmath1o - c4 .
vla 2 could be protostellar , but lacking a second positive indicator we exclude it .
finally , although we have presented strong evidence that sma 1b and 1d are independent sources , they may form a binary system . our sensitivity to binaries
is clearly limited by our angular resolution , so we choose to count these two objects as a single source in order to avoid biasing the following statistics by double counting some binaries but not others .
the final tally of sources is then @xmath96 .
the area of the protocluster on the sky is limited by the 1.3 mm primary beam .
the largest angular radius from the phase center of a detected source at 1.3 mm is @xmath97 , where @xmath98 mjy beam@xmath9 .
this implies that our sample of the 1.3 mm compact sources in this region is complete down to 25 mjy across a physical radius of 0.21 pc corresponding to a projected area of 0.14 pc@xmath99 and a spherical volume of 0.038 pc@xmath100 .
the presence of 25 sources within this volume yields an average number density of 660 pc@xmath3 .
we note that only 20 of the sources are above the 25 mjy completeness threshold ; therefore , this computed density is strictly a lower limit even if the underlying population does not extend beyond the faintest detection .
in other words , due to the radial decrease in sensitivity , the faintest sources detected in the central portion of the field would have been missed in the outer portions of the field . following the techniques applied to infrared and optical observations of clusters of stars and hii regions ( e.g. * ? ? ?
* ; * ? ? ?
* ) , we have constructed the minimal spanning tree ( mst ) formed from these 25 sources , as shown in figure [ mst ] .
the minimum spanning tree for a set of points is defined as the set of edges connecting them that possesses the smallest sum of edge lengths . to compute it , we used a python function based on the algorithm of @xcite . from this result
, we compute the q - parameter , which is defined as the ratio of @xmath101 , the normalized mean edge length of the mst , to @xmath102 , the correlation length , which is defined as the mean projected separation between sources normalized by the cluster radius , @xmath103 @xcite .
@xmath104 is defined as the distance from the mean position of all the stars to the furthest star from that point . for this cluster ,
we obtain @xmath105 and : @xmath106 interestingly , this value ( 0.82 ) is close to the critical value that separates clusters between the regime of multiscale ( fractal ) substructure ( @xmath107 ) typified by the taurus and chamaeleon clusters , and a centrally concentrated configuration typified by @xmath108 oph and ic348 @xcite .
it is also close to the value of 0.84 obtained for stars in the orion trapezium cluster @xcite . in the smoothed particle hydrodynamics simulation of the formation of a 1000 @xmath11 cluster @xcite
, the @xmath4-parameter of the resulting stars evolves steadily from initial values of @xmath190.5 to values greater than 1.1 @xcite .
it passes the critical 0.8 value after about 1.8 free - fall times .
these results suggest that our measured value for @xmath4 for this protocluster is not a fixed property but may instead provide a measure of its age .
however , we caution that our census of the cluster may be limited by the primary beam of the observations , i.e. the cluster may extend beyond this region in one or more directions .
it is also limited in terms of sensitivity depth .
future more sensitive observations with a wider field of view are needed to improve the robustness of this result . in any case , it is likely that this protocluster exists in a stage when its ultimate structure is still to be determined . using the virial theorem ,
the dynamical mass of a stellar cluster can be estimated from the mean square velocity of its members ( relative to their mean velocity ) as originally performed on the globular cluster m92 by @xcite .
for the ngc 6334 i(n ) protocluster , we compute the unbiased sample variance of the source - to - source lsr velocities : @xmath109 , by using the fitted velocities ( @xmath110 ) from the single - component ch@xmath5cn spectral models of the @xmath111 sources in table [ temperatures ] and their mean ( @xmath112 ) . as seen in figure [ mst ] , three of these sources lie in the central strip , two in the first pair of flanking strips , and one in the second pair of flanking strips .
thus , although small in number , these six sources provide a fair spatial sampling across the cluster .
the h@xmath1o maser emission provides a potential kinematic measurement for two additional sources : sma 20 and h@xmath1o - c4 .
however , h@xmath1o masers typically span a broad range in velocity , and the centroid can vary widely from the lsr of the thermal gas @xcite , as is the case for sma 18 , which is detected in both ch@xmath5cn and h@xmath1o .
therefore we have chosen not to include these two sources in this analysis .
the resulting value and uncertainty of the unbiased sample variance are @xmath113 km@xmath114 s@xmath115 , where the uncertainty in the variance is the variance times @xmath116 ( e.g. * ? ? ?
assuming a random orientation of space velocities , the mean square three dimensional velocity is then @xmath117 km@xmath114 s@xmath115 .
the proper effective distance to use in the virial theorem , @xmath118 , can be determined by numerical integration of the observed strip counts @xcite . in this case , the strip counts are the number of stars in each horizontal bin shown in figure [ mst ] .
we chose a bin width of 10 as it provides seven bins , which is a good compromise between having too few bins and having too few sources per bin .
using equation 1 of @xcite , we obtain @xmath119 , or 0.28 pc , yielding a dynamical mass estimate of @xmath120 @xmath11 . by comparison ,
the gas mass of ngc 6334 i(n ) plus sm2 based on single - dish submillimeter continuum observations is @xmath121 @xmath11 , assuming a uniform dust temperature of 30 k @xcite , after scaling to the new distance of 1.3 kpc .
because the @xcite mass is based on gaussian fits to these two sources , it is a lower limit to the total gas mass as it does not include the mass of the more extended material between them ( see figs .
[ fig0 ] & [ fig1 ] ) .
furthermore , the total gas mass is a lower limit to the total cluster mass because it does not include the mass of condensed ( proto)stars . considering these effects ,
the agreement between these two measurements is remarkable , and lends credence to the interpretation of the hot core velocity dispersion as tracing the dynamics of the protocluster .
we note that the velocity dispersion @xmath122 km s@xmath9 is in good agreement with the internal radial velocity dispersion of the scorpius ob2 cluster ( 1.0 - 1.5 km s@xmath9 ) as measured by kinematic modelling of hipparcos data @xcite .
we next consider the relaxation time of the protocluster , which is based on the crossing time : @xmath123 because the number of known protostars is still fairly small , the relaxation time ( see e.g. * ? ? ?
* ) is essentially the same : @xmath124 given the likelihood that the protostellar population continues below our sensitivity limit , this value should be considered an upper limit to the relaxation time .
the low value of the @xmath4-parameter suggests that the protocluster is dynamically young in the context of evolutionary simulations @xcite .
however , the good agreement between the dynamical mass and the gas mass suggests that the protocluster has already persisted for one or more crossing times , which theory suggests is sufficient for dynamical mass segregation to occur @xcite .
although the number of detected objects is too small to apply a statistical estimate of mass segregation @xcite , the probable most massive source in the protocluster ( sma 1b ) is by no measure located near the center of the distribution of members ( see figure [ fig1 ] ) . therefore , we do not see any clear evidence for mass segregation .
our single - component lte model fits ( table [ temperatures ] ) have provided gas temperature and column density measurements for six of the sources .
given the high volume densities implied by these fits , we make the usual assumption that the dust temperature ( @xmath125 ) is well coupled to the gas temperature through collisions . to obtain gas mass estimates from the 1.3 mm dust emission ,
we follow the procedure of @xcite using the dust mass opacity coefficient @xmath126 cm@xmath99 g@xmath9 appropriate for grains with ice mantles in regions of high gas density ( 10@xmath128 cm@xmath3 ) @xcite , and a gas to dust mass ratio of 100 . in the sources that do not show ch@xmath5cn emission
, we have no temperature measurement . in these cases
, we have assumed a range of temperatures that are fairly low ( 2050 k ) , and compute the range of masses implied by the measured flux density .
this temperature range encompasses both the typical temperature of gas surrounding young protostars prior to the hot core phase , as well as the expected mean temperature of gas in circumstellar disks around stars of type a3 b5 @xcite .
the masses were corrected for the dust opacity by comparing the observed brightness temperature to the assumed dust temperature ( @xmath129 ) and the correction factor is : @xmath130 .
the resulting mass estimates for all sources are shown in table [ dustmass ] , and for a subset of sources in figure [ seds ] . with the exception of sma 1a , b , c , d ,
sma 3 , and sma 9 , the mass estimates are in the range of 0.2 - 2 @xmath11 .
these masses cover a range similar to the masses of dense cores in the perseus molecular cloud @xcite ; however , the smallest perseus objects have a typical diameter @xmath395000 au ( 20 at 250 pc ) in submillimeter continuum images @xcite .
such objects would be nearly 4 in diameter at the distance of ngc 6334 i(n ) .
therefore , the emission we are detecting must arise from structures that are an order of magnitude ( or more ) smaller , which includes circumstellar disks .
the mass values are not as high as massive disk candidates ( e.g. the 0.8 - 8@xmath11 of iras 20126 + 4104 , * ? ? ?
* ) , but they are an order of magnitude higher than the typical masses of circumstellar disks around low - mass stars , for example 0.005 - 0.14 @xmath11 in the ophiuchus sample of @xcite . however , they are not too massive to be disks around intermediate mass stars .
there are a number of examples of herbig a stars with disk masses of @xmath190.2@xmath11 , such as mac ch12 @xcite .
its 1.3 mm flux density of 44 mjy @xcite would be 19 mjy at the distance of ngc 6334 i(n ) .
a more massive example is sma1 in g5.89 - 0.39 , which appears to be an intermediate mass protostar with @xmath191 @xmath11 of circumstellar material @xcite .
thus , along with the lower mass population of disks around high - mass protostars ( such as the 0.8 @xmath11 of afgl2591 vla3 ; * ? ? ?
* ) , we may also be detecting the most massive tip of the population of disks around intermediate - mass protostars in this protocluster . finally , in principle one can use the h@xmath1 gas column density estimated from the dust column density to calculate the ch@xmath5cn abundance , subject to the assumption that both tracers arise from the same volume of gas . in the case of sma 1b ,
the fitted size of the dust emission ( 0.87 at 1.3 mm and 0.47 at 0.87 mm ) is significantly larger than the modeled size of the ch@xmath5cn emission region ( 0.38 ) , so this assumption is not valid .
nevertheless , proceeding with this uncertainty , the gas column density toward sma 1b inferred from the dust emission is @xmath131 cm@xmath115 , yielding an abundance estimate of @xmath133 .
similarly , for the next two brightest line sources ( sma 2 and sma 4 ) , we obtain abundances of @xmath134 and @xmath135 . given all the uncertainties , these values are in reasonable agreement with the assumed value of @xmath136 ( see [ hotcore ] ) .
higher resolution observations are needed to obtain more accurate estimates of the molecular abundances via this method .
the position angle of the velocity gradient in the molecular gas in sma 1b ( @xmath1375 ) is equal to the position angle of the two - dimensional gaussian model fit to the dust emission ( figure [ sma1bd ] ) : @xmath1379at 1.3 mm ( table [ contsources ] ) and @xmath1388at 0.87 mm ( table [ contsourcesband7 ] ) .
furthermore , this position angle is 91 different from the position angle of @xmath139 determined for the sio 5 - 4 outflow ( * ? ? ?
* also see figure [ methanolmasers ] ) .
this perpendicularity is heavily suggestive of a disk - outflow system . in a few of the lower temperature transitions , including hc@xmath5n , ocs , and @xmath47cs
, emission extends beyond sma 1b by up to an arcsecond along the outflow direction on one or both sides .
this behavior is not expected from a compact disk . however , at these locations , the kinematic structure changes from a strong gradient to a narrower range of velocities , suggesting that these transitions are tracing an interaction of the outflow with ambient material near the disk . to further explore the velocity gradient toward sma 1b ,
we constructed position - velocity ( pv ) images ( figure [ pv ] ) for the same 12 transitions whose moment maps were shown in figure [ moments ] ( the position and orientation of the slice is shown on the ch@xmath5ocho panel of figure [ moments ] ) . in the context of a disk
, three main factors will govern the appearance of the pv image of a molecular line : ( 1 ) the overall physical structure of the circum - protostellar material , in particular the outer radius , inner radius ( if there is a central cavity ) , and the thickness ; ( 2 ) the radial gradient in the temperature , density , and molecular abundance ; and ( 3 ) the observer s viewing angle .
evidence for chemical segregation in the context of a hot core was recently reported for the inner 3000 au of the afgl2591 vla3 , in which different species appear to trace different radii with respect to the continuum peak @xcite .
one interpretation of this object is that it contains a massive disk @xcite . in the case of ngc 6334
i(n ) sma 1b , we observe the overall shapes of the structures in the pv images to be generally consistent with one another , but the differences are worth noting . in general , the high temperature transitions are peaked toward the central source , including the two highest temperature ch@xmath5oh lines , ch@xmath5cn k=7 , and ch@xmath5ch@xmath1cn .
in contrast , most of the lower temperature transitions show a double peak with a local minimum toward the central source , including @xmath47cs , ocs , ch@xmath5ocho and ch@xmath5cn k=3 .
this pattern is consistent with the combination of a radial temperature gradient in which the hotter gas lies closer to the central heating source , and a viewing angle that is edge - on ( or at least moderately so ) .
the two species that deviate from this picture are hnco , which shows a distinctive compact morphology , and hc@xmath5n , which extends to higher velocities .
regarding hnco , @xcite recently modelled its abundance in hot cores versus time using a wide range of assumed chemical reactions ( both grain and gas - phase ) , along with different initial cloud collapse temperatures and post - collapse ( hot core ) temperatures .
they find the best agreement with observed abundances from models that include both grain and gas - phase chemistry , cloud collapse temperatures of 10 k , and post - collapse temperatures @xmath140 k. furthermore , the late peak of hnco abundance with time ( @xmath141 yr ) suggests that hnco is not itself a `` first generation '' species liberated from dust grains . instead , it is the ejection of several larger hnco `` daughter '' molecules formed on the grains ( such as hncoho , hncocho , hnconh , and hncooh ) and their subsequent destruction that form hnco in the gas phase .
the compact size of the observed hnco emission relative to other species suggests that this formation scenario is efficiently proceeding only in the inner disk , perhaps due to the higher temperatures there .
regarding hc@xmath5n , not only does its emission extend to higher velocities , but these velocities arise exclusively from locations close to the center .
this pattern suggests faster rotation toward the inner radii .
however , it is somewhat surprising that hc@xmath5n is seen closer to the central source than other molecules given that it has a relatively high photodissociation rate @xcite and that ionizing radiation should be stronger at smaller radii . on the other hand , if there are locations with sufficient shielding , such as in the mid - plane of a high column density disk , the formation of hc@xmath5n can proceed quickly as a product of c@xmath1h@xmath1 and cn ( e.g. * ? ? ?
* ; * ? ? ?
* ) , and cn has a low photodissociation rate @xcite .
it is interesting to note that the hc@xmath5n abundance observed toward the continuum peak of afgl2591 vla3 by @xcite exceeds their gas - grain chemical evolution model predictions by more than two orders of magnitude .
indeed , these authors emphasize that molecular abundances are very sensitive to extinction , and hence the detailed geometry of the system .
as seen in figure [ pv ] , the maximum extent of the emission toward sma 1b is @xmath142 ( @xmath10800 au ) and the observed range of gas velocities at that radius is approximately @xmath143 km s@xmath9 with respect to the lsr . if this emission is interpreted as arising from a circular rotating disk , then these two quantities alone can be used to make a simple dynamical estimate of the central enclosed mass using kepler s law for circular rotation : @xmath144 @xmath11 , where @xmath145 is the angle of inclination of the rotation axis to the line of sight . however , rotation is not the only possible motion , as evidence for infall in this source has already been demonstrated by the trend of the increasingly redshifted absorption features with increasing line excitation seen in three transitions of cn and h@xmath1co @xcite .
therefore , to build a more illustrative physical model , we follow the approach of equation 1 of @xcite , where we interpret the pv images as arising from a rotating keplerian disk undergoing free - fall and having a specific inner and outer radius .
using this model , the white contour in figure [ pv ] encompasses the entire region where emission can arise ( ignoring any effects of line opacity and dust opacity ) from a disk with an enclosed mass of @xmath146 @xmath11 , and the specified inner and outer radii for each transition .
( for comparison , the model without free - fall is included as the black contour in the first panel of figure [ pv ] . )
the inner radius sets the total extent of the white contour along the velocity axis , while the outer radius sets its total extent along the position axis .
most of the molecular transitions are consistent with a model having an inner radius of 500 au , and outer radius of 800 au .
the hc@xmath5n line is more consistent with a smaller inner radius of @xmath19200 au , while the hnco line is more consistent with originating from a narrow range of inner radii ( 400 - 500 au ) .
the outer radius of 800 au seen in most species is comparable to the semi - major axis of the deconvolved gaussian model of the continuum from sma 1b , which is 660@xmath1033 au at 1.3 mm ( table [ contsources ] ) and 403@xmath1020 au at 0.87 mm ( table [ contsourcesband7 ] ) .
for example , an outer radius of 800 au for the molecular gas would correspond to the 7@xmath102 percent point of the 0.87 mm model dust source .
thus , the extent of the dust and gas emission yield similar values for the size of the structure . assuming circular symmetry , the observed ellipticity of the 0.87 mm deconvolved model can then be used to set a lower limit for @xmath145 of 55 , i.e. nearly edge - on .
the corresponding upper limit for the enclosed mass is then 30 @xmath11 .
we note that for a disk viewed edge - on , the line emission from the innermost radii will be obscured by the high optical depth of the foreground material in the disk , both from the gas and ( at some point ) the dust grains . to demonstrate , the ch@xmath5cn optical depths at line center for the @xmath41=7 and 8 components , derived from their peak brightness temperatures in the single - temperature cassis model , are 0.7 and 0.4 , respectively .
although the presence of a keplerian disk is tantalizing , future higher resolution observations are necessary to properly model the geometry and inclination angle and to confirm the expected kinematic structure in more detail .
radiative transfer modelling will no doubt be profitable in delineating and interpreting the chemical structure as it has been for low - mass protostellar disks @xcite .
using new high - resolution sma and vla images from 6 cm to 0.87 mm , we have found further multiplicity in ngc 6334 i(n ) , allowing us to perform the first structural analysis of a massive protocluster using techniques developed for optical / infrared studies of star clusters .
we also demonstrate the first use of the thermal gas velocities from an ensemble of hot cores to probe the dynamical properties of a protocluster .
our results are summarized as follows : * we have identified 16 new compact 1.3 mm continuum sources and two new 6 cm sources .
three of the newly discovered sources are associated with h@xmath1o masers .
combined with the previously - known 1.3 mm sources from @xcite , the total number of compact centimeter or millimeter sources is 28 . limiting our analysis to likely protostars , i.e. 25 sources ( see [ clusterstructure ] ) , we measure a protostellar density of @xmath2 pc@xmath3 and a minimum spanning tree @xmath4-parameter of 0.82 .
although our measurement of the @xmath4-parameter is likely limited in terms of sensitivity and extent , the value is close to the expected value for a uniform volume density of sources .
* all nine sources detected at more than one continuum wavelength have seds indicative of dust emission . the long wavelength emission toward sma 1b and sma 4
is well modeled by the additional presence of a hchii region .
thermal molecular line emission is detected towards six of the 1.3 mm continuum sources ( sma1b , sma2 , sma4 , sma6 , sma15 , and sma18 ) . from lte modeling of ch@xmath5cn ( @xmath6=12 - 11 ) using cassis we find gas temperatures ranging from 95 - 373 k , ch@xmath5cn column densities from ( 4 - 40)@xmath147 cm@xmath115 , and h@xmath1 gas densities of ( 0.8 - 9)@xmath148 cm@xmath3 ( assuming a ch@xmath5cn : h@xmath149 abundance of @xmath57 ) .
the radial velocities of the hot cores , measured from ch@xmath5cn , range from @xmath7 to @xmath8 km s@xmath9 , and the corresponding 1d velocity dispersion of 1.43 km s@xmath9 implies a dynamical mass of @xmath150 @xmath11 .
this mass is compatible with the gas mass of @xmath151 @xmath11 based on single dish imaging , and demonstrates that hot core line emission can be an important probe of protocluster dynamics .
* the dominant hot core sma 1b shows a consistent spatial - velocity structure in a wide range of hot core molecular lines that is consistent with a disk undergoing keplerian rotation and free - fall .
the orientation of the disk is in excellent agreement with the major axis of the dust continuum emission , and is perpendicular to the large scale outflow axis .
the outer radius of the disk is @xmath19800 au , and the enclosed mass is @xmath1910 - 30 @xmath11 ( depending on the inclination angle ) .
the radial distribution of hnco and hc@xmath5n appears to differ from the rest of the molecules .
* nine positions in the protocluster exhibit 229.7588 ghz class i methanol maser emission , generally in close proximity to previously - identified 44 ghz or 24.9 ghz class i masers .
the brightest position also exhibits 218.4400 ghz class i methanol maser emission .
these are the first observations with sufficient angular resolution to directly establish maser activity in these transitions ( by demonstrating brightness temperatures in excess of their excitation energies ) . based on analysis carried out with the cdms and jpl spectroscopic databases and splatalogue.net .
this research has made use of nasa s astrophysics data system bibliographic services .
this research made use of astropy , a community - developed core python package for astronomy ( http://www.astropy.org ; * ? ? ? * ) .
we thank fred schwab for assistance in interpreting the equations of @xcite .
we thank ed fomalont for his independent astrometric analysis of the alma commissioning data from project 2010.2.99001.csv described in jira ticket csv-2909 .
we thank r. friesen and r. indebetouw for providing useful comments on the manuscript .
we thank the anonymous referee for a thorough report .
alma is a partnership of eso ( representing its member states ) , nsf ( usa ) and nins ( japan ) , together with nrc ( canada ) and nsc and asiaa ( taiwan ) , in cooperation with the republic of chile .
the joint alma observatory is operated by eso , aui / nrao and naoj .
we thank h. mller for pointing out an error in the quantum numbers listed for the 218 ghz maser transition , which we corrected after publication .
lccc observing date & 23 feb 2010 & 18 feb 2010 & 07 jul 2011 + on - source time & 240 minutes & 210 minutes & 84 minutes + project code & 2009b - s036 & 2009b - s036 & 10c-186 + antennas & 7 & 6 & 27 + configuration & very extended ( vex ) & very extended ( vex ) & a + projected baseline lengths & 28 m
508 m & 26.5 m
508 m & 0.53 km 36.63 km + j2000 phase center & 17:20:55.0 , -35:45:07.0 & 17:20:55.0 , -35:45:07.0 & 17:20:53.33 , -35:46:00.0 + primary beam fwhm & 53 & 34 & 9.5 , 6.8 + synthesized beam & [email protected] ( + 5 ) & [email protected] ( + 9 ) & [email protected] ( -2 ) + synthesized beam ( ext+vex ) & [email protected] ( + 5 ) & ... & ... + synthesized beam ( vex - uv ) & [email protected] ( + 5 ) & [email protected] ( + 5 ) & ... + lower band center & 218.85 ghz & 335.54 ghz & 5.06 ghz + upper band center & 230.83 ghz & 347.54 ghz & 7.16 ghz + bandwidth & 4 @xmath152 1.968 ghz & 4 @xmath152 1.968 ghz & 2 @xmath152 1.024 ghz + subbands & 4 @xmath152 24 & 4 @xmath152 24 & 2 @xmath152 8 + polarization & single linear & single linear & dual circular + channel spacing & 1.1 km s@xmath9 & 0.7 km s@xmath9 & 60 km s@xmath9 , 43 km s@xmath9 + continuum rms noise ( ext+vex ) & 2.2 mjy beam@xmath9 & 8.7 mjy beam@xmath9 & 18 , 27 @xmath153jy beam@xmath9 + spectral line rms noise & 30 mjy beam@xmath9 & 130 mjy beam@xmath9 & ... + gain calibrator(s ) & j1733 - 1304 , j1924 - 2914 & j1733 - 1304 , j1924 - 2914 & j1717 - 3342 + bandpass calibrator & 3c273 & 3c273 & j1924 - 2914 + flux calibrator(s ) & titan , ceres & ceres & j1331 + 3030 + 225 ghz zenith opacity & 0.07 & 0.03 - 0.06 & n / a + median wind speed & 2 m / s & 5 m / s & 3 m / s + lcccccccc sma 1a & 17:20:55.137 & -35:45:05.76 & 0.0866 @xmath100.0022 & 39 & [email protected] & [email protected] @xmath152 [email protected] ( + 35@xmath105 ) & 1300@xmath152680 & 10@xmath102 + sma 1b+d & 17:20:55.192 & -35:45:03.93 & 0.3257 @xmath100.0022 & 148 & [email protected] & [email protected] @xmath152 [email protected] ( + 128@xmath109 ) & 1300@xmath152960 & 32@xmath104 + sma 1c & 17:20:55.267 & -35:45:02.89 & 0.1384 @xmath100.0022 & 63 & [email protected] & [email protected] @xmath152 [email protected] ( + 141@xmath1025 ) & 960@xmath152790 & 17@xmath103 + sma 2 & 17:20:54.870 & -35:45:06.40 & 0.1039
@xmath100.0022 & 47 & [email protected] & [email protected] @xmath152 [email protected] ( + 138@xmath106 ) & 750@xmath152350 & 24@xmath106 + sma 3 & 17:20:55.003 & -35:45:07.40 & 0.0331 @xmath100.0022 & 15 & [email protected] & [email protected] @xmath152 [email protected] ( + 10@xmath1025 ) & 1400@xmath1521200 & [email protected] + sma 4 & 17:20:54.627 & -35:45:08.73 & 0.0425 @xmath100.0022 & 19 & [email protected] & @xmath154 & @xmath155 & @xmath156 + sma 5 & 17:20:55.043 & -35:45:01.53 & 0.0433 @xmath100.0022 & 20 & [email protected] & [email protected] @xmath152 [email protected] ( + 6@xmath101 ) & 890@xmath152660 & [email protected] + sma 6 & 17:20:54.590 & -35:45:17.40 & 0.1812
@xmath100.0022 & 82 & [email protected] & @xmath154 & @xmath155 & @xmath157 + sma 7 & undetected & ... & ... & ... & ... & ... & ... + sma 8 & 17:20:53.782 & -35:44:46.16 & 0.0297 @xmath100.0034 & 8.7 & [email protected] & @xmath154 & @xmath155 & @xmath158 + sma 9 & 17:20:54.196 & -35:45:41.45 & 0.1070 @xmath100.0054 & 20 & [email protected] & @xmath154 & @xmath155 & @xmath159 + sma 10 & 17:20:54.283 & -35:45:09.38 & 0.0115 @xmath100.0022 & 5.2 & [email protected] & @xmath154 & @xmath155 & @xmath160 + sma 11 & 17:20:54.751 & -35:45:20.18 & 0.0197 @xmath100.0022 & 9.0 & [email protected] & [email protected] @xmath152 [email protected] ( + 76@xmath1025 ) & 940@xmath152350 & [email protected] + sma 12 & 17:20:54.849 & -35:45:33.51 & 0.0312 @xmath100.0034 & 9.2 & [email protected] & @xmath154 & @xmath155 & @xmath161 + sma 13 & 17:20:54.900 & -35:45:16.41 & 0.0425 @xmath100.0022 & 19 & [email protected] & @xmath154 & @xmath155 & @xmath158 + sma 14 & 17:20:54.962 & -35:45:08.92 & 0.0130 @xmath100.0022 & 5.9 & [email protected] & [email protected] @xmath152 [email protected] ( + 104@xmath1025 ) & 640@xmath152510 & [email protected] + sma 15 & 17:20:55.504 & -35:45:10.96 & 0.0469 @xmath100.0022 & 21 & [email protected] & @xmath154 & @xmath155 & @xmath162 + sma 16 & 17:20:55.503 & -35:44:55.90 & 0.0179 @xmath100.0022 & 8.1 & [email protected] & @xmath154 & @xmath155 & @xmath163 + sma 17 & 17:20:55.590 & -35:44:56.89 & 0.0145 @xmath100.0022 & 6.6 & [email protected] & @xmath154 & @xmath155 & @xmath164 + sma 18 & 17:20:55.645 & -35:45:32.61 & 0.0433 @xmath100.0034 & 13 & [email protected] & @xmath154 & @xmath155 & @xmath165 + sma 19 & 17:20:55.854 & -35:45:26.98 & 0.0179 @xmath100.0025 & 7.2 & [email protected] & @xmath154 & @xmath155 & @xmath166 + sma 20 & 17:20:56.061 & -35:45:32.66 & 0.0320 @xmath100.0034 & 9.4 & [email protected] & @xmath154 & @xmath155 & @xmath167 + sma 21 & 17:20:56.298 & -35:45:27.04 & 0.0362 @xmath100.0034 & 11 & [email protected] & @xmath154 & @xmath155 & @xmath168 + sma 22 & 17:20:56.571 & -35:45:17.01 & 0.0263 @xmath100.0025 & 11 & [email protected] & [email protected] @xmath152 [email protected] ( + 2@xmath104 ) & 750@xmath152400 & [email protected] + sma 23 & 17:20:56.629 & -35:45:09.84 & 0.0167 @xmath100.0025 & 6.7 & [email protected] & @xmath154 & @xmath155 & @xmath169 + lcccc & & + source & @xmath170 ( @xmath171 ) & @xmath172 ( @xmath173 ) & 5.06 ghz & 7.16 ghz + sma 1b & 17 20 55.192 & -35 45 03.83 & [email protected] & [email protected] + sma 4 & 17 20 54.619 & -35 45 08.57 & [email protected] & [email protected] + h@xmath1o - c4 & 17 20 54.149 & -35 45 13.70 & [email protected] & [email protected] + vla 2 & 17 20 54.871 & -35 45 36.49 & [email protected] & [email protected] + lcc sma 1b & 17 20 55.185 & -35 45 03.98 + sma 1c & 17 20 55.259 & -35 45 02.88 + sma 1d & 17 20 55.222 & -35 45 04.08 + sma 4 & 17 20 54.616 & -35 45 08.66 + sma 1a & 17 20 55.150 & -35 45 05.71 + sma 1b & 17 20 55.190 & -35 45 03.95 + sma 1c & 17 20 55.255 & -35 45 02.91 + sma 1d & 17 20 55.223 & -35 45 04.12 +
vla 3 & 17 20 54.992 & -35 45 06.92 + sma 4 & 17 20 54.618 & -35 45 08.71 + sma 6 & 17 20 54.591 & -35 45 17.39 + lccccccc sma 1c & 17:20:55.269 & -35:45:02.88 & [email protected] & [email protected] & [email protected] @xmath152 [email protected] ( + 113@xmath1010 ) & 820@xmath152550 & 24@xmath103 + sma
1b+d & 17:20:55.194 & -35:45:03.98 & [email protected] & [email protected] & [email protected] @xmath152 [email protected] ( + 129@xmath108 ) & 810@xmath152470 & 76@xmath1014 + sma 2 & 17:20:54.871 & -35:45:06.43 & [email protected] & [email protected] & [email protected] @xmath152 [email protected] ( + 164@xmath102 ) & 830@xmath152560 & [email protected] + sma 4 & 17:20:54.626 & -35:45:08.73 & [email protected] & [email protected] & [email protected] @xmath152 [email protected] ( + 178@xmath102 ) & 850@xmath152490 & [email protected] + sma 5 & 17:20:55.048 & -35:45:01.42 & [email protected] & [email protected] & @xmath154 & @xmath155 & @xmath161 + sma 6 & 17:20:54.595 & -35:45:17.40 & [email protected] & [email protected] & @xmath154 & @xmath155 & @xmath174
+ sma 13 & 17:20:54.883 & -35:45:16.32 & [email protected] & [email protected] & @xmath154 & @xmath155 & @xmath175 + sma 15 & 17:20:55.511 & -35:45:11.07 & [email protected] & [email protected] & @xmath154 & @xmath155 & @xmath176 + lcccccccccc hc@xmath5n & j=2423 & 218.32479 & 120.5 & jpl & 7.49 @xmath10 0.22 & @xmath1770.65 & 2.48 @xmath10 0.22 & 1.52 @xmath10 0.38 & @xmath1770.74 & 2.90 @xmath10 0.41 + ocs & j=1817 & 218.90336 & 89.3 & cdms & 4.71 @xmath10 0.21 & 1.38 @xmath10 0.21 & 2.64 @xmath10 0.21 & 2.03 @xmath10 0.28 & 0.6 @xmath10 0.2 & 2.36 @xmath10 0.36 + @xmath47cs & j=54 & 231.22069 & 22.2 & cdms & 2.06 @xmath10 0.30 & 1.22 @xmath10 0.30 & 2.06 @xmath10 0.30 & 1.02 @xmath10 0.29 & @xmath1770.70 & @xmath1771.54 + ch@xmath5oh ( @xmath178 ) & 8@xmath1797@xmath180 & 220.07849 & 86.1 & cdms & 5.45 @xmath10 0.33 & 3.16 @xmath10 0.33 & 4.48 @xmath10 0.33 & 2.04 @xmath10 0.32 & @xmath1770.79 & @xmath1771.27 + ch@xmath5oh ( @xmath178 ) & 20@xmath18120@xmath182 & 217.88639 & 497.9 & cdms & 3.68 @xmath10 0.22 & 1.07 @xmath10 0.22 & 2.63 @xmath10 0.22 & @xmath1770.86 & @xmath1770.84 & @xmath1771.31 + ch@xmath5oh v@xmath183=1 ( @xmath184 ) & 6@xmath1857@xmath186 & 217.29920 & 363.5 & cdms & 4.40 @xmath10 0.28 & 1.61 @xmath10 0.28 & 3.47 @xmath10 0.28 & @xmath1770.87 & @xmath1770.86 & @xmath1771.72 + ch@xmath5cn & j=1211 , k=3 & 220.70902 & 122.6 & jpl & 6.08 @xmath10 0.27 & 2.56 @xmath10 0.27 & 3.29 @xmath10 0.27 & 1.87 @xmath10 0.35 & @xmath1770.80 & 2.63 @xmath10 0.39 + ch@xmath5cn & j=1211 , k=7 & 220.53932 & 408.0 & jpl & 2.63 @xmath10 0.22 & @xmath1770.65 & @xmath1770.65 & @xmath1770.74 & @xmath1770.49 & @xmath1771.08 + ch@xmath5ch@xmath1cn & 25@xmath18724@xmath188 & 220.66092 & 132.4 & cdms & 4.34 @xmath10 0.29 & @xmath1770.88 & @xmath1770.88 & @xmath1770.71 & @xmath1770.66 & @xmath1771.18 + ch@xmath5ocho ( @xmath184 ) & 20@xmath18919@xmath181 & 216.96590 & 101.1 & jpl & 5.75 @xmath10 0.28 & 4.34 @xmath10 0.28 & 2.53 @xmath10 0.28 & 2.03 @xmath10 0.35 & @xmath1770.99 & @xmath1771.52 + ch@xmath5och@xmath5 & 17@xmath19016@xmath191 & 230.23376 & 136.6 & jpl & 3.19 @xmath10 0.23 & 1.23 @xmath10 0.23 & 1.06 @xmath10 0.23 & @xmath1770.72 & @xmath1770.63 & @xmath1771.25 + hnco & 10@xmath1929@xmath193 & 218.98101 & 90.6 & cdms & 5.28 @xmath10 0.23 & 2.13 @xmath10 0.23 & 2.07 @xmath10 0.23 & 0.85 @xmath10 0.27 & @xmath1770.89 & @xmath1771.44 + lccccccc sma 1b & 143@xmath107 & [email protected] & 490 & ( [email protected])e17 & [email protected] & [email protected] & 8e9 + sma 2 & 140@xmath106 & [email protected] & 390 & ( [email protected])e16 & [email protected] & [email protected] & 1e9 + sma 4 & 208@xmath102 & [email protected] & 340 & ( [email protected])e16 & [email protected] & [email protected] & 2e9 + sma 6 & 95@xmath103 & [email protected] & 330 & ( [email protected])e16 & [email protected] & [email protected] & 3e9 + sma 15 & 72@xmath1015 & [email protected] & 290 & ( [email protected])e16 & [email protected] & [email protected] & 1e9 + sma 18 & 139@xmath1010 & [email protected] & 330 & ( [email protected])e16 & [email protected] & [email protected] & 1e9 + sma 1b comp . 1 & 80@xmath105 & [email protected] & 520 & ( [email protected])e17 & [email protected] & [email protected] & 9e9 + sma 1b comp . 2 & 307@xmath1040 & [email protected] & 310 & ( [email protected])e17 & [email protected] & [email protected] & 9e9 + sma 4 comp . 1 & 135@xmath105 & [email protected] & 390 & ( [email protected])e16 & [email protected] & [email protected] & 8e8 + sma 4 comp .
2 & 280@xmath1035 & [email protected] & 290 & ( [email protected])e16 & [email protected] & [email protected] & 1e9 + lccccccc & & peak intensity & flux density & deconvolved & @xmath194 & @xmath195 + component & @xmath170 ( @xmath171 ) & @xmath172 ( @xmath173 ) & ( jy / beam ) & ( jy ) & size ( @xmath152 ) & ( k ) & ( km s@xmath9 ) + + 1 ( 8) & 17 20 54.327 & -35 45 22.31 & 0.76 & @xmath196 & @xmath154 & 145 & @xmath197 + 2 ( 9 ) & 17 20 54.392 & -35 45 22.68 & 0.57 & @xmath198 & @xmath154 & 180 & @xmath199 + 3 ( 11 ) & 17 20 54.447 & -35 45 22.76 & 1.46 & @xmath200 & @xmath201 & 480 & @xmath202 + 4 & 17 20 54.589 & -35 45 10.33 & 0.79 & @xmath203 & @xmath204 & 150 & @xmath205 + 5 ( 22 ) & 17 20 54.925 & -35 45 13.70 & 1.25 & @xmath206 & @xmath154 & @xmath207 & @xmath208 + 6 ( 25 ) & 17 20 54.952 & -35 45 13.99 & 0.84 & @xmath209 & @xmath210 & 380 & @xmath211 + 7 & 17 20 55.048 & -35 45 14.89 & 0.76 & @xmath212 & @xmath213 & 420 & @xmath214 + 8 & 17 20 55.024 & -35 45 15.32 & 0.60 & @xmath215 & @xmath216 & 110 & @xmath217 + 9 ( 55 ) & 17 20 55.670 & -35 45 00.38 & 6.40 & @xmath218 & @xmath219 & 3100 & @xmath220 + + 9 ( 55 ) & 17 20 55.671 & -35 45 00.34 & 2.28 & @xmath221 & @xmath154 & 270 & @xmath222 + | using the sma and vla , we have imaged the massive protocluster ngc 6334 i(n ) at high angular resolution ( @xmath0 au ) from 6 cm to 0.87 mm , detecting 18 new compact continuum sources .
three of the new sources are coincident with previously - identified h@xmath1o masers .
together with the previously - known sources , these data bring the number of likely protocluster members to 25 for a protostellar density of @xmath2 pc@xmath3 .
our preliminary measurement of the @xmath4-parameter of the minimum spanning tree is 0.82 close to the value for a uniform volume distribution .
all of the ( nine ) sources with detections at multiple frequencies have seds consistent with dust emission , and two ( sma 1b and sma 4 ) also have long wavelength emission consistent with a central hypercompact hii region .
thermal spectral line emission , including ch@xmath5cn , is detected in six sources : lte model fitting of ch@xmath5cn ( @xmath6=1211 ) yields temperatures of 72373 k , confirming the presence of multiple hot cores .
the fitted lsr velocities range from @xmath7 to @xmath8 km s@xmath9 , with an unbiased mean square deviation of 2.05 km s@xmath9 , implying a protocluster dynamical mass of 410@xmath10260 @xmath11 . from analysis of a wide range of hot core molecules , the kinematics of sma
1b are consistent with a rotating , infalling keplerian disk of diameter 800 au and enclosed mass of 10 - 30 @xmath11 that is perpendicular ( within @xmath12 ) to the large - scale bipolar outflow axis . a companion to sma 1b at a projected separation of @xmath13 ( 590 au ; sma 1d ) , which shows no evidence of spectral line emission ,
is also confirmed .
finally , we detect one 218.4400 ghz and several 229.7588 ghz class - i ch@xmath5oh masers . |
bursts of immediate breakings triggered by damage sequences are analogous to acoustic outbreaks in loaded specimens .
however , damage breakings can not be recorded by experimental means they determine the waiting time @xmath3 between consecutive bursts .
the competition of the two failure modes has the consequence that the system drives itself towards failure under a constant subcritical external load @xmath4 .
the global acceleration of the dynamics that can be observed in fig .
[ fig : timeseries ] is the consequence of the increasing load on the intact part of the system due to subsequent load redistributions , while the fluctuations of the burst size @xmath2 and waiting time @xmath3 emerge due to the quenched heterogeneity of fibers strength in qualitative agreement with experiments @xcite . in order to quantify how the accelerating dynamics appears on the microscale we determined the rate of bursts @xmath5 as a function of the distance from the critical point @xmath6 .
[ fig : omori_illesztes]@xmath7 shows that at the beginning of the creep process the event rate monotonically increases having a power law functional form .
approaching catastrophic failure @xmath5 saturates and converges to a constant .
the most remarkable result is that the functional form of @xmath5 can be described by the modified omori law @xcite @xmath8^p } } , \label{eq : omori}\ ] ] where @xmath9 is the saturation rate or productivity at catastrophic failure , @xmath10 denotes the characteristic time scale , and @xmath11 is the omori exponent .
perfect agreement can be observed in fig .
[ fig : omori_illesztes]@xmath7 between the simulated data and the analytic form of equation ( [ eq : omori ] ) . in the case of earthquakes ,
the omori law describes the decay rate of aftershocks following major earthquakes @xcite . for creep rupture
we observe the inverse process : considering the macroscopic failure as the main shock , the breaking bursts are foreshocks whose increasing rate is described by the ( inverse ) omori law . as a crucial point , our approach makes it posssible to clarify how the characteristic time scale @xmath10 of the omori law emerges : fig.[fig : omori_illesztes]@xmath12 illustrates that due to the increasing load on intact fibers shorter and shorter damage sequences are sufficient to trigger bursts .
however , this acceleration is limited such that for @xmath13 the average length of damage sequences @xmath14 saturates between @xmath0 and @xmath1 .
the origin of this high susceptibility is that the load of intact fibers @xmath15 gradually increases to its quasi - static critical value @xmath16 ( see fig.[fig : omori_illesztes]@xmath12 ) @xcite .
hence , the omori time scale @xmath10 is determined by the condition @xmath17 , where @xmath16 can be obtained from the quasi - static constitutive equation of fbms @xcite .
note that the condition @xmath18 marks the point of instability where the avalanche can not stop anymore and it becomes catastrophic .
[ fig : omori_illesztes]@xmath7 also demonstrates that the saturation rate @xmath9 does not depend on the external load @xmath19 , however , the characteristic time scale @xmath10 linearly increases @xmath20 , indicating that at higher load saturation sets on earlier .
our simulations revealed that the omori exponent is @xmath21 , it does not depend on any details of the damage law @xcite such as the @xmath22 exponent and the disorder distributions until the load redistribution is long ranged .
the event rate @xmath5 is practically the inverse of the average waiting time between events occurring at time @xmath23 .
more detailed characterization is provided by the probability distribution of waiting times @xmath24 which is presented in fig .
[ fig : waiting_time ] corresponding to the system of fig .
[ fig : omori_illesztes ] . along the distributions
two characteristic time scales can be identified : for waiting times below a threshold @xmath25 , the distributions have constant values , while in the limit of large waiting times @xmath26 a rapidly decreasing exponential form is obtained . for the intermediate regime
@xmath27 the waiting time distributions exhibit a power law behavior @xmath28 where the exponent proved to be universal @xmath29 . increasing the external load @xmath19 the upper cutoff @xmath30 decreases
, however , the lower characteristic time @xmath31 is independent of @xmath19 . since the temporal occurrence of events
is determined by the global increase of the breaking probability due to the increasing load on intact fibers , the above results suggest that the time evolution of crackling noise of heterogeneous materials undergoing creep rupture can be described as a non - homogeneous poissonian process ( nhpp ) . for nhpp
the waiting time distribution of a series of @xmath32 events with duration @xmath33 can be obtained analytically from the event rate as @xcite @xmath34 to verify the consistency of the nhpp picture for our creeping system , first we fitted the event rate functions by the omori law determining the value of the parameters @xmath9 , @xmath10 , and @xmath11 .
then the analytic form equation ( [ eq : omori ] ) of @xmath5 with the numerical parameters was plugged into equation ( [ eq : wait_nhpp ] ) and the integral was calculated numerically taking into account the load dependent lifetime @xmath35 of the sample @xcite , as well . in fig .
[ fig : waiting_time ] an excellent agreement can be observed between the waiting time distributions obtained from the simulations @xmath36 and the analytic prediction @xmath37 of equation ( [ eq : wait_nhpp ] )
. an important consequence of the above results is that the lower cutoff @xmath31 of @xmath36 can be obtained from the saturation event rate @xmath38 , which does not depend on the external load .
the upper cutoff @xmath30 is determined by the other time scales as @xmath39 . inserting the basquin law of creep life @xmath40 reproduced by our model @xcite and
the linear load dependence of it follows that the upper cutoff @xmath30 scales with the external load as @xmath41 .
the results show that details of the damage process control the global time scale of rupture , however , they do not affect the omori and waiting time exponents .
power law distributions of interoccurrence times in fracture phenomena are usually considered to be the fingerprint of correlations between consecutive events @xcite .
our analysis revealed that for creep phenomena this is not necessarily true , the global acceleration of a heterogeneous system can lead to power law distributed inter - event times without any local correlations .
the comparison of theoretical results to the experimental findings and different types of measurements to each other can be problematic because in laboratory experiments the time series of bursts is never complete : small size bursts generate only low amplitude signals which may fall in the range of background noise @xcite .
devices also have a finite time resolution resulting in a deadtime of detection , during which bursts generated at different positions in space can not be distinguished from each other @xcite .
incompleteness of the time series , especially in field observations , may also be caused by the fact that recording does not start exactly at the time when the load was set .
hence , the beginning of the time series is missing and the measurement is more focused on the vicinity of the failure point where intensive crackling occurs @xcite . in order to capture the effect of the detection threshold of the measuring equipment in the data evaluation , we introduced a threshold value @xmath42 for the size of bursts @xmath2 , i.e. bursts with size @xmath43 are ignored in the time series .
since the size of bursts increases when approaching global rupture ( see fig .
[ fig : timeseries ] ) , the detection threshold removes events typically at the beginning of the time series decreasing the rate of events in this regime .
it can be observed in fig.[fig : omori_hatter]@xmath7 that as @xmath42 increases the functional form of the event rate @xmath44 remains nearly the same described by the omori law equation ( [ eq : omori ] ) , however , the exponent @xmath11 monotonically increases with the threshold value @xmath42 . for the corresponding waiting time distributions @xmath45 in fig.[fig : omori_hatter]@xmath12
, the exponent @xmath46 of the power law regime also increases with increasing @xmath42 , however , the nhpp nature of the event series is preserved at any values of @xmath42 . for nhpps
the two exponents @xmath46 and @xmath11 have the simple relation @xcite @xmath47 which holds with a high accuracy in our system for the numerically determined exponents at any values of @xmath42 ( see fig.[fig : omori_hatter]@xmath48 ) .
it is important to emphasize that the exponents increase due to the non - stationary nature of creep rupture : the simultaneous increase of the rate and size of events towards failure has the consequence that the finite detection threshold mainly affects the beginning of the time series resulting in a few long waiting times up to the first bigger bursts .
their statistics is characterized by a peak or small bump in fig.[fig : omori_hatter]@xmath12 , while the rest of waiting times have a steeper power law distribution .
the results demonstrate that the detection threshold has a dramatic effect on the outcomes of the analysis of crackling time series , just varying @xmath42 practically any values can be obtained for the waiting time exponent @xmath46 between @xmath0 and @xmath1 .
the finite time resolution @xmath49 of the detectors has the consequence that bursts pile up , i.e. since bursts can not be distinguished within the duration @xmath49 , the size of bursts sums up giving rise to larger event sizes in the time series .
the effect of the deadtime is captured in the data evaluation such that if an avalanche of size @xmath50 occurred at time @xmath51 , all those avalanches which appeared in the interval @xmath52 are added to @xmath50 . in fig.[fig : holdido_dist]@xmath7 at zero deadtime @xmath53 where all avalanches are distinguished the size distribution @xmath54 has a power law form @xmath55 followed by an exponential cutoff . the value of the exponent @xmath56 is equal to the usual mean field burst exponent of fbms @xcite .
as @xmath49 increases the pile up of bursts promotes large events while the small ones get suppressed . due to the acceleration of the failure process
pile - up gets dominating in the vicinity of macroscopic failure , hence , in fig.[fig : holdido_dist]@xmath7 the value of @xmath49 is compared to @xmath31 of the waiting time distribution . as a consequence ,
the waiting time distributions hardly change , however , the burst size distribution @xmath57 has a crossover to a power law of a significantly lower exponent @xmath58 showing the higher frequency of large events in the statistics . in fig.[fig : holdido_dist]@xmath7 pile up becomes dominating already at @xmath59 , which shows the importance of the results for real experiments .
recently , laboratory experiments on earth materials have revealed that the b - value , i.e. the exponent of the probability distribution of the energy of the time series of acoustic events of rupture cascades decreases in the vicinity of failure @xcite . to investigate the possibility of an analogous phenomenon in creep rupture
, we constrained the data evaluation to events occurring in a time interval of duration @xmath60 preceding macroscopic failure and determined the probability distribution @xmath61 .
it can be observed in fig .
[ fig : holdido_dist]@xmath12 that approaching macroscopic rupture @xmath62 , where the largest avalanches are triggered , the burst size distribution @xmath61 exhibits a crossover : at a characteristic burst size @xmath63 the exponent of @xmath61 has a striking change from @xmath56 to a surprisingly low value @xmath64 .
the value of @xmath63 extends to the largest avalanche as @xmath65 decreases .
the crossover is accompanied by the change of the waiting time distribution , as well : since in the regime @xmath62 the event rate is constant , the power law regime of @xmath66 disappears and the distribution turns to a pure exponential as it is expected for homogeneous ( constant event rate ) poissonian processes @xcite .
acoustic outbreaks generated by nucleating and propagating cracks provide the main source of information on the microscopic temporal dynamics of creep rupture . for the understanding of acoustic monitoring data of engineering constructions and of field measurements on steep slopes or rock walls in mountains
requires the application of statistical physics .
our analysis showed that time - to - failure power laws of macroscopic quantities such as creep rate commonly observed in experiments are accompanied by the emergence of omori type acceleration of the bursting activity on the microscale .
the origin of the omori time scale is that the aging system drives itself to a critical state where a few breakings are sufficient to trigger extended bursts .
the omori law is known to describe the relaxation of the rate of aftershocks following major earthquakes @xcite , and it has also been confirmed for foreshocks when observed @xcite .
our results suggest the interpretation that acoustic bursts in creep behave like foreshocks of the imminent catastrophe .
our investigations unveiled that the evolving time series of crackling events is the result of an underlying non - homogeneous poissonian process .
it has the striking consequence that observing power law distributed waiting times in fracture may not imply the presence of dynamic correlations , up to a large extent it can be caused by the global acceleration of the system .
we showed that special care should be taken when comparing results of model calculations to measurements on crackling noise , since the deadtime of devices and the finite background noise to signal ratio can even affect the measured value of critical exponents .
varying solely the detection threshold of events , for the distribution of waiting times any exponents can be obtained between @xmath0 and @xmath1 covering the range of experimental results @xcite . due to the finite deadtime of electronics bursts pile up which gives rise to a crossover to a lower exponent of the size distribution of bursts .
recently , avalanches have been identified with a high spacial resolution along a propagating crack front using optical imaging techniques @xcite .
considering global avalanches in the same experiment integrates bursts along the front , giving rise to a significantly lower exponent in agreement with our predictions @xcite .
components of engineering constructions are mainly subject to creep loads @xcite , and creep rupture often lies at the core of natural catastrophes such as landslides , snow and stone avalanches , as well @xcite
. we demonstrated that restricting the measurement to the close vicinity of ultimate failure , the size distribution of bursts exhibits a crossover to a significantly lower exponent , which is accompanied by the change of the functional form of the waiting time distribution .
since the crossover is controlled by the time scale of the omori law these results can be exploited for forecasting the imminent catastrophic failure event .
here we focused on the case of long range load redistribution following failure events .
when the load sharing is localized the spatial correlation of failure events leads to the emergence of a propagating crack front .
the load accumulated along the crack front gives rise to an overall acceleration of the failure process again with a non - homogeneous poissonian character .
however , at short time scales correlated clusters of events may arise inside the time series .
the results imply the interesting question to clarify when studying burst time series under creep whether there is anything in the dynamics beyond non - homogeneous poissonian processes .
we use a generic fiber bundle model @xcite of the creep rupture of heterogeneous materials which has succesfully reproduced measured creep behavior @xcite .
the sample is discretized in terms of a bundle of @xmath67 parallel fibers having a brittle response with identical young modulus @xmath68 .
the bundle is subject to a constant external load @xmath19 below the fracture strength of @xmath69 the system .
it is a crucial element of the model that the fibers break due to two physical mechanisms : immediate breaking occurs when the local load @xmath70 on fibers exceeds their fracture strength @xmath71 , @xmath72 .
time dependence is introduced such that those fibers , which remained intact , undergo an aging process accumulating damage @xmath73 @xcite .
the rate of damage accumulation @xmath74 is assumed to have a power law dependence on the local load @xmath75 of fibers @xmath76 , where @xmath77 is a constant and the exponent @xmath22 controls the time scale of the accumulation process .
fibers can tolerate only a finite amount of damage so that when the total damage @xmath73 accumulated up to time @xmath23 exceeds a local damage threshold @xmath78 the fiber breaks .
the two breaking thresholds @xmath79 and @xmath78 , @xmath72 of fibers are independent random variables which are for simplicity uniformly distributed between @xmath80 and @xmath0 . after each breaking event the load dropped by the broken fiber
is equally redistributed over the remaining intact ones @xcite .
the separation of time scales of slow damage and of immediate breaking leads to the emergence of a bursty evolution of the system : damaging fibers break slowly one - by- one , gradually increasing the load on the remaining intact fibers .
after a certain number of damage breakings the load increment becomes sufficient to induce the immediate breaking of a fiber which in turn triggers an entire burst of breakings . as a consequence , the time evolution of creep rupture occurs as a series of bursts corresponding to the nucleation and propagation of cracks , separated by silent periods of slow damaging .
the size of burst @xmath2 is defined by the number of fibers breaking in a correlated trail , while the waiting time @xmath3 between consecutive events is the duration of the damage sequence of @xmath81 breakings which triggers the next event .
macroscopic failure occurs in the form of a catastrophic avalanche at time @xmath33 which defines the lifetime of the system @xcite .
our model reproduces the basquin law of creep life , i.e. the lifetime @xmath33 decreases as a power law of the external load @xmath40 , where the exponent coincides with the exponent of damage accumulation @xmath22 @xcite .
computer simulations were carried out in our fbm with @xmath82 fibers averaging over @xmath83 samples for each parameter set except for fig .
[ fig : timeseries ] where intensionally a small bundle of @xmath84 fibers were considered .
the exponent @xmath22 of the damage law mainly sets the global time scale of the evolution , hence , the simulation results are presented only for @xmath85 .
other parameters of the model are fixed as @xmath86 and @xmath87 @xcite .
this work was supported by the projects tamop-4.2.2.a-11/1/konv-2012 - 0036 , tamop-4.2.2/b-10/1 - 2010 - 0024 , tmop4.2.4.a/2 - 11 - 1 - 2012 - 0001 , otka k84157 , and eranet_hu_09 - 1 - 2011 - 0002 . | creep rupture of heterogeneous materials occurring under constant sub - critical external loads is responsible for the collapse of engineering constructions and for natural catastrophes .
acoustic monitoring of crackling bursts provides microscopic insight into the failure process .
based on a fiber bundle model , we show that the accelerating bursting activity when approaching failure can be described by the omori law . for long range load redistribution the time series of bursts proved to be a non - homogeneous poissonian process with power law distributed burst sizes and waiting times .
we demonstrate that limitations of experiments such as finite detection threshold and time resolution have striking effects on the characteristic exponents , which have to be taken into account when comparing model calculations with experiments .
recording events solely within the omori time to failure the size distribution of bursts has a crossover to a lower exponent which is promising for forecasting the imminent catastrophic failure .
materials subject to a constant external load below their fracture strength typically exhibit a time dependent response and fail in a finite time .
such creep rupture phenomena have an enormous technological importance and human impact since they are responsible for the collapse of engineering constructions and they lie at the core of natural catastrophes such as landslides , snow and stone avalanches and earthquakes @xcite .
the acoustic waves generated by the nucleation and propagation of cracks allow for the monitoring of the failure process on the meso- and micro scales .
crackling noise is usually characterized by the integrated statistics accumulating all the events of the time series up to failure @xcite .
experiments revealed that the probability distribution of the energy of crackling bursts and of the interoccurrence times have power law functional form , which are considered to be the fingerprint of correlations in the microscopic breaking dynamics @xcite .
the value of the exponents measured on different types of heterogenous materials show a surprisingly large scatter between @xmath0 and @xmath1 which has not been captured by theoretical studies @xcite .
the approach to failure is usually characterized on the macroscale by the strain rate which proved to have a power law divergence as a function of time to failure @xcite . here
we take a different strategy and investigate the details of the crackling time series in order to understand how the creeping system evolves towards catastrophic failure .
we consider a generic fiber bundle model @xcite ( fbm ) of damage enhanced creep rupture which successfully reproduces measured creep behaviour of heterogeneous materials both on the micro and macro scales ( methods ) @xcite . in the model under a constant subcritical external load the fibers break due to two physical mechanisms : immediate breaking occurs when the load of fibers exceeds the local failure strength .
time dependence emerges such that intact fibers accumulate damage which results in failure in a finite time .
the separation of time scales of slow damaging and of immediate breaking together with the load redistribution following failure events lead to a highly complex time evolution where slowly proceeding damage sequences trigger bursts of immediate breakings @xcite .
an example of the time series of bursts can be seen in fig.[fig : timeseries ] where the increasing burst size @xmath2 and the decreasing waiting time @xmath3 between consecutive events clearly mark the acceleration of the system towards failure . as a novel approach to creep we focus on the evolution of the rate of bursts and show that the temporal occurrence of crackling events and the power law statistics of waiting times can fully be described based on non - homogeneous poissonian processes without assuming correlations of bursts .
our investigation unveils that limitations of measuring devices in experiments have astonishing effects on the outcomes of crackling noise analysis which can explain the strong scatter of measured critical exponents of crackling noise in creep , and the discrepancy between experimental findings and theoretical approaches .
studying how the time series evolves when approaching the catastrophe we address the possibility of forecasting the imminent failure .
the importance of the results goes beyond fracture phenomena and catastrophic failures , recently the human activity has been found to exhibit similar bursty character where analogous problems of the evolution of time series and waiting time statistics occur @xcite . |
chua , l.o . , kocarev , lj . , eckert , k. & itoh , m. [ 1992 ] experimental chaos synchronization in chua s circuit " int . j. bifurcation and chaos * 2 * 705 - 708 . | the idea of secure communication of digital signals via chaos synchronization has been plagued by the possibility of attractor reconstruction by eavesdroppers as pointed out by prez and cerdeira . in this letter
, we wish to present a very simple mechanism by which this problem can be overcome , wherein the signal is transmitted via a multistep parameter modulation combined with alternative driving of different transmitter variables , which makes the attractor reconstruction impossible .
the method is illustrated by means of the lorenz system and chua s circuit as examples .
the work of pecora and carroll [ 1990,1991 ] on synchronization of chaotic signals has suggested the possibility of secure communication using chaos synchronization .
they have shown that two chaotic systems can be synchronized when they are linked by a common signal ( driving signal ) provided the lyapunov exponents of the subsystem are all negative . following this suggestion a number of methods
have been proposed for secure communication [ lakshmanan & murali , 1996 ; hayes et al . , 1993 ; cuomo & oppenheim , 1993 ; murali & lakshmanan , 1993a,1993b ; kocarev & parlitz , 1995 ; zhang et al . , 1998 ] with the help of chaotic signals .
cuomo and oppenheim [ 1993 ] have shown that a small difference between the corresponding parameters in the drive and response systems will cause synchronization frustration between the transmitter and receiver variables . using this property ,
they have suggested a very simple method for digital signal transmission and illustrated it with the lorenz system .
however , prez and cerdeira [ 1995 ] have pointed out that it is possible to reconstruct the message by an eavesdropper from a simple return map formed by the extrema of the modulated driving signal ( here , the driving signal is used as a carrier signal in the digital signal transmission . in our method
the digital message is imposed on this carrier signal through parameter variation .
so , we call the driving signal bearing the digital message as modulated driving signal and the modulation as parameter modulation . ) , even without synchronization with the receiver system . to overcome the above problem several methods
have been suggested by many authors recently [ murali & lakshmanan , 1998 ; mensour & longtin , 1998 ; minai & pandian , 1998 ] : private communication using compound chaotic signal technique , using delay - differential equations technique , communication through noise and so on .
however , most of the above methods are very complicated and are often difficult to implement . in this letter , we describe a simple method by which the digital signal can be highly masked with the chaotic signal by using a multistep parameter modulation and show how it can be further complicated by introducing two different drive signals alternatively instead of a single drive signal . as a result , reconstruction of
the message becomes virtually impossible and this provides further development in the technique of secure communication using chaotic signals . in order to present our method , we make use of the dynamical system used by cuomo and oppenheim [ 1993 ] namely the lorenz system . the associated evolution equations at the transmitter end are @xmath0 while at the receiver end , corresponding to an @xmath1-drive , they read as @xmath2 where @xmath3 and the parameter @xmath4 chosen for the purpose of modulation can have either the value @xmath5 or @xmath6 .
the numerical value of the parameter @xmath4 at the receiver end is kept at a constant value 4.0 , while in the transmitter circuit it is changed between two values , namely 4.0 and 4.4 , when the digital signal is zero and one , respectively .
the transmitted digital message is reconstructed by using the synchronization error power @xmath7 .
it is negligible when the transmitter and the receiver systems are synchronized , while it has some finite value when they are not synchronized .
thus the receiver will synchronize with the transmitter when the parameter @xmath4 in the latter takes a value @xmath5 , while asynchronization takes place when @xmath8 at the transmitter .
it may be noted that instead of the @xmath1-drive at the receiver end one can as well use the @xmath9-drive in which case the eq.(2 ) will be modified as @xmath10 in the work of prez and cerdeira [ 1995 ] , it has been shown that even without any receiver circuit , one can unmask the message from the modulated drive signal . in this reference , the maxima @xmath11 and minima
@xmath12 are collected from the modulated driving signal and the new varibles @xmath13 and @xmath14 are defined .
the plot of the return map for @xmath15 vs @xmath16 is shown in fig .
1 . note that there are 3 segments in the attractor each one of which is further split into two strips ( one corresponding to the high state and the other corresponding to low state of the digital message ) .
it is obvious that the split in the attractor is due to the change of the parameter @xmath4 at the sender end between the values @xmath5 and @xmath6 . from the return map
one can easily unmask the message by finding the strip on which the point@xmath17 falls in each segment of the attractor .
we can call the above procedure as a single - step parameter modulation , since @xmath4 can have only one value for one state of the message ( that is , @xmath4 can take only a single value @xmath6 for any high state in the digital message and only @xmath5 for the low state digital signal at the transmitter ) .
the crux of the problem is then that from the nature of the chaotic attractor and the return map of the lorenz system , one can have only eight possibilities of message reconstruction from which one can easily discover the actual message .
then , one possible way to overcome such a reconstruction is to increase the number of possible ways by which message can be reconstructed to an unmanageable level and thereby eliminating the possibility of identifying the correct message . in our work , we have effected two changes in the mode of transmission of the digital message by the parameter modulation in order to complicate the attractor in the return map .
firstly , in the sender part , instead of using a single - step parameter modulation , we use a multi - step parameter modulation , where for one state of the digital message ( either @xmath18 or @xmath19 ) , the parameter @xmath4 in the sender equation is assigned to have any one of the @xmath20 preassigned values ( @xmath21 and sufficiently large ) , while at the receiver part we use @xmath20 receiver subsystems with different parameters .
we call the number @xmath20 as the step of the modulation .
secondly , we will use both @xmath1 and @xmath9 signals to drive the receiver subsystems alternatively during the transmission as indicated in eqs.(2 ) and ( 3 ) .
the block diagram of the transmitter and the receiver with modulation step @xmath20 is shown in fig .
2 , where r@xmath22,r@xmath23, ... r@xmath24 are the subsystems of the receiver .
each one of the @xmath25s is assigned a specific value among the @xmath20 chosen parameter values which are used to impose the high state of the digital message on the driving signal .
the synchronization error power at the output of each subsystem is fed into low pass filters(lpfs ) separately .
then the filtered signal is converted into a digital signal at threshold detectors(td)[cuomo & oppenheim , 1993 ] , which is then passed through not gates in order to invert and to confirm with the original assignment of high and low states of the digital signal .
an or gate at the receiver combines all the outputs from the not gates .
thus , when the digital message imposed on the driving signal is @xmath18 , any one of the @xmath20 subsystems at the receiver will synchronize and the output at the or gate will be @xmath18. none of the subsystems will synchronize , if the modulated digital message is zero , and hence @xmath19 will be the output of the or gate .
we illustrate our method in the case of the lorenz system with the same parameters used by prez and cerdeira [ 1995 ] and with the modulation step as small as @xmath26 .
the lorenz system ( 1 ) is known to exibit chaotic behavior when the parameter @xmath4 takes any value between @xmath27 and @xmath28 .
one can assign a specific value to @xmath4 within this range for the purpose of modulation .
we select the region @xmath29 for the illustration of our method . as an example , the parameter @xmath4 in the transmitting system is allowed to take any one of the five values @xmath30 for high state of the digital signal , while it can have any one the values @xmath31 for the low state . in the receiver part
, we use 5 subsystems with the modulation parameter @xmath4 fixed at @xmath30 respectively .
the method works as follows .
suppose we have a high state in the digital message , to start with .
then , the receiver is driven by the modulated driving signal with modulation parameter @xmath32 .
for the next high state in the message the modulation is done with @xmath4 = @xmath33 and this process continues upto the @xmath4 value @xmath34 .
then the value of @xmath4 is reset to @xmath35 .
this procedure is also applicable to the low state of the message but with the modulation parameter(@xmath4 ) taking the values @xmath36 in that order . in the receiver
part , all the subsystems are driven by the modulated driving signal . however , the subsystems @xmath37 and @xmath38 will synchronize only for the values of modulation parameter at the transmitter , @xmath39 respectively . when the synchronization is achieved in a subsystem
, the error power will be below the threshold level and so the message obtained from td will be at the low state .
it is inverted to the high state using the not gate .
but a low state is attained by the digital message at the not gate , if the subsystem is not synchronized .
thus a high state message is obtained as an output at the or gate if any one of the subsystems is synchronized and a low state digital message is recovered when asynchronization takes place at all the subsystems . a test digital message transmitted and recovered by this scheme by numerical simulation is shown in fig .
3 .
actually the above possibility is only a test case and has been discussed for illustrative purpose .
one can modulate @xmath4 in a random way also ( that is , at the transmitter circuit , @xmath4 can be assigned any one of the @xmath20 chosen values in a random manner ) rather than in a predetermined way .
the method presented here is applicable equally well to such possibilities also .
the return map constructed using the modulated driving signal is shown in fig .
4 . here
we have 10 strips in each segment .
this corresponds to @xmath40 strips , where @xmath20 is the step of the parameter modulation .
because of the complicated nature of the return map it is very difficult to find the strip on which the point@xmath17 falls in each segment of the attractor and so it is difficult to unmask the message from the return map . even if one can identify the location of the points in the strip , still one has a serious difficulty in assigning 0 and 1 to the strips since there are @xmath41 chances to make a mistake , for a successful reconstruction of the message . for @xmath42
, there are 7 chances [ pre & cerdeira , 1995 ] of making a mistake during the unmasking of the message from the modulated driving signal and for @xmath26 , this chances increases to a high value of 1,067,462,647 ( of the order of @xmath43 ) .
so it is almost impossible to unmask the message masked by our scheme , even at as low value as @xmath26 .
we can complicate the attractor in the return map further by the second method . in the lorenz system , synchronization [ pecora & carroll , 1991 ] is possible by both @xmath1 driving and @xmath9 driving , since the conditional lyapunov exponents are all negative in both the cases .
we use this property in the second method . here
the principle of transmitting and receiving the message is the same as described above , but instead of using a single driving signal @xmath44 , we now use both @xmath44 and @xmath45 as the driving signals alternatively . for the odd digital bits , we drive the receiver by modulated @xmath44 whereas the even digital bits are transmitted using modulated @xmath45 as the driving signal . the block diagram of the transmitting and receiving systems with modulation step @xmath20 is shown in fig .
5 . here
we use @xmath46 subsystems in the receiver , @xmath47 subsystems for each value of modulation parameter ( which are used for the transmission of high state of the digital message ) , one with modulated @xmath1 driving and the other with modulated @xmath9 driving .
the recovered message is similar to the one obtained in our previous method .
the return map constructed from the modulated drive signals ( modulated x(t ) and modulated y(t ) ) for the modulation step @xmath20=5 is shown in fig .
6 . from the return map ,
it is impossible to unmask the message since the attractor of the @xmath9 drive signal is merged with the attractor of the @xmath1 driving signal .
we have also applied our scheme to the simplest autonomous chua circuit generator of the chaotic signal [ madan , 1993 ; chua et al . ,
1992 ] .
we have used the unnormalized circuit equations @xmath48 where @xmath49 $ ] during our verification with @xmath50 as the modulation parameter .
the other parameters value used are @xmath51 , @xmath52 and @xmath53 .
the chua circuit can also be synchronized either with the driving signal @xmath54 or with @xmath55 and so we can again use them as alternative drive variables .
one can assign any value to the parameter @xmath50 between @xmath56 to @xmath57 , except in some small regions where periodic windows occur . here
the largest lyapunov exponent calculated has positive values and hence the circuit exibits chaotic behaviour .
we find from our numerical studies that it is even more difficult to extract message in this case with multiparameter modulation compared to the lorenz system , due to the nature of the chaotic attractor .
thus , in this letter , we have described two simple and straightforward methods to complicate the attractor in the return map of the modulated driving signal : one with multi - step parameter modulation by using a number of values to the modulation parameter @xmath4 and the other by using alternative @xmath1 and @xmath9 driving combined with multistep parameter modulation .
we have illustrated them with the lorenz system and chua s circuit as examples .
we conclude that it is almost impossible to reconstruct the digital message transmitted in our scheme by the use of the return map formed by the modulated driving signal .
this work has been supported by the national board of higher mathematics , department of atomic energy , government of india and the department of science and technology , government of india through the research projects .
the authors thank dr .
k. murali for many valuable suggestions . |
ultracompact ( uc ) hii regions are very small ( @xmath50.1 pc ) and dense ( @xmath1@xmath11 @xmath12 ) ionized regions ( wood @xmath13 churchwell 1989a , hereafter wc89 ; kurtz , churchwell , @xmath13 wood 1994 , hereafter kcw ) .
they are bright radio and infrared sources and seem to be still embedded in the parental molecular clouds . thus these objects have been thought to represent an early evolutionary stage of massive stars .
one of the most interesting problems related to them is that too many uc hii regions are observed in the galaxy , i.e. , the `` age problem '' : the number of uc hii regions is greater by about an order of magnitude than expected from other indicators of massive star formation rate based on their dynamical age ( wc89 ; wood @xmath13 churchwell 1989b ) .
hence , it has been considered that hii regions should remain in the ultracompact phase for a much longer ( @xmath110@xmath8 yr ) period than their sound crossing time ( @xmath510@xmath6 yr ) in a uniform medium .
several different models were proposed to resolve this problem : the infall model , photo - evaporating disk model , pressure - confined model , and stellar - wind supporting bow shock model ( see churchwell 2000 and references therein ) .
these models present various mechanisms which constrain hii regions to remain longer in the ultracompact phase .
on the other hand , there is some observational evidence to suggest that uc hii regions are _ not _ as young as they appear .
the ultracompactness of uc hiiregions might arise because interferometric observations are insensitive to structures larger than a certain extent depending on the baseline range . using the very large array ( vla ) d - array , which is sensitive to large ( @xmath215@xmath3 ) structures , koo et al .
( 1996 ) found that [email protected] , known as an uc hii region , is actually a large hii region complex that contains structures of diverse length scales ranging from 0.04 to 40 pc : it has an ultracompact core , a compact core , an extended halo , and a diffuse large plateau .
the presence of the extended envelope could be direct evidence supporting that the uc hii region is much older than 10@xmath6 yr ( see 5 for more details ) .
it also implies that there are much more ionizing photons than what would have been estimated from the ultracompact core alone .
a few other groups have also noted the importance of extended emission around uc hii regions in understanding the nature of hii regions ( garay et al .
1993 ; kurtz et al .
for example , kurtz et al . ( 1999 ) detected extended emission towards 12 out of 15 randomly selected sources from the kcw survey , and suggested on the basis of its morphology that the extended emission may be directly connected with uc hii regions in about half ( 8/15 ) of the sources .
however , the physical relationship between uc hii regions and their extended envelopes is still poorly understood . in this study
, we undertook vla radio continuum observations of 16 uc hii regions in search of associated extended emission and detected diffuse extended emission in all the sources .
a preliminary version of these results was presented by kim @xmath13 koo ( 1996 ) .
we also made h76@xmath0 and he76@xmath0 radio recombination line ( rrl ) observations in order to study the physical association between the uc hii regions and their extended envelopes .
the observations are described in 2 and the results are presented in 3 .
we investigate the relationship between the parameters of the uc hii regions and their envelopes in 4 and discuss the physical association between the two , the origin of extended envelopes , and their implication in resolving the age problem in 5 .
some individual sources are discussed in 6 and our main conclusions are summarized in 7 .
we have carried out radio continuum observations of 16 uc hii regions with @xmath240@xmath1420@xmath15 resolution at 21 cm ( table 1 ) .
the observations were made during 1995 february using the vla of the national radio astronomy observatory ( nrao ) in the dnc hybrid configuration .
so our observations are sensitive to structures smaller than 15@xmath3 .
we have selected 16 uc hii regions from the wc89 catalog ( table 2 ) .
these uc hii regions are ones with simple morphology and large ( @xmath110 ) ratios of single - dish to vla fluxes . here simple morphology means that they each are not resolved into two or more clearly distinct components .
this requirement is to avoid possible confusion by nearby uc hiiregions in exploring the association between an uc hii region and extended emission .
they consist of one shell - type , two spherical , three core - halo type , four cometary , and six irregular ( or multiply peaked ) uc hii regions . in order to improve the _ u - v _ coverage ,
each source was observed at several hour angles .
the flux density was referred to 3c 286 , which was assumed to have flux densities of 15.03 jy ( at 1.385 ghz ) and 14.64 jy ( at 1.465 ghz ) .
the data were edited , calibrated , imaged , and cleaned following the standard vla procedures in the astronomical image processing system ( aips ) .
self - calibration was accomplished for several appropriate sources to reduce the rms noise level of their final images .
we observed h76@xmath0 rrls , which have a rest frequency of 14,689.99 mhz , toward all of the objects in our sample .
the observations were undertaken with the 140 foot ( 43 m ) telescope of the nrao in 1997 february and june .
the telescope has an angular resolution ( fwhm ) of about 2@xmath3 and a main beam efficiency of 0.58 at the observing frequency . in order to diminish the effects of standing waves between the telescope and receiver , we observed with focus
alternately displaced @xmath16@xmath17/8 from its peak value .
both circular polarizations were observed simultaneously using two 1024 channel autocorrelators with 40 mhz bandwidth each , and thus the velocity resolution was 1.59 km s@xmath10 after hanning smoothing .
each spectrum was obtained by integrating typically for 20 minutes using frequency switching . the system temperature varied in the range 40@xmath460 k during the observing sessions depending on weather conditions and elevation of the source .
we mapped [email protected] ( 16@xmath3@xmath1810@xmath3 ) and [email protected] ( 18@xmath3@xmath1814@xmath3 ) in full - beam spacing .
for the other sources , we made observations at 1@xmath413 positions including the peak positions in each envelope ( see figs .
1@xmath191@xmath20 ) .
he76@xmath0 ( 14,695.97 mhz )
rrls were also observed at a total of 18 positions in 6 sources ( see table 5 ) .
these positions are ones where strong h76@xmath0 line emission was detected .
we detected extended emission in all our sources .
figures 1@xmath191@xmath20 show their images .
the extended envelopes are from @xmath21 to @xmath22 in angular extent , and their morphologies range from a relatively simple structure with a single peak ( e.g. , g8.14 + 0.23 , g23.71 + 0.17 , and g23.96 + 0.15 ) to a fairly complicated structure with several compact components ( e.g. , [email protected] , [email protected] , and [email protected] ) .
all the uc hii regions except for [email protected] and g25.72 + 0.05 are located at the peak positions .
both of the latter are spherical uc hii regions lying near the edge of extended hii regions .
if we adopt the distances given by wc89 ( table 2 ) , the geometric mean diameters range from 0.56 pc to 5.68 pc for the compact components and from 3.8 pc to 18.7 pc for the extended envelopes .
based on their linear sizes ( and physical parameters ) , the compact components would have been classified as compact or dense hii regions while the extended components would have been categorized as classical hii regions ( habing & israel 1979 ) .
table 3 presents the basic observed parameters of our sources . in the first column of this table ,
a@xmath4c represent the brightest three compact components in order of decreasing peak brightness .
the first entry for each source refers to the extended component including uc hii region and compact component(s ) . in the sources with two or more compact components ,
the uc hii regions tend to lie in the smallest and densest ones .
this is shown in figure 2 , where we plot angular and linear sizes versus distance for the compact components .
most ( 11/14 ) of the compact components with uc hii regions are smaller than 2 pc , whereas all but two without uc hii regions are larger .
the average diameter ( 1.6 pc ) of the compact components with uc hii regions is about half of that of those without uc hii regions ( 3.5 pc ) . on the other hand
, there seems to be no relationship between the morphologies of uc hii region and its extended envelope .
assuming that the hii region is spherically symmetric , optically thin , homogeneous , dust - free , and ionization - bounded , we derived the electron number density @xmath23 , emission measure @xmath24 , excitation parameter @xmath25 , and mass @xmath26 of the ionized gas from the observed integrated flux densities ( schraml @xmath13 mezger 1969 ; panagia @xmath13 walmsley 1978 ) .
we also estimated the lyman continuum photon flux @xmath27 required to ionize each source using the formulae of rubin ( 1968 ) .
here we used the electron temperatures given by downes et al .
( 1980 ) and wink et al .
table 4 summarizes the results .
the table also lists the spectral type of the ionizing star of each source ( panagia 1973 ) .
this assumes that a single zero - age main sequence ( zams ) star is responsible for the ionization .
the spectral types of the compact components with uc hii regions are approximately three subclasses earlier than those measured from uc hii regions alone in wc89 , which are given in parentheses .
if a source is produced by a stellar cluster , rather than by a single star , then the most luminous star in the cluster would be about two subclasses later than the equivalent single - star spectral type ( see , e.g. , kcw ) .
we determined the fraction of ultraviolet ( uv ) photons absorbed by dust within the ionized gas , @xmath28=1@xmath4@xmath27/@xmath29 , where @xmath29 is the total rate of lyman continuum photons of the ionizing star ( table 4 ) . in this calculation , we adopted the value of @xmath29 from wc89 , which was derived primarily from the @xmath30 data under the assumption that a single star provides the entire infrared luminosity .
we used the value of the associated ( or the nearest ) compact component for @xmath27 , considering the low ( @xmath24@xmath3 ) resolution of @xmath30 observations .
our values of @xmath28 might be overestimates because the sizes of the compact components are frequently smaller than the @xmath30 beamsize . in the majority ( 11/16 ) of our sources , nevertheless , the estimated fractions ( @xmath3160% ) are significantly less than those ( @xmath180% ) determined by wc89 and kcw for most uc hii regions in their catalogs .
this discrepancy may be mainly due to the contribution from extended emission around uc hii regions .
accordingly , the dust absorption of stellar uv radiation within the hii regions seems to be less important than suggested by previous studies .
we detected h76@xmath0 line emission in all the sources except [email protected] , which was the weakest radio continuum source in our sample .
figure 3 shows a sample from our rrl spectra at [email protected] ( + 05.@xmath322 , + 00@xmath3 10@xmath15 ) , where the strongest h76@xmath0 line emission was detected .
we performed a least - squares fit of each spectrum to a single gaussian profile .
table 5 gives the gaussian parameters of the h76@xmath0 lines .
in this table , columns ( 2)@xmath4(5 ) are the equatorial coordinates , the central velocity , the line intensity , and the line width ( fwhm ) , respectively .
the equatorial coordinates are given as offsets with respect to the position of uc hii region .
errors quoted are statistical @xmath161 @xmath33 errors in the fit only .
the central velocities of our rrls detected toward uc hii regions are in good agreement with those of previous single - dish observations ( downes et al .
1980 ; wink et al . 1982 ; lockman 1989 )
. we can find no significant velocity difference between the spectrum observed toward each uc hii region and the average spectrum of its extended envelope with one exception ( g25.72 + 0.05 ) .
this indicates that the two components are physically associated . in the case of g25.72 + 0.05 , h76@xmath0 line emission
was observed at 98 km s@xmath10toward the uc hii region while it was detected at 54 km s@xmath10 in the nearby compact component , suggesting that they constitute a chance coincidence in the line of sight . in seven sources ( [email protected] , [email protected] , [email protected] , [email protected] , [email protected] , [email protected] , and [email protected] ) , on the other hand , the ionized gas in some part of the envelope is blue- or redshifted by 3@xmath410 km s@xmath10 with respect to the compact component associated with an uc hii region .
for instance , the eastern part of [email protected] is blueshifted from its central region by @xmath210 km s@xmath10 ( fig .
4 ) . we will discuss in detail the velocity structure of individual sources in 6 .
the h76@xmath0 line width is usually larger in the compact components , especially in the ones with uc hii regions , than in the extended envelopes for the resolved sources ( see 6.1 and 6.2 ) .
this trend probably occurs because turbulent motions , including large - scale expansion or contraction motions , are larger in the compact components than in the extended envelopes , as proposed by garay & rodrguez ( 1983 ) .
he76@xmath0 line emission was detected at 10 out of the 18 observed positions ( table 6 ) .
the singly ionized helium abundance ( by number ) y@xmath34 was estimated at these positions from the ratio of the he76@xmath0 to h76@xmath0 line integrals .
the derived values ( [email protected] ) for [email protected] , [email protected] , and [email protected] are comparable to the average value ( [email protected] ) in our galaxy ( shaver et al .
1983 ) , while the values ( [email protected] ) for [email protected] , [email protected] , and [email protected] are significantly lower .
the low value of y@xmath34 implies that either the sizes of heii regions are significantly smaller than those of hii regions or the he abundance y is low at the observed positions . for [email protected] and [email protected] ,
the observed low value is likely to be due to the low effective temperature of the ionizing stars ( o7 ) ( mezger 1980 ) .
additionally these two objects are at galactocentric distances of 6@xmath47 kpc where the metal abundance is known to be significantly higher than in the solar neighborhood ( shaver et al .
the high metallicity could make the sizes of heii regions smaller .
this is not true for [email protected] , which has ionizing stars of spectral type o5 or earlier .
such stars emit hard enough photons to singly ionize helium in the entire hii region .
this view is supported by the fact that normal values were observed for y@xmath34 in [email protected] , [email protected] , and [email protected] , which have ionizing stars of o5 or o5.5 .
we can also exclude the possibility that y is low in the region because the value ( @xmath350.063 ) is approximately normal in [email protected] , which is located in the same molecular cloud ( see 6.3 ) .
a possible explanation for the low y@xmath34 observed in [email protected] is that helium ionizing photons are preferentially absorbed by dust in the nebula as in sgr b2 ( e.g. , chaisson , lichten , & rodrguez 1978 ) .
what is the physical relationship between the ultracompact , compact , and extended components in each source ? according to our rrl observations , the compact component(s ) and extended envelope appear to be physically associated in the individual sources . on the other hand ,
our measurements of the h76@xmath0 line emission toward uc hii regions were likely to be dominated by their associated compact components , rather than that due to the uc hii regions themselves , since our beam size ( @xmath22@xmath3 ) is much larger than the typical size ( @xmath510@xmath15 ) of the uc hii regions . in order to investigate the physical relationship between uc hii regions and their associated compact components
, therefore , we compare our h76@xmath0 line data with the rrl and molecular line data of uc hii regions taken from the literature ( table 7 ) . in this comparison
, we do not include [email protected] , in which the h76@xmath0 line was not detected .
[email protected] and g25.72 + 0.05 are also excluded because they have no associated compact components .
figure 5 exhibits the results . on the whole ,
the central velocities of the two components are in rough agreement and there is no systematic trend between the two parameters . for 5 uc hii regions ,
our results are compared with those of high - resolution rrl observations ( andrews et al .
1985 ; churchwell et al .
1990b ; afflerbach et al .
. the central velocity of the uc hii region is equal to that of its associated compact component within 2 km s@xmath10 for [email protected] , [email protected] , and [email protected] , and within 6 km s@xmath10 for [email protected] and g23.96 + 0.15 . for the other uc hii regions ,
our h76@xmath0 lines are compared with their associated nh@xmath36 or cs ( 7@xmath46 ) lines ( churchwell et al .
1990a ; cesaroni et al .
1992 ; plume et al .
these molecular lines trace the densest part of a cloud , and their velocities should be close to those of uc hii regions ( e.g. , forster et al .
we could not find any significant ( @xmath16 km s@xmath10 ) difference between the two velocities except for g23.71 + 0.17 , for which the difference is @xmath210 km s@xmath10 .
this large velocity difference is probably due to a relative motion between the ionized and molecular gas .
table 7 also presents the average difference between the velocities of the ultracompact and compact components , @xmath37 , for each line emission .
figure 6 displays @xmath38/@xmath39 versus @xmath39/@xmath40 for 14 sources in which the uc hii region has an associated compact component . here
@xmath38 , @xmath39 , and @xmath40 are the lyman continuum photon fluxes of the ultracompact , compact , and extended components , respectively .
for @xmath38 , we used the values estimated by wc89 from flux densities at 2 cm or 6 cm .
the ratios of @xmath39/@xmath40 are greater than 65% for all the sources with a single peak except for [email protected] , which has a ratio of about 30% . on the other hand ,
@xmath38/@xmath39 is less than 35% for 12 of the 14 sources .
our values of @xmath38/@xmath39 and @xmath39/@xmath40 may be , respectively , overestimated and underestimated and so should be taken as upper limits and lower limits , since uc hii regions are optically thin at 2 cm or 6 cm but optically thick at 21 cm .
the extreme cases are [email protected] and [email protected] , for which @xmath39 is equal to or less than @xmath38 .
the uc hii region , compact component(s ) , and extended envelope in each source have approximately the same radial velocity and thus the three components are likely to be physically associated in almost all of our sources .
are they ionized by the same source or by separate sources ?
although they are physically associated , they can still be ionized by separate sources , e.g. , by stars of different ages in associations .
this is closely related to how massive stars form from molecular clouds and how they destroy the natal clouds .
first , in order to examine whether the compact components in each source are excited by the same ionizing source or not , we estimated how much lyman continuum photon flux the ionizing star in the strongest compact component can provide for nearby components in sources with two or more compact components .
here it was assumed that the strongest component is density - bounded and the star maintains the ionization of the entire hii region ( including all compact components and extended envelope ) .
the flux supplied depends on the solid angle subtended by the compact component from the ionizing star .
the derived fluxes are much less ( 0.3%@xmath430% ) than the required values in all the sources . in the case of [email protected] , for example
, the ionizing source in the eastern compact component can supply the western one with at most 15% of the required flux .
_ hence , the compact components in each source appear to be ionized by separate sources .
_ what about the compact components and their envelopes ? as stated in
4.2 , the ratios of @xmath39/@xmath40 are greater than 65% for all the sources with a single peak except for [email protected] .
the surface brightness decreases monotonically from the compact component peak to the envelope edge in these sources .
_ therefore , it seems probable that the extended envelope is entirely or at least significantly ionized by the exciting star of the compact component .
_ in the case of sources with two or more compact components , the envelope of each compact component could have been produced by uv photons escaping from the compact component in a similar manner to the sources with a single peak . the individual envelopes might have merged to form a larger envelope at later times .
however , we can not exclude absolutely the possibility that the large envelope may have been created by separate ionizing sources , such as massive stars in an earlier generation as in the sequential star formation model ( elmegreen & lada 1977 ) .
we can also ask whether the uc hii region and its associated compact component are excited by the same source or by separate sources . as noted in 3.1 , the uc hii regions correspond to the peaks of their associated compact components in all the sources and the compact components with uc hii regions are usually smaller and denser than those without uc hii regions . for three sources ,
i.e. , [email protected] , [email protected] , and [email protected] , we were able to confirm that the uc hii region is still located at the peak of radio continuum emission in available data obtained at higher resolution ( zijlstra et al .
1990 ; andrews , basart , & lamb 1985 ; fey et al .
if they are ionized by independent sources , one would not expect these correlations .
_ hence the uc h@xmath41 region and its associated compact component are likely to be excited by the same ionizing source in our samples . _
if this is true , the fractions of uv photons that escape from the uc hiiregions , @xmath42=1@xmath4@xmath38/@xmath39 , are estimated to be 65%@xmath497% for our sources .
these are probably underestimates ( see 4.2 ) the age problem of uc hii regions originates from the results of wood & churchwell ( 1989b ) , who found that uc hii regions have peculiar @xmath30 60 - 12 @xmath43 m and 25 - 12 @xmath43 m colors , log@xmath44 and log@xmath45 , and evaluated the total number of uc hii regions in the galaxy to be 1650@xmath43300 based on these color criteria .
since the @xmath30 sample is sensitivity - limited , most of them are likely to be embedded o stars .
this estimate is about 10 times greater than the number derived from the known formation rate of o stars , e.g. , 4@xmath1810@xmath46 o stars yr@xmath10 ( see garay & lizano 1999 ) , and the dynamical age ( @xmath510@xmath6 yr ) of uc hii regions .
however , almost all of the uc hii regions in our sample have extended ( @xmath1 a few 1 pc ) envelopes ionized by the same sources , which implies that their actual ages are @xmath110@xmath8 yr ( see next section ) . if most uc hii regions are associated with extended emission as in our sources , therefore , the age problem is expected to be resolved although we need to explain how structures with very different length scales can coexist in an hii region for a long time . in order to address this issue , we calculated the ratios of single - dish ( @xmath47 ) to vla fluxes ( @xmath48 ) for 30 uc hii regions with simple morphology in the wc89 catalog .
here we used the 11 cm flux densities observed with the effelsberg 100 m telescope ( fwhm@xmath494.@xmath503 ) ( f@xmath51rst et al .
most of the uc hii regions have large ratios ( fig . 7 ) .
for instance , about 70% have ratios @xmath75 or @xmath48/@[email protected] .
the derived ratios are likely to be underestimated , since the uc hii regions are quite optically thick at 11 cm . however , this may have little effect on our result .
one may suspect that the result is attributed to selection effects of the wc89 survey , because their original target list was created by selecting strong compact sources from previous single - dish surveys .
thus we also determined the flux ratios for 22 simple uc hii regions in the survey of kcw , which were chosen by the @xmath30 color criteria of wood & churchwell ( 1989b ) , and obtained a very similar result ( fig .
7 ) . for comparison , we computed the ratios for the 15 sources of kurtz et al .
( 1999 ) , which consist of 8 simple and 7 complex uc hii regions .
about 80% of the 12 sources with extended emission have ratios @xmath110 , while two of the 3 sources without extended emission have ratios @xmath315 .
_ consequently , most sources known as uc h@xmath41 regions are likely to be associated with extended emission , analogous to our sources . _ the @xmath30 color criteria of wood & churchwell ( 1989b ) were established using the uc hii regions in the wc89 catalog .
so the color criteria may select compact or extended hii regions as well as uc hii regions .
this idea is consistent with the results of codella , felli , & natale ( 1994 ) who found that more than half of 445 diffuse hii regions are related to @xmath30 point sources that satisfy the color criteria .
we have investigated whether or not there is a systematic trend between the @xmath30 colors and @xmath47/@xmath48 .
the @xmath30 colors may be expected to become bluer with increasing @xmath47/@xmath48 because the color criteria for the selection of @xmath30 point sources associated with diffuse hii regions are known to be bluer than those for uc hii regions ( hughes & macleod 1989 ) .
however , we could not find any apparent variation in the @xmath30 colors with @xmath47/@xmath48 , which implies that the @xmath30 colors are not very sensitive to the presence of extended emission .
our observations show that an uc hii region is usually surrounded by more extended emission , which appears to be excited by the same ionizing source .
various models were suggested to explain the longevity of uc hii regions , as mentioned in 1 , but none of them predicted the presence of extended emission around them .
we consider the origin of the extended envelopes of uc hii regions in this section .
molecular clouds are well known to be hierarchically clumpy at all observable length scales .
_ thus it is not surprising that h@xmath41 regions , which form in molecular clouds , have substructures as in our sources .
_ recent high - resolution studies of massive star - forming regions have revealed `` molecular clumps '' and `` hot cores '' therein ; the molecular clumps have diameters @xmath21 pc , masses @xmath710@xmath52 @xmath53 , and densities @xmath210@xmath54 @xmath12 , while the hot cores have diameters @xmath50.1 pc , masses of ( 1@xmath43)@xmath1810@xmath55 @xmath53 , densities @xmath710@xmath56 @xmath12 , and temperatures @xmath1100 k ( see kurtz et al 2000 and references therein ) .
the signposts of massive star formation , such as uc hii regions and maser sources , are commonly observed in the molecular clumps .
the hot cores are closely associated with uc hii regions .
they are believed to be the sites of massive star formation where uc hii regions have not yet developed ( e.g. , cesaroni et al .
1994 ) or remnants of the gas from which the massive stars formed ( e.g. , garay & rodrguez 1990 ) .
the sizes of the hot cores and molecular clumps agree roughly with those of uc hii regions and their associated compact components in our sample , respectively .
we can first think of the possibility that the extended envelope could be formed very soon after the central star turns on , e.g. , before the hii region starts to expand dynamically .
it may depend on the density of interclump medium and volume filling factor of dense clumps . according to earlier studies on the internal structures of the molecular clumps in massive star - forming regions ( snell et al .
1984 ; stutzki et al .
1988 ; stutzki & gesten 1990 ; plume et al . 1997 ) , the interclump medium density is generally @xmath110@xmath6 @xmath12 and the volume filling factor varies between @xmath310.1 and 0.5 and tends to increase with approaching the peak . assuming that the spectral type of ionizing star is o6 , the initial strgren radii of hii regions would be @xmath310.15 pc , which are much less than the typical size of molecular clumps . to our knowledge , there is no previous detailed study on the substructures of the hot cores .
however , both the interclump medium density and volume filling factor are expected to be greater in the hot cores , since they have much higher mean densities .
therefore , the extended envelope is likely to formed by the dynamical expansion of ionized gas sphere .
the existence of uc hii regions without extended envelopes , such as [email protected] , g25.72 + 0.05 ( 6.4 ) , and g78.4 + 2.6 ( kurtz et al . 1999 ) , also argues for this view .
we propose a simple model in which _ the existence of the extended emission around uc h@xmath41 regions can be explained by combining the champagne flow model with the hierarchical structure of massive star - forming regions _
( tenorio - tagle 1982 and references therein ) .
figure 8 is a schematic representation of our model .
here we consider the case where a massive star forms off - center within a hot core with @xmath57@xmath210@xmath58 @xmath12and @xmath59@xmath2100 @xmath53 , which is placed in a molecular clump of @xmath57@xmath210@xmath8 @xmath12 .
the initial and final equilibrium strmgren radii in this case are , respectively , about 0.0015 pc and 0.05 pc assuming @xmath27=10@xmath60 s@xmath10 for the star ( o6 ) .
a champagne flow would develop when an ionization front ( of d - type ) breaks out of the core .
the timescale that the ionization front needs to reach the core s edge depends on the depth of the exciting star inside the core .
assuming that the depth is 0.03 pc , the timescale will be @xmath210@xmath6 yr from the expansion law for hii regions in a uniform medium ( spitzer 1968 ) .
if @xmath27 is significantly less than 10@xmath60 s@xmath10 and/or @xmath57 is greater than 10@xmath58 @xmath12 , the hii region can reach pressure equilibrium without breaking through the core s boundary as in the pressure - confined model of uc hii regions ( de pree , rodrguez , @xmath13 goss 1995 ; garca - segura @xmath13 franco 1996 ) .
the supersonic gas flow drives a strong isothermal shock moving away from the core and a rarefaction wave traveling toward the exciting star .
the velocity of the champagne flow , which is determined by the pressure contrast at the edge , could increase up to 30 km s@xmath10 . according to whitworth ( 1979 ) who estimated the efficiency with which o stars disperse their parental molecular clouds ,
the erosion timescale can be calculated from the following formula @xmath61 if this result is adopted , the lifetime of the hot core would be about 3@xmath1810@xmath8 yr
. meanwhile , the hii region continues to be ultracompact within the hot core while it could grow up to a few 1 pc outside the core .
another champagne flow would develop when the ionization front crosses the edge of the molecular clump .
lyman continuum photons that escape from the clump ionize the surrounding region with a lower density , which forms a more extended hii region .
we could find the champagne flows in several sources
3.2 based on the velocity gradient of h76@xmath0 line emission and/or the morphology of radio continuum image ( see also 6 ) .
this is compatible with expectation of our model .
it is difficult to detect the champagne flows in all the sources because of the large line width and non - gaussian profile of h76@xmath0 lines as well as various viewing angles .
we may not expect to obtain direct evidence for the large velocity gradients on length scales of molecular clumps or smaller owing to the low resolution and coarse sampling of our rrl observations .
however , the larger line width of rrls in the compact components noted earlier is consistent with our model . in the light of the model presented here , among our sources [email protected] and g8.14 + 0.23 are good examples of hii regions observed at viewing angles of about 0@xmath62 and 90@xmath62 , respectively ( see 6.2 and 6.6.2 ) . for uc and compact hii regions , @xmath23 was found to be on the average proportional to d@xmath10 ( garay et al .
1993 ; garay & lizano 1999 ) .
figure 9 compares electron density with diameter for the compact and extended components of our sources .
there is a fairly good correlation between the two parameters . a least - squares fit yields @xmath23 = 630 d@xmath63 .
figure 9 also displays the relation between @xmath23 and d for more compact hii regions in the catalogs of wc89 , kcw , and garay et al .
( 1993 ) . we performed a least - squares fit to all the data points in the plot and obtained @xmath23 = 790 d@xmath64 .
if an hii region evolves with a constant flux of ionizing photons , @xmath23 would be roughly proportional to d@xmath65 . in view of this , the previous studies suggested two possible explanations for the shallower power law : either uc hii regions are excited by stars with lower luminosities than those exciting compact hii regions or a part of uc hii regions in the surveys of wc89 and kcw are externally ionized dense clumps within inhomogeneous hii regions . according to our model in 5.3 , however , the observed d@xmath4@xmath23 relationship is not a manifestation of the evolutionary effect but mainly of the variation in the ambient density ; more compact hii regions are located in denser parts of molecular clouds . in this regard , it is natural to find a d@xmath4@xmath23 relationship which is similar to the d@xmath4@xmath57 relationship of molecular clouds , e.g. , @xmath66
d@xmath67 ( e.g. , larson 1981 )
. there could be some evolutionary effects for larger hii regions , but figure 9 shows that the effect does not make a significant contribution to the relationship for hii regions with d@xmath520 pc .
[email protected] , also known as w28 a2(1 ) , is situated in w28 a2 , which is @xmath250@xmath3 south of the w28 supernova remnant ( snr ) .
this uc hii region was found to drive one of the most powerful ( @[email protected] @xmath53 km s@xmath10 yr@xmath10 ) molecular bipolar outflows ( e.g. , accord , walmsley , & churchwell 1997 ) and thus has been extensively studied . in the vla radio continuum maps with high ( @xmath311@xmath15 ) resolution , [email protected] appears as a @xmath25@xmath15 shell ( wc89 ) or a disk - like structure ( zijlstra et al .
previous estimates of the spectral type of the ionizing star range from o5 to o7 .
our radio continuum observations show that there are two compact components separated by [email protected] in this field : a large ( 1.1 pc ) central one and a small ( 0.7 pc ) western one ( fig . 1@xmath69 ) .
[email protected] is located in the western compact component .
the two compact components are embedded in low - level emission extending over @xmath70 ( or @xmath71 pc@xmath55 at 2.6 kpc ) .
the brightness distribution within the extended structure is fairly complicated . a bright ridge , originating from the central compact component , extends in the southwest direction .
the surface brightness declines outward from the central region and then increases again at the edge of the extended structure .
figures 10@xmath69 and 10@xmath72 show the distributions of the h76@xmath0 line integral and equivalent line width ( @xmath73/@xmath74 ) in this field , respectively .
the line integral peak does not correspond to the central continuum peak but to the western peak with the uc hii region .
the line width is largest ( 47 km s@xmath10 ) at the same position , as pointed out in 3.2 .
this large line width is certainly attributed to the energetic outflow .
the central velocity is slightly lower in the compact components than in the extended envelope .
[email protected] lies in the well - observed hii region m8 , the lagoon nebula .
m8 is placed in the sagittarius arm of the galaxy and contains an extremely young ( about 1.5 myr ) open cluster , ngc 6530 , in the eastern part ( sung , chun , & bessell 2000 ) .
the dense region ( @xmath210@xmath3 ) of m8 in optical emission corresponds roughly to our radio hii region .
the central part ( 30@xmath15@xmath1815@xmath15 in ns@xmath18ew ) of the dense region is known as the hourglass nebula . [email protected] and the hourglass , respectively , lie @xmath23@xmath15 to the southeast and @xmath215@xmath15 east of the o7 v star herschel 36 , which is known to be a newly formed star recently emerged from its dust cocoon ( wc89 ; woolf 1961 ) .
it was suggested that the dense region is ionized mainly by herschel 36 while the more extended nebulosity is excited by the other o stars in ngc 6530 , such as 9 sgr ( o4(f ) ) and hd 165052 ( oe ) ( lada et al .
1976 ; elliot et al .
1984 ) . based on their optical and radio observations showing that the ionized gas is blueshifted by about 7 km s@xmath10 with respect to the underlying molecular cloud , lada et al .
( 1976 ) proposed that the dense ionized region is on the near edge of the cloud , similar to the orion nebula .
stecklum et al .
( 1998 ) recently suggested that [email protected] is not a real uc hii region but a circumstellar disk surrounding a young star ( later than b5 ) , which is being photoevaporated by herschel 36 .
our radio continuum observations reveal a single compact component surrounded by extended emission of 14.@xmath75 ( or @xmath76 pc@xmath55 at 1.9 kpc ) ( fig .
1@xmath72 ) .
[email protected] and the hourglass nebula are located in the compact component .
since the surface brightness declines monotonically from the uc hii region to the edge of the extended envelope , the ionizing source of our radio hiiregion is expected to be located in the compact component .
herschel 36 may be the best candidate .
an interesting feature is the arclike structure emanating from the southeastern part of the envelope .
figures 11@xmath69 and 11@xmath72 show maps of the h76@xmath0 line integral and equivalent line width for this field , respectively .
the distribution of the line integral is similar to that of continuum emission , but the line width is extraordinarily large at the bottom of the arclike structure .
this large velocity dispersion is likely to be a consequence of the champagne flow , although the velocity difference between the arc and compact component is not large ( @xmath23 km s@xmath10 ) .
the central velocity appears to be larger by a small amount in the northeastern part than in the remaining regions .
these two uc hii regions are located in the w31 hii region / molecular cloud complex . in lower - resolution radio continuum maps ,
w31 comprises two strong components which contain [email protected] and [email protected] , respectively .
the snr [email protected] , which was identified as the soft @xmath77-ray repeater sgr 1806@xmath420 ( kulkarni & frail 1993 ) , is @xmath29@xmath3 southwest of [email protected] .
our radio continuum observations show in the field of [email protected] that there are several compact components including two strong ones and that they are embedded in extended emission of @xmath78 ( or @xmath79 pc@xmath55 at 6.0 kpc ) ( fig .
1@xmath80 ) .
the uc hii region lies in the western compact component .
the central region is extended in the east - west direction . based on their vla continuum observations at 5 ghz , woodward et al .
( 1984 ) suggested that this elongated region consists of about 20 dense ionized clumps and that the individual clumps are hii regions excited by separate embedded stars of spectral types [email protected] .
the surface brightness declines rapidly to the west , while it decreases slowly to the east . as mentioned in 3.2 , our rrl observations strongly suggest that the eastern protuberance may be a result of the champagne flow ( fig .
4 ) . in the field [email protected] ( fig 1@xmath81 )
, we can see two compact components surrounded by diffuse emission extending over @xmath82 ( or @xmath83 pc@xmath55 at 6.0 kpc ) .
the two compact components form an ionized ridge elongated in the ne - sw direction and the western one contains [email protected] .
the diffuse envelope extends straight in the nw - se direction perpendicular to the central ridge .
such a bipolar hii region can be produced by the champagne flow in the case where the ionizing source is located in a thin , flat molecular cloud ( bodenheimer , tenorio - tagle , & yorke 1979 ) . according to our molecular line observations ( kim & koo 2000 )
, the distribution of the ambient molecular gas strongly supports the champagne flow model . however
, we could not observe a significant velocity difference between the central region and the envelope .
this is presumably because the inclination angle of the extended structure is close to zero .
we will present the result of a detailed infrared and radio study on the w31 region in a forthcoming paper ( kim & koo 2000 ) .
these two spherical uc hii regions have no associated compact components .
the extended hii region associated with [email protected] consists of three compact components and a diffuse envelope elongated in the east - west direction ( fig .
1@xmath84 ) .
[email protected] is located near the edge of the extended hii region .
the velocities of the northern and western parts are lower than that of the central region by @xmath24 km s@xmath10 ( see table 5 ) .
g25.72 + 0.05 is also located on the sky near the edge of an extended hii region with a single compact component ( fig 1@xmath85 ) but the velocity of g25.72 + 0.05 is very different from that of the extended hii region as noted in 3.2 .
this strongly suggests that they are a chance coincidence in the line of sight .
nh@xmath36 ( 2,2 ) lines with almost same velocity ( @xmath86@xmath49100 km s@xmath10 ) as the h76@xmath0 lines were detected towards [email protected] and g25.72 + 0.05 ( churchwell et al .
thus these two uc hii regions are likely to be real uc hii regions .
their exciting stars were determined to be b0 stars by wc89 .
however , we can not rule out absolutely the possibility that they may be externally ionized dense clumps , especially in the case of [email protected] .
[email protected] is a prototypical cometary uc hii region and has been studied in detail at infrared and radio wavelengths .
this uc hii region appears as a @xmath25@xmath15 diameter edge - brightened arclike structure in the vla maps of wc89 . in order to explain the cometary morphology and velocity structure of this source ,
two competing models were proposed : the bow shock model ( wood & churchwell 1991 ; mac low et al .
1991 ; afflerbach et al .
1994 ) and the champagne flow model ( fey et al . 1995 ; pratap , megeath , & bergin 1999 ; lumsden & hoare 1999 ) .
recent near - infrared images show that [email protected] lies in an embedded cluster ( fey et al .
1995 ; pratap et al .
watson & hanson ( 1997 ) restricted the exciting star to spectral classes o5@xmath4o8 on the basis of its @xmath87-band spectrum .
our radio continuum observations reveal several compact components including two strong ones in this field ( fig .
1@xmath88 ) .
[email protected] exists in the northern compact component from which two protuberances stretch to the east and to the southwest , respectively .
most compact components are embedded in extended emission of @xmath89 ( or @xmath90 pc@xmath55 at 9 kpc ) .
there does not exist a significant velocity difference between the compact components and extended envelope , suggesting a physical association between them .
[email protected] , also known as w28 a1 , is located on the sky near the center of the w28 snr , which appears to be an incomplete shell with a diameter of @xmath230@xmath3 in our radio continuum map ( fig .
1@xmath91 ) .
the distance to w28 is known to be @xmath22 kpc ( kaspi et al .
1993 ) , whereas [email protected] seems to be located at a much greater distance of 16.7 kpc on the basis of h@xmath92co absorption line observations ( downes et al .
hence , the two objects are not physically associated but happen to lie along the line of sight .
[email protected] appears to be elongated in the north - south direction in figure 1@xmath91 ( see also fig . 1 of andrews et al .
the central velocity of the h76@xmath0 line emission from this source is 13.2 km s@xmath10 , which is in good agreement with that ( @xmath215 km s@xmath10 ) of andrews et al .
( 1985 ) who made high - resolution ( 8.@xmath153@xmath184.@xmath150 ) h76@xmath0 line observations using the vla .
g8.14 + 0.23 is located at the peak of an extended ( @xmath93 ) structure with a bipolar morphology extended in the north - south direction ( fig .
1@xmath94 ) .
such a radio continuum appearance can be interpreted as a result of the champagne flow originating from a thin , flat molecular cloud .
the central velocity of the southern protuberance is equal to that of the compact component within 2 km s@xmath10 , which is possible if the axis of the extended hii region is in the plane of the sky , similar to [email protected] .
since the surface brightness declines more rapidly on the eastern side than on the western side , the gas density and/or the thickness appear to decrease from the east to the west along the thin molecular cloud .
shepherd & churchwell ( 1996 ) found high - velocity molecular gas , which is suggestive of bipolar outflow , in this object .
[email protected] is embedded in extended emission with three compact components ( fig .
1@xmath95 ) .
the surface brightness declines slowly to the northeast , while it drops steeply on the other side .
[email protected] appears as a @xmath24@xmath15 cometary structure in the vla map of wc89 , and is located in the northern compact component in our radio continuum map .
the compact components have the same velocity , which is different from that of the northeastern part by @xmath23 km s@xmath10 .
high - velocity molecular gas was detected toward this object ( shepherd & churchwell 1996 ) .
we made vla radio continuum observations of 16 simple uc hii regions with large ratios of @xmath47/@xmath48 and detected extended emission towards all the sources .
the extended emission consists of one to several compact components and a diffuse envelope .
all the uc hii regions except for two spherical ones ( [email protected] and g25.72 + 0.05 ) are located in the compact components .
there is no significant velocity difference between the ultracompact , compact , and extended components in each source with one exception ( g25.72 + 0.05 ) , which strongly suggests that they are physically associated in almost all of our sources .
the uc hii regions always correspond to the peaks of their associated compact components and the compact components with uc hii regions are usually more compact than those without uc hii regions .
it is therefore likely that the associated compact components and their immediate envelopes have been produced by uv photons escaping from the uc hii regions , although we can not completely exclude the possibility that they could have been created by separate ionizing sources in a cluster or by sequentially formed massive stars .
consequently , almost all of the uc hii regions in our sample are unlikely to be ` real ' uc hii regions and their actual ages may be much greater than 10@xmath6 yr .
the two uc hii regions with no associated compact components may be ` real ' uc hii regions or externally ionized dense clumps .
we calculated the ratios of @xmath47/@xmath48 for the previously known uc hii regions with simple morphology and found that most of them have large ratios like our sources .
therefore , the age problem of uc hii regions would be significantly alleviated .
we presented a simple model in which the presence of extended emission around uc hii regions can be understood using both the champagne flow model and the hierarchical structure of massive star - forming regions . in our model
the uc hii regions and their associated compact and extended components may not represent an evolutionary sequence .
the three components could coexist for a long ( @xmath710@xmath8 yr ) time for an hii region which evolves within a hierarchically clumpy molecular cloud . in this context ,
the compact components without uc hii regions seem to be in a later evolutionary phase than those with uc hii regions . the relation @xmath9
d@xmath10 may also be a corollary of the hierarchical structure of molecular clouds .
we thank dana balser for his help with the rrl observations and data reduction .
we are very grateful to ed churchwell and stan kurtz for carefully reading the manuscript and for many helpful comments and suggestions .
k. acknowledges support from korea research foundation grant for junior researcher in 1996 .
this work has been supported by korea science and engineering foundation through grant no .
961 - 0203 - 014 - 02 , and also bk21 program , ministry of education , korea through sees .
afflerbach , a. , churchwell , e. , hofner , p. , @xmath13 kurtz , s. 1994 , apj , 437 , 697 afflerbach , a. , churchwell , e. , accord , j. m. , hofner , p. , kurtz , s. , & depree , c. g. 1996 , apjs , 106 , 423 andrews , m. d. , basart , j. p. , & lamb , r. c. 1985 , aj , 90 , 310 bodenheimer , p. , tenorio - tagle , g. , & yorke , h. w. 1979 , apj , 233 , 85 cesaroni , r. , walmsley , c. m. , & churchwell , e. 1992 , a&a , 256 , 618 cesaroni , r. , churchwell , e. , hofner , p. , & walmsley , c. m. 1994 , a&a , 288 , 903 chaisson , e. j. , lichten , s. m. , & rodrguez , l. f. 1978 , apj , 221 , 810 churchwell , e. 2000 , in the origin of stars and planetary systems , ed . c. j. lada & n. d. kylafis ( dordrecht : kluwer ) , 515 churchwell , e. , walmsley , c. m. , & cesaroni , r. 1990a , a&as , 83 ,
119 churchwell , e. , walmsley , c. m. , wood , d. o. s. , & steppe , h. 1990b , in radio recombination lines : 25 years of investigations , ed . m. a. gordon & r. l. sorochenko ( dordrecht : kluwer ) , 83 codella , c. , felli , m. , & natale , v. 1994 , a&a , 284 , 233 de pree , c. g. , rodrguez , l. f. , @xmath13 goss , w. m. 1995 , rev .
mexciana astron .
astrofis . , 31 , 39 downes , d. , wilson , t. l. , bieging , j. , @xmath13 wink , j. 1980 , a@xmath13as , 40 , 379 elmegreen , b. g. , @xmath13 lada , c. j. 1977 , apj , 214 , 725 elliot , k. h. , goudis , c. , hippelein , h. , & meaburn , j. 1984 , a&a , 138 , 451 fey , a. l. , gaume , r. a. , claussen , m. j. , @xmath13 vrba , f. j. 1995 , apj , 453 , 308 forster , j. r. , caswell , j. l. , okumura , s. k. , hasegawa , t. , & ishiguro , m. 1990 , a&a , 231 , 473 f@xmath51rst , e. , reich , w. , reich , p. , & reif , k. 1990 , a&as , 85 , 805 garay , g. , & lizano , s. 1999 , pasp , 111 , 1049 garay , g. , & rodrguez , l. 1983 , apj , 266 , 263 garay , g. , & rodrguez , l. 1990 , apj , 362 , 191 garay , g. , rodrguez , l. , moran , j. m. , @xmath13 churchwell , e. 1993 , apj , 418 , 368 garca - segura , g. , @xmath13 franco , j. 1996 , apj , 469 , 171 habing , h. j. , & israel , f. p. 1979
, ara&a , 17 , 345 kaspi , v. m. , lyne , a. g. , manchester , r. n. , johnston , s. , damico , n. , & shemar , s. l. 1993 , apj , 409 , l57 kim , k .- t . , @xmath13 koo , b .- c . 1996 , j. korean astron .
soc . , 29 , s177 kim , k .-
t . , @xmath13 koo , b .-
2000 , in preparation koo , b .-
c . , kim , k .-
, yun , m .- s . , @xmath13 ho , p. t. 1996 , apj , 456 , 662 kulkarni , s. r. , & frail , d. a. 1993 , nature , 365 , 33 kurtz , s. , cesaroni , r. , churchwell , e. , hofner , p. , & walmsley , c. m. 2000 , in protostars and planets iv , ed . v. mannings , a. boss , & s. russell ( tucson : univ .
arizona press ) , 299 kurtz , s. , churchwell , e. , @xmath13 wood , d. o. s. 1994 , apjs , 91 , 659 ( kcw ) kurtz , s. , watson , a. m. , hofner , p. , @xmath13 otte , b. 1999 , apj , 514 , 232 lada , c. j. , gull , t. r. , gottlieb , c. a. , & gottlieb , e. w. 1976 , apj , 203 , 159 lockman , f. j. 1989 , apjs , 71 , 469 lumsden , s. l. , @xmath13 hoare , m. g. 1999 , mnras , 305 , 701 mac low , m .- m . , van buren , d. , wood , d. o. s. , @xmath13 churchwell , e. 1991 , apj , 369 , 395 mezger , p. g. 1980 , in proc .
workshop on recombination lines , ed .
p. a. shaver ( dordrecht : reidel ) , 81 panagia , n. 1973 , aj , 78 , 928 panagia , n. , @xmath13 walmsley , c. m. 1978 , a@xmath13a , 70 , 411 plume , r. , jaffe , d. t. , & evans , n. j. ii 1992 , apjs , 78 , 505 plume , r. , jaffe , d. t. , evans , n. j. ii , martn - pintado , j. , & gmez - gonzlez , j 1997 , apj , 476 , 730 pratap , p. , megeath , s. t. , @xmath13 bergin , e. a. 1999 , apj , 517 , 799 rubin , r. h. 1968 , apj , 154 , 391 schraml , j. , @xmath13 mezger , p. g. 1969 , apj , 156 , 269 shaver , p. a. , mcgee , r. x. , newton , l. m. , danks , a. c. , & pottasch , s. r. 1983 , mnras , 204 , 53 shepherd , d. s. , & churchwell , e. 1996 , apj , 472 , 225 snell , r. l. , goldsmith , p. f. , erickson , n. r. , mundy , l. g. , & evans , n. j. ii 1984 , apj , 276 , 625 spitzer , l. jr .
1968 , diffuse matter in space ( new york : interscience publishers ) , 183 stecklum , b. , henning , t. , feldt , m. , hayward , t. l. , hoare , m. g. , hofner , p. , & richter , s. 1998 , aj , 115 , 767 stutzki , j. , & gesten , r. 1990 , apj , 356 , 513 stutzki , j. , stacey , g. j. , genzel , r. , harris , a. i. , jaffe , d. t. , & lugten , j. b. 1988 , apj , 332 , 379 sung , h. , chun , m .- y . , & bessell , m. s. 2000 , aj , 120 , 333 tenorio - tagle , g. 1982 , in regions of recent star formation , ed .
r. s. roger @xmath13 p.
e. dewdney ( dordrecht : kluwer ) , 1 watson , a. m. , @xmath13 hanson , m. m. 1997 , apj , 490 , l165 wink , j. e. , altenhoff , w. j. , @xmath13 mezger , p. g. 1982 , a@xmath13a , 108 , 227 wood , d. o. s. , @xmath13 churchwell , e. 1989a , apjs , 69 , 831 ( wc89 ) wood , d. o. s. , @xmath13 churchwell , e. 1989b , apj , 340 , 265 wood , d. o. s. , @xmath13 churchwell , e. 1991 , apj , 372 , 199 woodward , c. e. , helfer , h. l. , & pipher , j. l. 1984 , mnras , 209 , 209 woolf , n. j. 1961 , pasp , 73 , 206 zijlstra , a. a. , pottasch , s. r. , engels , d. , roelfsema , p. r. , hekkert , p. te - lintel , & umana , g 1990 , mnras , 246 , 217 observation dates & 1995 feb 11 , 13 , 14 , @xmath13 17 + total time & 19 hours + configuration & dnc + number of antennas & 27 + center frequency & 1.38510 @xmath13 1.46490 ghz + bandwidth & 50 mhz + largest structure visible & 15@xmath3 + synthesized beam fwhm & @xmath240@xmath15 @xmath18 20@xmath15 + primary beam fwhm & 30@xmath3 + flux calibrator source & 3c286 + phase calibrator source & 1819 - 096 + typical on - source time & 40 minutes + number of fields observed & 16 + typical rms noise & 1 mjy beam@xmath10 + rccccccc [email protected] & @xmath96 & @xmath97 & 40 & @xmath98 & 1.70 & 2.6 & shell + [email protected] & 18 00 36.4 & @xmath99 @xmath10022 54 & 35 & @xmath101 & 2.75 & 1.9 & core - halo + [email protected] & 17 57 47.4 & @xmath102 @xmath10020 30 & 35 & @xmath103 & 2.50 & 16.7 & irregular + g8.14 + 0.23 & 18 00 00.2 & @xmath104 @xmath10048 15 & 35 & @xmath105 & 0.53 & 4.2 & irregular + [email protected] & 18 06 22.5 & @xmath106 @xmath10020 05 & 40 & @xmath107 & 1.71 & 6.0 & irregular + [email protected] & 18 05 57.9 & @xmath106 @xmath10006 26 & 40 & @xmath107 & 1.62 & 6.0 & cometary + [email protected] & 18 09 43.7 & @xmath108 @xmath10025 09 & 40 & @xmath109 & 0.49 & 16.1 & cometary + [email protected] & 18 09 58.9 & @xmath108 @xmath10011 58 & 40 & @xmath110 & 0.90 & 16.7 & cometary + [email protected] & 18 32 01.2 & @xmath111 @xmath10033 33 & 45 & @xmath112 & 0.91 & 9.0 & spherical + g23.71 + 0.17 & 18 31 10.3 & @xmath111 @xmath10009 36 & 45 & @xmath112 & 0.73 & 9.0 & core - halo + g23.96 + 0.15 & 18 31 42.6 & @xmath113 @xmath10057 11 & 60 & @xmath112 & 0.60 & 6.0 & irregular + g25.72 + 0.05 & 18 35 21.6 & @xmath114 @xmath10026 27 & 55 & @xmath115 & 0.85 & 14.0 & spherical + g27.28 + 0.15 & 18 37 55.7 & @xmath116 @xmath10000 35 & 60 & @xmath117 & 0.65 & 15.2 & irregular + [email protected] & 18 43 27.1 & @xmath118 @xmath10042 36 & 50 & @xmath119 & 1.62 & 9.0 & cometary + [email protected] & 18 54 37.1 & @xmath120 @xmath10035 01 & 35 & @xmath121 & 1.37 & 12.7 & irregular + [email protected] & 18 57 46.8 & @xmath122 @xmath10058 55 & 35 & @xmath121 & 0.61 & 12.0 & core - halo rcccccc [email protected] & @xmath123 & @xmath124 & @xmath125 & @xmath126 & 0.91 & 20.86 + a ............ & 17 57 37.7 & @xmath99 @xmath10004 08 & @xmath127 & @xmath128 & 0.91 & 7.79 + b ............ & 17 57 27.3 & @xmath99 @xmath10004 13 & @xmath129 & @xmath130 & 0.32 & 1.30 + [email protected] & 18 00 36.8 & @xmath99 @xmath10022 49 & @xmath131 & @xmath132 & 3.57 & 48.29 + a ............ & 18 00 36.6 & @xmath99 @xmath10022 48 & @xmath133 & @xmath134 & 2.69 & 14.93 + [email protected] & 17 57 47.0 & @xmath102 @xmath10020
25 & @xmath135 & @xmath136 & 1.07 & 2.32 + a ............ & 17 57 47.0 & @xmath102 @xmath10020 24 & @xmath137 & @xmath138 & 1.03 & 2.06 + g8.14 + 0.23 & 18 00 00.9 & @xmath104 @xmath10048 15 & @xmath139 & @xmath140 & 2.04 & 6.67 + a ............ & 18 00 00.5 & @xmath104 @xmath10048 13 & @xmath141 & @xmath142 & 1.74 & 4.71 + [email protected] & 18 06 27.8 & @xmath106 @xmath10020 00 & @xmath143 & @xmath144 & 2.94 & 55.22 + a ............ & 18 06 27.5 & @xmath106 @xmath10019 59 & @xmath145 & @xmath146 & 2.77 & 23.20 + b ............ & 18 06 22.0 & @xmath106 @xmath10020 00 & @xmath147 & @xmath148 & 2.24 & 9.94 + [email protected] & 18 06 02.1 & @xmath106
@xmath10005 41 & @xmath149 & @xmath150 & 1.42 & 15.92 + a ............ & 18 06 02.5 & @xmath106 @xmath10005 31 & @xmath151 & @xmath152 & 1.32 & 6.63 + b ............ & 18 05 57.8 & @xmath106 @xmath10006 19 & @xmath153 & @xmath154 & 1.03 & 3.53 + [email protected] & 18 09 42.0 & @xmath108
@xmath10026 05 & @xmath155 & @xmath156 & 0.32 & 3.06 + a ............ & 18 09 42.1 & @xmath108 @xmath10025 59 & @xmath157 & @xmath158 & 0.32 & 0.67 + b ............ & 18 09 47.7 & @xmath108
@xmath10026 33 & @xmath159 & @xmath160 & 0.24 & 1.27 + c ............ & 18 09 44.0 & @xmath108 @xmath10025 15 & @xmath161 & @xmath162 & 0.18 & 0.46 + [email protected] & 18 09 58.9 & @xmath108 @xmath10011 58 & @xmath163 & @xmath164 & 0.14 & 0.69 + a ............ & 18 09 58.7 & @xmath108 @xmath10011 56 & @xmath165 & @xmath166 & 0.11 & 0.45 + [email protected] & 18 32 00.5 & @xmath111 @xmath10034 43 & @xmath167 & @xmath168 & 0.41 & 11.31 + a ............ & 18 32 00.7 & @xmath111 @xmath10034 48 & @xmath169 & @xmath170 & 0.37 & 3.47 + b ............ & 18 32 01.1 & @xmath111 @xmath10036 10 & @xmath171 & @xmath172 & 0.28 & 6.80 + g23.71 + 0.17 & 18 31 10.0 & @xmath111 @xmath10009 41 & @xmath173 & @xmath174 & 0.46 & 1.93 + a ............ & 18 31 10.0 & @xmath111 @xmath10009 50 & @xmath175 & @xmath176 & 0.44 & 1.85 + g23.96 + 0.15 & 18 31 42.2 & @xmath113 @xmath10057 11 & @xmath177 & @xmath178 & 1.08 & 2.13 + a ............ & 18 31 42.1 & @xmath113 @xmath10057 10 & @xmath179 & @xmath180 & 1.02 & 1.71 + g25.72 + 0.05 & 18 35 22.9 & @xmath114 @xmath10028 12 & @xmath181 & @xmath182 & 0.18 & 1.74 + a ............ & 18 35 22.3 & @xmath114 @xmath10028 04 & @xmath183 & @xmath184 & 0.16 & 1.43 + g27.28 + 0.15 & 18 37 55.7 & @xmath116 @xmath10000 35 & @xmath185 & @xmath186 & 0.44 & 0.96 + a ............ & 18 37 55.7 & @xmath116 @xmath10000 35 & @xmath187 & @xmath188 & 0.40 & 0.74 + [email protected] & 18 43 27.4 & @xmath118 @xmath10042 36 & @xmath189 & @xmath90 & 1.50 & 12.69 + a ............ & 18 43 27.4 & @xmath118 @xmath10042 34 & @xmath190 & @xmath191 & 1.36 & 2.61 + b ............ & 18 43 32.6 & @xmath118 @xmath10044 57 & @xmath192 & @xmath193 & 0.61 & 6.08 + [email protected] & 18 54 31.8 & @xmath120 @xmath10034 26 & @xmath194 & @xmath195 & 0.19 & 1.64 + a ............ & 18 54 37.0 & @xmath120 @xmath10034 56 & @xmath196 & @xmath197 & 0.17 & 0.27 + b ............ & 18 54 31.9 & @xmath120 @xmath10034 44 & @xmath198 & @xmath199 & 0.14 & 1.37 + [email protected] & 18 57 46.8 & @xmath122 @xmath10058 55 & @xmath200 & @xmath201 & 0.45 & 1.12 + a ............ & 18 57 46.8 & @xmath122 @xmath10058 53 & @xmath202 & @xmath203 & 0.39 & 0.96 + rccccclcc [email protected] & 64 & 3.5 & 69.1 & 480 & 49.08 & o6 ( o7 ) & 95 & 8500 + a ............ & 473 & 36.9 & 50.0 & 25 & 48.66 & o7 & & + b ............ & 391 & 15.7 & 27.4 & 5 & 47.88 & o9.5 & & + [email protected] & 98 & 6.6 & 73.3 & 380 & 49.20 & o6 ( b0 ) & @xmath2040 & 7700 + a ............ & 1036 & 103.9 & 49.6 & 11 & 48.69 & o7 & & + [email protected] & 66 & 6.0 & 111.7 & 1970 & 49.80 & o4.5 ( o8.5 ) & 34 & 6700 + a ............ & 443 & 72.5 & 107.3 & 262 & 49.75 & o4.5 & & + g8.14 + 0.23 & 111 & 6.6 & 62.0 & 200 & 49.10 & o6 ( b0 ) & @xmath2040 & 5600 + a ............ & 917 & 98.3 & 55.2 & 17 & 48.95 & o6.5 & & + [email protected] & 99 & 14.5 & 158.7 & 3800 & 50.33 & @xmath7o4 ( o8 ) & 59 & 5500 + a ............ & 555 & 108.3 & 118.9 & 285 & 49.95 & o4 & & + b ............ & 726 & 116.8 & 89.6 & 93 & 49.58 & o5 & & + [email protected] & 64 & 5.5 & 107.5 & 1820 & 49.74 & o4.5 ( o8.5 ) & 73 & 6800 + a ............ & 505 & 64.5 & 80.2 & 96 & 49.36 & o5.5 & & + b ............ & 540 & 57.2 & 65.0 & 48 & 49.09 & o6 & & + [email protected] & 46 & 3.9 & 120.2 & 3550 & 49.88 & o4 ( o6.5 ) & 88 & 7000 + a ............ & 261 & 24.2 & 72.4 & 137 & 49.22 & o5.5 & & + b ............ & 136 & 11.9 & 89.6 & 518 & 49.50 & o5 & & + c ............ & 160 & 11.0 & 63.9 & 153 & 49.05 & o6 & & + [email protected] & 44 & 2.4 & 78.2 & 1030 & 49.19 & o6 ( o9 ) & @xmath2040 & 10000 + a ............ & 99 & 6.2 & 68.0 & 297 & 49.00 & o6 & & + [email protected] & 49 & 4.5 & 124.6 & 3700 & 49.96 & o4 ( b0 ) & 49 & 6300 + a ............ & 191 & 18.4 & 84.0 & 292 & 49.45 & o5.5 & & + b ............ & 126 & 13.3 & 105.1 & 864 & 49.74 & o5 & & + g23.71 + 0.17 & 112 & 7.6 & 70.1 & 290 & 49.17 & o6 ( o9 ) & 38 & 7100 + a ............ & 341 & 32.8 & 69.0 & 91 & 49.15 & o6 & & + g23.96 + 0.15 & 145 & 8.1 & 53.1 & 100 & 48.93 & o6.5 ( o9.5 ) & 42 & 5100 + a ............ & 1299 & 140.0 & 49.4 & 9 & 48.84 & o6.5 & & + g25.72 + 0.05 & 63 & 4.5 & 90.2 & 1100 & 49.52 & o5 ( b0 ) & 35 & 6700 + a ............ & 106 & 8.5 & 84.5 & 534 & 49.43 & o5.5 & & + g27.28 + 0.15 & 82 & 5.7 & 79.5 & 570 & 49.30 & o5.5 ( o9.5 ) & 48 & 7700 + a ............ & 418 & 45.5 & 72.9 & 87 & 49.19 & o6 & & + [email protected] & 71 & 7.5 & 128.2 & 2780 & 50.03 & o4 ( o5.5 ) & 85 & 5800 + a ............ & 1022 & 155.7 & 75.7 & 40 & 49.34 & o5.5 & & + b ............ & 241 & 30.1 & 100.3 & 393 & 49.71 & o5 & & + [email protected] & 50 & 3.1 & 84.8 & 1160 & 49.37 & o5.5 ( o9 ) & 83 & 8100 + a ............ & 356 & 23.5 & 46.7 & 27 & 48.59 & o7 & & + b ............ & 112 & 8.7 & 79.9 & 426 & 49.29 & o5.5 & & + [email protected] & 96 & 6.1 & 69.5 & 330 & 49.21 & o6 ( o7.5 ) & 52 & 6100 + a ............ & 376 & 35.9 & 66.0 & 72 & 49.15 & o6 & & + rccccc [email protected] & @xmath96 @xmath97 & & & & + & ( @xmath205 , @xmath100@xmath206 ) & [email protected] & [email protected] & [email protected] & 5484 + & ( @xmath413.4 , 00 00 ) & [email protected] & [email protected] & [email protected] & 771 + & ( @xmath405.4 , 00 00 ) & [email protected] & [email protected] & [email protected] & 684 + & ( + 02.6 , 00 00 ) & [email protected] & [email protected] & [email protected] & 5436 + & ( + 10.6 , @xmath404 00 ) & [email protected] & [email protected] & [email protected] & 757 + & ( + 10.6 , @xmath402 00 ) & [email protected] & [email protected] & [email protected] & 1253 + & ( + 10.6 , 00 00 ) & [email protected] & [email protected] & [email protected] & 3955 + & ( + 10.6 , + 02 00 ) & [email protected] & [email protected] & [email protected] & 1404 + & ( + 10.6 , + 04 00 ) & [email protected] & [email protected] & [email protected] & 709 + & ( + 18.6 , 00 00 ) & [email protected] & [email protected] & [email protected] & 3594 + & ( + 26.6 , 00 00 ) & [email protected] & [email protected] & [email protected] & 1083 + & ( + 34.6 , 00 00 ) & [email protected] & [email protected] & [email protected] & 739 + & ( + 42.6 , 00 00 ) & & @xmath207 & & + [email protected] & 18 @xmath10000 36.4 @xmath208 @xmath10022 54 & & & & + & ( 00.0 , @xmath10000 00 ) & [email protected] & [email protected] & [email protected] & 7856 + & ( @xmath424.0 , 00 00 ) & [email protected] & [email protected] & [email protected] & 789 + & ( @xmath416.0 , 00 00 ) & [email protected] & [email protected] & [email protected] & 1568 + & ( @xmath408.0 , 00 00 ) & [email protected] & [email protected] & [email protected] & 2976 + & ( 00.0 , @xmath406 00 ) & [email protected] & [email protected] & [email protected] & 874 + & ( 00.0 , @xmath404 00 ) & [email protected] & [email protected] & [email protected] & 1236 + & ( 00.0 , @xmath402 00 ) & [email protected] & [email protected] & [email protected] & 3183 + & ( 00.0 , + 02 00 ) & [email protected] & [email protected] & [email protected] & 4906 + & ( 00.0 , + 04 00 ) & [email protected] & [email protected] & [email protected] & 1827 + & ( 00.0 , + 06 00 ) & @[email protected] & [email protected] & [email protected] & 1244 + & ( + 08.0 , 00 00 ) & [email protected] & [email protected] & [email protected] & 3420 + & ( + 16.0 , 00 00 ) & [email protected] & [email protected] & [email protected] & 2140 + & ( + 24.0 , 00 00 ) & [email protected] & [email protected] & [email protected] & 1492 + & ( + 32.0 , 00 00 ) & [email protected] & [email protected] & [email protected] & 865 + & ( + 40.0 , 00 00 ) & [email protected] & [email protected] & [email protected] & 1456 + & ( + 08.0 , @xmath404 00 ) & [email protected] & [email protected] & [email protected] & 1182 + & ( + 16.0 , @xmath404 00 ) & [email protected] & [email protected] & [email protected] & 2002 + & ( + 24.0 , @xmath404 00 ) & [email protected] & [email protected] & [email protected] & 1669 + & ( + 32.0 , @xmath404 00 ) & [email protected] & [email protected] & [email protected] & 1341 + & ( + 40.0 , @xmath404 00 ) & [email protected] & [email protected] & [email protected] & 976 + [email protected] & 17 @xmath10057 47.4 @xmath210 @xmath10020 30 & & & & + & ( 00.0 , @xmath10000 00 ) & [email protected] & [email protected] & [email protected] & 2025 + g8.14 + 0.23 & 18 @xmath10000 00.2 @xmath211 @xmath10048 15 & & & & + & ( 00.0 , @xmath10000 00 ) & [email protected] & [email protected] & [email protected] & 1700 + & ( + 06.4 , @xmath401 10 ) & [email protected] & [email protected] & [email protected] & 405 + [email protected] & 18 @xmath10006 22.5 @xmath212 @xmath10020 05 & & & & + & ( 00.0 , @xmath10000 00 ) & [email protected] & [email protected] & [email protected] & 5170 + & ( @xmath410.8 , 00 00 ) & [email protected] & [email protected] & [email protected] & 2193 + & ( @xmath405.8 , + 02 00 ) & [email protected] & [email protected] & [email protected] & 7473 + & ( @xmath402.8 , @xmath402 00 ) & [email protected] & [email protected] & [email protected] & 344 + & ( @xmath402.8 , 00 00 ) & [email protected] & [email protected] & [email protected] & 8819 + & ( 00.0 , + 05 00 ) & [email protected] & [email protected] & [email protected] & 859 + & ( + 03.8 , + 02 25 ) & [email protected] & [email protected] & [email protected] & @xmath21310598 & ( + 05.2 , @xmath404 00 ) & & @xmath2148 & & + & ( + 05.2 , @xmath402 00 ) & [email protected] & [email protected] & [email protected] & 1517 + & ( + 05.2 , + 00 10 ) & [email protected] & [email protected] & [email protected] & @xmath21312226 & ( + 06.3 , @xmath405 40 ) & [email protected] & [email protected] & [email protected] & 806 & ( + 13.2 , @xmath402 00 ) & [email protected] & [email protected] & [email protected] & 2419 + & ( + 13.8 , 00 00 ) & [email protected] & [email protected] & [email protected] & 6712 + & ( + 22.7 , @xmath400 10 ) & [email protected] & [email protected] & [email protected] & 840 + & ( + 30.4 , 00 00 ) & [email protected] & [email protected] & [email protected] & 567 + [email protected] & 18 @xmath10005 57.9 @xmath212 @xmath10006 26 & & & & + & ( 00.0 , @xmath10000 00 ) & [email protected] & [email protected] & [email protected] & 1479 + & ( @xmath406.8 , + 02 10 ) & [email protected] & [email protected] & [email protected] & 833 + & ( + 03.5 , @xmath401 55 ) & [email protected] & [email protected] & [email protected] & 1006 + & ( + 04.2 , + 00 50 ) & [email protected] & [email protected] & [email protected] & 3772 + [email protected] & 18 @xmath10009 43.7 @xmath215 @xmath10025 09 & & & & + & ( @xmath401.0 , @xmath400 20 ) & [email protected] & [email protected] & [email protected] & 787 + & ( + 04.2 , @xmath401 25 ) & [email protected] & [email protected] & [email protected] & 960 + & ( + 05.6 , + 00 45 ) & [email protected] & [email protected] & [email protected] & 545 + [email protected] & 18 @xmath10009 58.9 @xmath215 @xmath10011 58 & & & & + & ( 00.0 , @xmath10000 00 ) & & @xmath2147 & & + [email protected] & 18 @xmath10032 01.2 @xmath216 @xmath10033 33 & & & & + & ( 00.0 , @xmath10000 00 ) & [email protected] & [email protected] & [email protected] & 912 + & ( @xmath410.8 , @xmath402 00 ) & [email protected] & [email protected] & [email protected] & 1166 + & ( @xmath404.7 , + 00 30 ) & [email protected] & [email protected] & [email protected] & 852 + & ( @xmath401.3 , @xmath403 00 ) & [email protected] & [email protected] & [email protected] & 1224 + & ( @xmath400.8 , @xmath402 00 ) & [email protected] & [email protected] & [email protected] & 3293 + & ( 00.0 , @xmath401 10 ) & [email protected] & [email protected] & [email protected] & 2508 + & ( + 09.1 , @xmath402 00 ) & [email protected] & [email protected] & [email protected] & 984 + & ( + 27.0 , @xmath402 05 ) & & @xmath2146 & & + g23.71 + 0.17 & 18 @xmath10031 10.3 @xmath216 @xmath10009 36 & & & & + & ( 00.0 , @xmath10000 00 ) & [email protected] & [email protected] & [email protected] & 1292 + g23.96 + 0.15 & 18 @xmath10031 42.6 @xmath407 @xmath10057 11 & & & & + & ( 00.0 , @xmath10000 00 ) & [email protected] & [email protected] & [email protected] & 1295 + g25.72 + 0.05 & 18 @xmath10035 21.6 , @xmath217 @xmath10026 27 & & & + & ( 00.0 , @xmath10000 00 ) & [email protected] & [email protected] & [email protected] & 405 + & ( + 01.3 , @xmath401 45 ) & [email protected] & [email protected] & [email protected] & 948 + g27.28 + 0.15 & 18 @xmath10037 55.7 @xmath218 @xmath10000 35 & & & & + & ( 00.0 , @xmath10000 00 ) & [email protected] & [email protected] & [email protected] & 472 + [email protected] & 18 @xmath10043 27.1 @xmath219 @xmath10042 36 & & & & + & ( 00.0 , @xmath10000 00 ) & [email protected] & [email protected] & [email protected] & 3730 + & ( @xmath408.0 , @xmath403 05 ) & [email protected] & [email protected] & [email protected] & 1667 + & ( @xmath401.7 , @xmath406 15 ) & [email protected] & [email protected] & [email protected] & 1122 + & ( @xmath400.6 , @xmath404 45 ) & [email protected] & [email protected] & [email protected] & 1388 + & ( @xmath400.6 , @xmath403 15 ) & [email protected] & [email protected] & [email protected] & 3714 + & ( @xmath400.6 , @xmath401 37 ) & [email protected] & [email protected] & [email protected] & 2196 + & ( + 05.6 , @xmath401 55 ) & [email protected] & [email protected] & [email protected] & 3646 + & ( + 07.6 , @xmath403 10 ) & [email protected] & [email protected] & [email protected] & 3346 + [email protected] & 18 @xmath10054 37.1 @xmath220 @xmath10035 01 & & & & + & ( 00.0 , @xmath10000 00 ) & [email protected] & [email protected] & [email protected] & 470 + & ( @xmath408.0 , + 00 55 ) & [email protected] & [email protected] & [email protected] & 339 + & ( @xmath405.7 , @xmath400 25 ) & [email protected] & [email protected] & [email protected] & 388 + & ( @xmath402.9 , @xmath400 13 ) & [email protected] & [email protected] & [email protected] & 600 + [email protected] & 18 @xmath10057 46.8 , @xmath221 @xmath10058 55 & & & & + & ( 00.0 , @xmath10000 00 ) & [email protected] & [email protected] & [email protected] & 858 + rcccccc [email protected] & ( + @xmath222 , @xmath206 ) & [email protected] & [email protected] & [email protected] & 81 & 0.020 + [email protected] & ( 00.0 , @xmath10000 00 ) & [email protected] & [email protected] & [email protected] & 275 & 0.035 + [email protected] & ( 00.0 , @xmath10000 00 ) & [email protected] & [email protected] & [email protected] & 179 & 0.035 + & ( + 03.8 , + 02 25 ) & [email protected] & [email protected] & [email protected] & 170 & 0.016 & ( + 05.2 , + 00 10 ) & [email protected] & [email protected] & [email protected] & 593 & 0.049 [email protected] & ( + 04.2 , + 00 50 ) & [email protected] & [email protected] & [email protected] & 237 & 0.063 + [email protected] & ( @xmath400.8 , @xmath402 00 ) & [email protected] & [email protected] & [email protected] & 262 & 0.080 + [email protected] & ( 00.0 , @xmath10000 00 ) & [email protected] & [email protected] & [email protected] & 254 & 0.068 + & ( @xmath400.6 , @xmath403 15 ) & [email protected] & [email protected] & [email protected] & 283 & 0.076 + & ( + 05.6 , @xmath401 55 ) & [email protected] & [email protected] & [email protected] & 250 & 0.069 + ccccccccl [email protected] & 9.9 & 54.8 & & 4.9 & 9.8 & 9.3 & 11.0 & 1 , 5 , 6 , 7 + [email protected] & 4.5 & 26.8 & & & & 9.6 & & 6 + [email protected] & 13.2 & 26.5 & & @xmath215 & & 12.8 & & 2 , 6 + g8.14 + 0.23 & 20.3 & 27.3 & & & 18.6 & 19.3 & 20.3 & 5 , 6 + [email protected] & 16.4 & 29.7 & & & & 9.9 & & 6 + [email protected] & 7.7 & 28.5 & & & & 13.2 & 12.7 & 6 , 7 + [email protected] & 28.9 & 22.2 & & 27.6 & 23.7 & 24.3 & 24.7 & 3 , 5 , 6 , 7 + [email protected] & & & & & & 20.8 & & 6 + g23.71 + 0.17 & 103.0 & 30.3 & & & & 113.0 & 114.6 & 6 , 7 + g23.96 + 0.15 & 78.9 & 28.0 & & 73.7 & 80.4 & 79.7 & & 1 , 5 , 6 + g27.28 + 0.15 & 35.0 & 23.9 & & & & 31.5 & & 6 + [email protected] & 95.9 & 28.6 & & 95.6 & 97.7 & 97.6 & 98.1 & 4 , 5 , 6 , 7 + [email protected] & 51.2 & 17.9 & & & & 49.4 & & 6 + [email protected] & 51.5 & 30.3 & & & & @xmath252 & & 6 + @xmath223 & & & & @[email protected] & @[email protected] & [email protected] & [email protected] & + | we have carried out 21 cm radio continuum observations of 16 uc hii regions using the vla ( d - array ) in a search for associated extended emission .
we have also observed h76@xmath0 recombination line towards all the sources and he76@xmath0 line at the positions with strong h76@xmath0 line emission .
the uc hii regions have simple morphologies and large ( @xmath110 ) ratios of single - dish to vla fluxes .
we detected extended emission towards all the sources .
the extended emission consists of one to several compact ( @xmath21@xmath3 or 0.5@xmath45 pc ) components and a diffuse extended ( 2@xmath3@xmath412@xmath3 or 4@xmath419 pc ) envelope .
all the uc hii regions but two are located in the compact components , where the uc hii regions always correspond to their peaks .
the compact components with uc hii regions are usually smaller and denser than those without uc hii regions . for individual sources ,
we derive the spectral types ( o7@xmath4o4 ) of the ionizing stars and the fractions of uv photons absorbed by dust within the nebulae , which are significantly different from previous estimates based on the uc hii regions alone .
our recombination line observations show that the ultracompact , compact , and extended components have approximately the same velocity in the individual sources with one exception ( g25.72 + 0.05 ) , implying that they are physically associated .
the compact components in each object appear to ionized by separate sources , while the uc hii regions and their associated compact components are likely to be ionized by the same sources on the basis of the morphological relations mentioned above .
this suggests that almost all of the observed uc hii regions are not ` real ' uc hii regions and that their actual ages are much greater than their dynamical age ( @xmath510@xmath6 yr ) .
we find that most of simple uc hii regions previously known have large ratios of single - dish to vla fluxes , similar to our sources .
therefore , the ` age problem ' of uc hii regions does not seem to be as serious as earlier studies argued .
we present a simple model in which the coexistence of the ultracompact , compact , and extended components for a long ( @xmath710@xmath8 yr ) time is easily explained by combining the champagne flow model with the hierarchical structure of massive star - forming regions .
the well - known relation between the density and diameter of hii regions , @xmath9 d@xmath10 , is a natural consequence of the hierarchical structure according to our model .
we discuss some individual sources . |
while we are preparing the manuscript , the paper @xcite showed up on arxiv , which has some overlap with this letter .
see also refs .
@xcite and @xcite for recent related works .
this work was supported by jsps kakenhi grant no .
22244021 ( k.h .
, t.m . ) , no .
22540263 ( t.m . ) , no .
23104008 ( t.m . ) and also by world premier international research center initiative ( wpi initiative ) , mext , japan . the work of t.t .
was supported by the program for leading graduate schools , mext , japan .
99 p. a. r. ade _ et al .
_ [ bicep2 collaboration ] , arxiv:1403.3985 [ astro-ph.co ] . a. a. starobinsky , phys . lett .
b * 91 * ( 1980 ) 99 .
a. a. starobinsky , sov .
* 9 * ( 1983 ) 302 ; v. f. mukhanov and g. v. chibisov , jetp lett .
* 33 * ( 1981 ) 532 [ pisma zh .
fiz . * 33 * ( 1981 ) 549 ] . s. ferrara and p. van nieuwenhuizen , phys .
b * 74 * ( 1978 ) 333 .
k. s. stelle and p. c. west , phys .
b * 74 * ( 1978 ) 330 .
e. s. fradkin and m. a. vasiliev , lett .
nuovo cim .
* 22 * ( 1978 ) 651 .
m. f. sohnius and p. c. west , phys .
b * 105 * ( 1981 ) 353 .
s. cecotti , phys .
b * 190 * ( 1987 ) 86 .
s. ferrara , r. kallosh and a. van proeyen , jhep * 1311 * ( 2013 ) 134 [ arxiv:1309.4052 [ hep - th ] ] . s. v. ketov and t. terada , jhep * 1312 * ( 2013 ) 040 [ arxiv:1309.7494 [ hep - th ] ] .
s. cecotti , s. ferrara , m. porrati and s. sabharwal , nucl .
b * 306 * ( 1988 ) 160 .
f. farakos , a. kehagias and a. riotto , nucl .
b * 876 * ( 2013 ) 187 [ arxiv:1307.1137 ] .
j. ellis and n. e. mavromatos , phys .
d * 88 * ( 2013 ) 085029 [ arxiv:1308.1906 [ hep - th ] ] .
r. kallosh and a. linde , jcap * 1306 * ( 2013 ) 028 [ arxiv:1306.3214 [ hep - th ] ] .
j. ellis , d. v. nanopoulos and k. a. olive , phys .
* 111 * ( 2013 ) 111301 [ arxiv:1305.1247 [ hep - th ] ] ; j. ellis , d. v. nanopoulos and k. a. olive , jcap * 1310 * ( 2013 ) 009 [ arxiv:1307.3537 ] ; j. ellis , d. v. nanopoulos and k. a. olive , phys .
d * 89 * ( 2014 ) 043502 [ arxiv:1310.4770 [ hep - ph ] ] .
s. ferrara , r. kallosh , a. linde and m. porrati , arxiv:1307.7696 [ hep - th ] ; s. ferrara , r. kallosh , a. linde and m. porrati , jcap * 1311 * ( 2013 ) 046 [ arxiv:1309.1085 [ hep - th ] , arxiv:1309.1085 ] .
p. fre and a. s. sorin , arxiv:1308.2332 [ hep - th ] ; p. fre and a. s. sorin , arxiv:1310.5278 [ hep - th ] ; s. ferrara , p. fre and a. s. sorin , arxiv:1311.5059 [ hep - th ] ; s. ferrara , p. fre and a. s. sorin , arxiv:1401.1201 [ hep - th ] .
w. buchmller , v. domcke and k. schmitz , jcap * 1304 * ( 2013 ) 019 [ arxiv:1210.4105 [ hep - ph ] ] ; w. buchmller , v. domcke and k. kamada , phys .
b * 726 * ( 2013 ) 467 [ arxiv:1306.3471 [ hep - th ] ] .
d. roest , m. scalisi and i. zavala , jcap * 1311 * ( 2013 ) 007 [ arxiv:1307.4343 ] .
s. cecotti and r. kallosh , arxiv:1403.2932 [ hep - th ] .
r. kallosh , a. linde and d. roest , jhep * 1311 * ( 2013 ) 198 [ arxiv:1311.0472 [ hep - th ] ] .
i. antoniadis , e. dudas , s. ferrara and a. sagnotti , arxiv:1403.3269 [ hep - th ] .
s. v. ketov , prog .
phys . * 2013 * 123b04 [ arxiv:1309.0293 [ hep - th ] ] . j. wess and j. bagger , princeton , usa : princeton univ . press .
( 1992 ) 259 p. s. ferrara , m. t. grisaru and p. van nieuwenhuizen , nucl .
b * 138 * ( 1978 ) 430 .
h. m. lee , jcap * 1008 * ( 2010 ) 003 [ arxiv:1005.2735 [ hep - ph ] ] .
s. ferrara , r. kallosh , a. linde , a. marrani and a. van proeyen , phys .
d * 83 * ( 2011 ) 025008 [ arxiv:1008.2942 [ hep - th ] ] . | motivated by the recent observation of the @xmath0-mode signal in the cosmic microwave background by bicep2 , we study the starobinsky - type inflation model in the framework of old - minimal supergravity , where the inflaton field in the original ( non - supersymmetric ) starobinsky inflation model is promoted to a complex field .
we study how the inflaton evolves on the two - dimensional field space , varying the initial conditions .
we show that ( i ) one of the scalar fields has a very steep potential once the trajectory is off from that of the original starobinsky inflation , and that ( ii ) the @xmath0-mode signal observed by bicep2 is too large to be consistent with the prediction of the model irrespective of the initial conditions .
thus , the bicep2 result strongly disfavors the complexified starobinsky inflation in supergravity . ut-14 - 12 + ipmu14 - 0090 1.5 in * * complexified starobinsky inflation in supergravity + in the light of recent bicep2 result .5 in koichi hamaguchi@xmath1 , takeo moroi@xmath1 , takahiro terada@xmath2 , 0.25 in _ @xmath3department of physics , the university of tokyo , tokyo 113 - 0033 , japan + @xmath4kavli institute for the physics and mathematics of the universe ( kavli ipmu ) , + wpi , todias , the university of tokyo , chiba 277 - 8568 , japan _
.5 in recently , @xmath0-mode polarization of cosmic microwave background ( cmb ) has been observed by bicep2 , which indicated a large tensor - to - scalar ratio of @xcite @xmath5 the observation of bicep2 provides a significant constraint on inflationary models because the value of @xmath6 is directly related to the scale of inflation ( i.e. , the expansion rate during inflation ) . in particular , the bicep2 result strongly disfavors one of the interesting possibilities , i.e. , starobinsky inflation model @xcite which utilizes a scalar degree of freedom in the gravitational sector as an inflaton .
this is because the starobinsky inflation predicts @xmath6 of the order of @xmath7 , which is significantly smaller than the bicep2 result .
if one extends the model , this conclusion may change .
the extension we consider in the present study is to supersymmetrize the model because supersymmetry is a prominent candidate of the physics beyond the standard model .
in such a model , the inflation can still be realized solely by the gravitational sector , while new scalar degrees of freedom are automatically introduced , which may affect the dynamics of inflation .
the starobinsky model is based on a modified theory of gravity , so we need to consider a modified theory of supergravity .
there are two minimal off - shell formulations of supergravity : the old - minimal @xcite and the new - minimal @xcite supergravity .
supergravity embedding of starobinsky model has been studied both in the old - minimal @xcite and the new - minimal @xcite supergravities .
these studies share the original philosophy of the starobinsky model in the sense that the supergravity generalizations of the model rely solely on ( super)geometrical or ( super)gravitational quantities .
the old - minimal realization of starobinsky model is possible with generic `` khler potential '' and `` superpotential '' of scalar curvature supermultiplet with extra propagating scalar degrees of freedom other than the inflaton ( also called scalaron ) @xcite . imposing a constraint @xmath8
, one can construct the old - minimal higher - curvature supergravity with only one ( pseudo)scalar in addition to the scalaron @xcite . even in this case , the discussion after eq .
( [ eq : retimt ] ) holds . on the contrary
, the new - minimal realization has a higgsed ( massive ) vector field as well as the inflaton @xcite .
thus , we consider the old - minimal supergravity because it automatically introduces new scalar degrees of freedom . in this letter , we study the starobinsky - type inflation model in the framework of old - minimal supergravity .
we pay particular attention to the fact that there exist two scalar degrees of freedom originating from the gravity multiplet in such a model .
we study the evolution of the inflaton on the two - dimensional field space .
we will see that the potential of one of the scalar fields becomes very steep once the trajectory is off from that of the original starobinsky inflation .
we also show that the tensor - to - scalar ratio in the supergravity starobinsky model is too small to be consistent with the bicep2 result even though the field space is enlarged .
the generic action of the old - minimal supergravity @xcite is , in chiral curved superspace language , , where @xmath9 is the reduced planck scale . ]
@xmath10 + \text{h.c.}\end{aligned}\ ] ] where @xmath11 and @xmath12 are the hermitian and the holomorphic functions of the scalar curvature chiral superfield @xmath13 , respectively .
the superfield @xmath13 contains ricci scalar curvature @xmath14 in its @xmath15 component and gravitino in its @xmath16 and @xmath15 components .
it also contains a complex scalar @xmath17 and real vector @xmath18 . these are auxiliary fields in the case of the minimal action with @xmath19 and @xmath20 . for generic functions @xmath21 and @xmath22 , however , these become dynamical .
the theory is classically equivalent @xcite to the standard matter - coupled supergravity @xcite @xmath23 + \text{h.c.}\end{aligned}\ ] ] with the following no - scale type khler potential and superpotential : @xmath24 linearized analysis of the original picture ( higher - curvature supergravity ) for a simple function @xmath11 has been performed in ref .
@xcite .
bosonic lagrangian of the original picture and comparisons of both pictures are described in ref .
@xcite .
note that _ any _ @xmath21 and @xmath22 functions lead to the unique khler and superpotentials for @xmath25 because the origin of @xmath25 is a lagrange multiplier .
in particular , canonically normalized field @xmath26 along the real axis ( @xmath27 ) has the starobinsky potential ( cf .
eq .
( [ eq : lxy ] ) ) . roughly speaking , @xmath28 , @xmath29 , @xmath30 , and @xmath31 in this picture
correspond to @xmath14 , @xmath32 , @xmath17 , and @xmath33 in the original geometrical picture , respectively .
in this letter , we focus on the standard matter - coupled supergravity picture . consider a khler potential for @xmath30 , @xmath34 the first term ( constant )
is needed to reproduce einstein supergravity .
the second term leads to the kinetic term of the new degrees of freedom .
however , this term produces the scalar potential unbounded below in the region of large @xmath35 .
instability for radial @xmath35 direction is stabilized by the third term proportional to @xmath36 ( see _ e_.g
. refs .
@xcite and references therein ) .
small @xmath36 makes other local minima near the original minimum ( @xmath37 ) .
because of these reasons , we take a sufficiently large value of @xmath36 .
note that , for sufficiently large @xmath36 , @xmath30 is stabilized for any value of @xmath25 .
we also assume @xmath38 so that the potential value at the vacuum is zero .
thus , @xmath30 is set to the minimum @xmath39 , and the resultant effective theory has two fields @xmath28 and @xmath29 with only one parameter @xmath40 .
after stabilization of @xmath30 , the lagrangian density is given by @xmath41 canonical normalization of both fields at the same time is impossible in this case .
we find it useful to define the semi - canonical basis that does not have kinetic mixing and realizes canonical normalization at the vacuum ( @xmath42):alternatively , one may transform @xmath29 into canonically normalized form by @xmath43 .
then the potential for @xmath44 is also simplified , @xmath45 , where @xmath46 is the starobinsky potential . however , in this basis , @xmath47 is no more canonically normalized and there is a kinetic mixing between @xmath47 and @xmath44 .
@xmath48 then , the lagrangian density becomes @xmath49 the third term is the starobinsky potential .
looking at the second and fourth terms , one may naively guess that chaotic inflation @xcite is possible neglecting the common factor @xmath50 .
however , as we shall see , this exponential factor strongly drives @xmath47 to the positive direction in the large @xmath51 region .
now let us investigate if the fields @xmath47 and/or @xmath51 play the role of inflaton which are responsible for the present density fluctuations of our universe .
for this purpose , we first study the evolution of these fields .
the evolution equations for @xmath47 and @xmath51 are given by @xmath52 where the `` dot '' denotes the derivative with respect to time @xmath53 and @xmath54 ( with @xmath55 being the scale factor ) is the expansion rate of the universe .
when the energy density of the universe is dominated by that of @xmath25 , we obtain @xmath56 where @xmath57 is the total energy density : @xmath58 by solving the above equations numerically , we follow the trajectories of @xmath47 and @xmath51 with various initial values . plane .
the green closed contour at around the minimum ( _ i.e. _ , @xmath59 ) corresponds @xmath60 , while other green lines represent the contours of @xmath61 , @xmath62 , @xmath63 , @xmath64 , @xmath65 and @xmath66 from right to left .
the solid red , solid blue , dashed red , and dashed blue lines represent the evolutions of the fields with initial conditions @xmath67 , @xmath68 , @xmath69 and @xmath70 , respectively .
note that dashed lines overlap with the solid lines for @xmath71 .
points with numbers show the @xmath72-folding numbers for each trajectory .
the trajectories are terminated at the end of inflation ( _ i.e. _ , @xmath73 ) .
, width=377 ] in fig .
[ fig1:trajectory ] , we show the contours of the potential and the evolutions of the fields on @xmath74 plane . as representative initial conditions , we choose @xmath75 , @xmath68 , @xmath69 and @xmath70 .
( the initial values of @xmath76 and @xmath77 are taken to be zero . ) with such initial conditions , we can see that @xmath25 starts to move to the @xmath47 direction first , then it settles to the real axis ( i.e. , @xmath78 ) . after reaching to the real axis
, the motion of @xmath25 is well approximated by the single - field inflation with @xmath47 ; the situation is almost the same as the non - supersymmetric original starobinsky inflation . as can be seen from the dashed lines , the trajectories are almost unchanged even if @xmath47 starts from @xmath79 . on each contour , in particular for @xmath80 , we show several points which give rise to some specific values of the @xmath72-folding numbers until the end of inflation . here
, the @xmath72-folding number is defined as @xmath81 where @xmath82 is the time at the end of inflation . in our analysis
, we define it by @xmath83 , where the slow - roll parameter @xmath84 is given by @xmath85 we have used the einstein equation in the last equality .
we can see that the change of the @xmath72-folding value in the period of @xmath86 is small .
therefore , a large value of the @xmath72-folding number during inflation , which is necessary to solve the horizon and flatness problems , should be accumulated when @xmath25 is on the real axis . as a function of the @xmath72-folding number @xmath87 , for the initial conditions @xmath88 ( red line ) and @xmath89 ( blue line ) . the dashed line , corresponding to @xmath90 , is drawn to guide the eyes .
note that @xmath87 decreases with time . , width=377 ] for @xmath91 and @xmath92 , the expansion of the universe is accelerating and decelerating , respectively .
thus , for inflation to happen , @xmath91 is necessary . to see when the expansion is accelerating , in fig .
[ fig2:epsh ] , we plot @xmath84 as a function of @xmath87 , taking @xmath93 and @xmath94 .
we can see that , just after the start of the motion , @xmath84 significantly increases and soon becomes larger than @xmath95 . in this period ,
the expansion of the universe is decelerating and not inflating .
the drop of @xmath84 at @xmath96 ( 45 ) in the red ( blue ) line corresponds to the point at which @xmath51 becomes most negative and @xmath97 ( cf .
fig .
[ fig1:trajectory ] ) .
thus , the universe transits from the decelerating epoch to the starobinsky - type inflation .
we call the period in between as `` transition period , '' and the period of the starobinsky - like expansion as `` starobinsky - inflation period . ''
the important point is that the transition period is very short ; during the transition period , @xmath87 changes @xmath98 or so .
( for the case of @xmath99 , for example , the transition period corresponds to @xmath100 . )
this is due to the fact that the motion of @xmath51 becomes suppressed soon after the condition @xmath91 is satisfied .
if we require that the causal connection be realized for the scale much longer than @xmath101 ( with @xmath102 being the wavenumber corresponding to the present hubble scale ) , the mode with the wavenumber @xmath102 should leave the horizon in the starobinsky - inflation period . then , the tensor - to - scalar ratio becomes @xmath103 and is too small to be consistent with the value given in eq . .
thus , in the light of the recent bicep2 result , the starobinsky inflation is disfavored even if the field space is complexified in the framework of old - minimal supergravity .
one of the possibilities to change this conclusion may be to consider the case where the mode with @xmath102 exits the horizon in the transition period .
however , such a solution looks unlikely . even though the density fluctuations with the wavenumber @xmath104 may be altered
, fluctuations with the wavenumber @xmath105 larger than @xmath106 have almost the same property as those in the case of starobinsky inflation .
consequently , for the angular scale of @xmath107 with @xmath108 , the density perturbations behave as those in the starobinsky model .
the bicep experiment is sensitive to the @xmath0-mode signal with @xmath109 , while the scalar - mode fluctuations for such an angular scale is well studied by using cmb and other observables .
thus , in the present model , it is difficult to enhance the tensor - mode fluctuations without conflicting observations . |
despite recent scepticism concerning early announcements of the discovery of exotic strange baryon @xmath0 two collaborations diana and leps confirmed their original results @xcite .
we refer the reader to recent experimental reviews @xcite . in this article following @xcite
we assume that @xmath1 exists with a mass equal 1540 mev and total width @xmath2 mev .
an immediate consequences is the existence of the whole exotic su(3 ) multiplet : @xmath3 .
apart from truly exotic states antidecuplet contains cryptoexotic nucleon- ( @xmath4 ) and @xmath5-like states ( @xmath6 ) .
the interpretation of these states is not well understood : one may try to associate them with some known resonances , or one may postulate the existence of new resonances with nucleon or @xmath5 quantum numbers .
this is the approach which we adopt here @xcite . following refs .
@xcite@xcite we assume that there exists new , narrow nucleon resonance @xmath7 which we will interpret as a member of @xmath3 .
if so , @xmath8 decays have to satisfy the following constraints @xcite originally discussed in @xcite:@xmath9 unfortunately the small partial width of @xmath7 to @xmath10 contradicts su(3 ) symmetry relations between the decay constants of @xmath8 . in this short note
we would like to emphasize the importance of mixing both for decays and mass spectra of the putative exotic antidecuplet baryons .
the exotic states have been already anticipated by the founders of the quark model although they did not elaborate on them .
later the arguments have been raised that they should be heavy and wide .
in contrast , chiral soliton models predicted that the pentaquark masses were generically small ( _ i.e. _ in the range of 1.51.6 gev ) @xcite .
it was much more difficult to accommodate the small decay width of @xmath1 @xcite . in sect .
[ decaywidths ] we explain how small decay width arises naturally in the chiral soliton quark model ( @xmath11qsm ) .
next , in sect .
[ gmo ] we argue that the decay coupling @xmath12 is further reduced due to gell - mann okubo ( gmo ) mixing caused by the nonzero @xmath13 . in sect .
[ rmixing ] we show how additional mixing of @xmath3 with roper octet can change decay patterns of @xmath14 .
we estimate allowed range of mixing angles and present predictions for remaining members of @xmath3 : @xmath6 and @xmath15 .
conclusions are presented in sect .
in in ref . @xcite the following nonrelativistic formula for the decay width has been used : @xmath16 it follows from the goldberger - treimann relation between axial and strong decay constants . here
@xmath17 is the mass of the decaying baryon , @xmath18 the mass of the decay product and @xmath19 is the outgoing meson momentum .
generically @xmath20 given by ( [ gammadef ] ) would be still in the range of a few hundreds of mev @xcite if not for the terms non - leading in @xmath21 expansion .
indeed , even without mixing the decay constant @xmath22 , which stands for the matrix element of the tensor decay operator @xmath23 between the physical baryon wave functions @xmath24:@xmath25 is the sum of three different contributions that are formally of different order in @xmath21 .
however , they are multiplied by the su(3 ) clebsch - gordan coefficients that also depend on @xmath26 @xcite .
for example : @xmath27 where @xmath28 .
similarly@xmath29 chiral soliton models provide us with specific predictions for constants @xmath30 @xcite .
had we neglected @xmath31 and @xmath32 in eqs.([gtnk],[gdnpi ] ) ( which would be inconsistent for @xmath12 because of @xmath26 enhancement of @xmath31 ) we would have obtained for @xmath33:@xmath34 estimating @xmath35 from the experimental value of @xmath36 decay width , and consequently@xmath37 we see that small decay width of @xmath1 results from the cancelation in ( [ gtnk ] ) .
indeed , the authors of ref . @xcite have shown that in the nonrelativistic limit of @xmath11qsm one obtains that @xmath38 with @xmath39 and consequently @xmath40 ! in the same limit @xmath11qsm predicts that @xmath41 and @xmath42 .
it follows that antidecuplet decay constants are small .
treating @xmath13 corrections as perturbation introduces mixing @xcite@xmath43 where subscripts refer to spin .
for some specific states mixing constants @xmath44 may be equal zero due to the isospin .
for example @xmath1 mixes only with @xmath45 , but this component of the wave function does not contribute to the decays to octet .
therefore only mixing of the final nucleon with @xmath3 and @xmath46 modifies the decay constant:@xmath47 ^{2}. \label{gtnkmix}%\ ] ] since @xmath48 is small the admixtures proportional to the reduced matrix elements@xmath49 are important even if mixing parameters @xmath50 and @xmath51 are not large ( for definitions see ref .
neither @xmath52 nor @xmath53 vanish in the nonrelativistic limit . therefore in this limit @xmath1 decay
occurs entirely due to the mixing . for realistic model parameters
( when @xmath54 ) there is a cancelation ( note that @xmath55 and @xmath56 ) between different terms in ( [ gtnkmix ] ) and @xmath12 is further suppressed . mixing affects decay patterns of pentaquarks violating su(3 ) relations between the decay constants @xmath22 @xcite . on somewhat more phenomenological ground
let us consider only @xmath57 mixing @xcite defining@xmath58 which can be directly extracted from ( [ gammadef ] ) if @xmath59 is known . throughout this paper
we assume that @xmath60 mev , hence @xmath61 .
it is then possible to express decay constants of other members of antidecuplet in terms of measurable physical parameters such as @xmath62 mixing angle @xmath63 and one a priori unknown parameter @xmath64 which can be estimated from @xmath11qsm calculations . here
we take @xmath65 @xcite
. then we have for example:@xmath66 if we want to interpret @xmath7 as @xmath8 we need to satisfy bounds ( [ conditions ] ) .
this is quite difficult within one angle scenario .
it is possible to nullify @xmath67 by a suitable choice of mixing angle @xmath68 , but then the mean octet mass ( i.e. the nucleon mass before mixing ) which is experimentally @xmath69 mev comes out wrong @xcite:@xmath70 the mixing angle that satisfies ( [ conditions ] ) is order of magnitude too small to account for baryon masses . for realistic mixing angles @xmath67 is dominated by @xmath71 and @xmath72 in contradiction with experimental data for @xmath7 .
in order to satisfy conditions ( [ conditions ] ) with more realistic mixing angles we have considered in ref .
@xcite scenario in which exotic antidecuplet can mix with roper resonance octet ( mixing angle @xmath73 ) . for roper octet gmo mass formulae work with much worse accuracy than for the ground state octet @xcite , so there is a need for additional mixing . since @xmath74 @xcite roper decay to @xmath75 is negligible
. however roper admixture contributes to @xmath76 and other decay modes:@xmath77 since @xmath78 is comparable with @xmath71 one may suppress @xmath67 without changing much @xmath79 . in ref .
@xcite we have found that conditions ( [ conditions ] ) are satisfied in the vicinity of the line @xmath80 we see that mixing angles are reasonable .
the decay widths and branching ratio to @xmath75 along ( [ or_lin ] ) are plotted in fig .
[ fig : twoangles ] to @xmath10 ( long dashed red line ) in mev together with branching ratio ( short dashed blue line ) of @xmath81 as functions of mixing angle @xmath63 along the line ( [ or_lin ] ) .
thin vertical lines correspond to the limits on the mixing angle @xmath63 .
the plot is made for @xmath82 . ] in fig .
[ fig : gn10bar ] we plot @xmath67 and @xmath83 together with their different components along the line ( [ or_lin ] ) .
we see that indeed @xmath67 is small due to the cancellation between @xmath71 and @xmath84 , while @xmath85 rises moderately when mixing increases .
( solid lines ) with different components defined in eq .
( [ gn10bar ] ) shown by dashed lines .
the plot is made for @xmath82.,title="fig : " ] ( solid lines ) with different components defined in eq .
( [ gn10bar ] ) shown by dashed lines .
the plot is made for @xmath82.,title="fig : " ] having established the range of mixing angles we can predict masses of the remaining antidecuplet members @xcite:@xmath86 note that bound ( [ xirange ] ) contradicts the result of na49 @xmath87 mev @xcite . from our analysis
it follows that total decay width of @xmath15 to @xmath88 and @xmath89 is of the order of 10 mev .
total width of @xmath6 does not exceed 30 mev but is also constrained from below to be larger that 10 mev .
most prominent decay channels are @xmath90 and @xmath91 with branching ratios approximately 60% and 20% respectively . due to the mixing su(3 ) forbidden decays to decuplet are possible , but small , at the level of 5 to 9% .
mixing induced by @xmath13 was first studied within the @xmath11qsm already in ref . @xcite but only in the leading order in @xmath26 .
it was extended to nonleading terms in ref .
@xcite and @xcite . mixing apears also in other approaches to pentaquarks .
for example in a diquark model @xcite antidecuplet mixes with an accompanying cryptoexotic octet .
diagonalization of strangeness induces _ ideal _ ( large ) mixing between these two representations .
the resulting physical states of nucleon quantum numbers have been interpreted as roper and @xmath92 .
however due to the ideal mixing these two states should have comparable widths @xcite , while experimentally they are differ substantially .
the discussion of masses and decay widths of the @xmath93 states under assumption that they correspond to the roper and @xmath94 done in ref .
@xcite still indicates that it is impossible to match the mass splittings with the observed branching ratios for these two resonances even for arbitrary mixing .
whether any different assignment of the diquark @xmath93 states would be compatible with the decay patterns deserves a separate study . in this short note
we have argued that due to the smallness of the reduced matrix elements of @xmath95 , which is natural in chiral soliton models , mixing with other su(3 ) representations has to be taken into account .
unfortunately , for the time being we can only speculate which mixing scenario is phenomelogically impossible , allowed or desired .
here we have examined a possibility that antidecuplet mixes with the roper octet .
mixing of roper and the ground state octets is presumably very small . indeed
the first order gmo mass formulae work very well for the ground state octet so there is almost no space for additional mixing . * acknowledgements * this paper is based on a common work with maxim polyakov and klaus goeke .
i would like to thank the organizers of the workshop `` excited qcd '' for stimulative and creative atmosphere .
biedenharn and y. dothan , _ monopolar harmonics in su(3)@xmath96 as eigenstates of the skyrme - witten model for baryons _ , e. gotsman and g. tauber ( eds . ) , _ from su(3 ) to gravity _ ,
p. 15 - 34 ; l.c .
biedenharn , y. dothan and a. stern , phys .
b * 146 * , 289 ( 1984 ) . | exotic antidecuplet baryons are predicted to be not only surprisingly light but also very narrow .
first we explain how small decay width arises in the quark soliton model .
next , we study possible mixing of exotic antidecuplet with roper octet and discuss its phenomenological consequences . |
lyman break galaxies ( lbgs ) are by definition characterized by a strong absorption longward of the wavelength due to intervening intergalactic gas and intrinsic uv emission of strong star formation , and so can easily be found using the color
color technique ( steidel et al . , 1996 ) .
this technique is efficient for redshifts larger than 2 , where intergalactic absorption is more pronounced and the galaxy uv flux is redshifted to the optical , allowing observations from the ground .
the rest frame uv spectrum of a galaxy bears the signature of its interstellar medium through absorption lines and stellar activity through line and broad band emission .
although low resolution uv spectra of galaxies are available for the local universe , high resolution data are strongly limited by the low efficiency of space telescopes .
only recently the far uv spectrograph ( fuse ) started to deliver high resolution ( hr , @xmath17 ) spectra of nearby star forming galaxies , revealing for the first time complex structures ( heckman et al . , 2001 ) . in principle
, high redshift galaxies can benefit of the fact that their uv emission is redshifted to the optical and can , therefore , be studied by large ground - based telescopes . however ,
their apparent optical magnitude is generally fainter than @xmath18 and hr spectroscopy can not be performed .
other possible candidates for hr spectroscopy are the very powerful radio galaxies , but no radio galaxy at redshift larger than @xmath19 and brighter than @xmath20 has been found so far .
this limit is just a little beyond the hr capabilities of the largest telescopes .
the chance to observe high redshift galaxies at high spectral resolution is then confined to gravitational lensed galaxies and gamma ray burst ( grb ) afterglows .
the latter are indeed potentially very interesting for future observations ( fiore et al . , 2000 )
thanks to the advent of new x ray satellites ( hete2 and in the near future swift ) that are capable of prompt identification of grbs .
the former are still penalized by a limited number of sufficiently bright targets .
a unique case is the galaxy ms1512cb58 ( @xmath0 ) .
this very bright gravitational lens lbg ( @xmath21 , ellingson et al . , 1996 ) is within the state of the art capabilities of hr spectrographs of the largest ground based telescopes .
it was serendipitously discovered by yee et al .
( 1996 ) during the cnoc cluster redshift survey and is gravitationally lensed by the cluster ms1512 + 36 at @xmath22 ( seizt et al .
the gravitational magnification has been studied in detail by seitz et al .
( 1998 ) who found that the @xmath23 arcsec@xmath24 arclet brightness is amplified @xmath25 times by the cd galaxy of the cluster , 6@xmath26 apart .
since this is the brightest lbg galaxy known , it has been widely studied in the optical , submillimeter and radio bands , both in imaging and spectroscopy .
optical low resolution spectra obtained with keck / liris ( pettini et al . , 2000 ) showed the complex structure of the interstellar medium of the galaxy .
observations of cb58 were performed using the uv visual echelle spectrograph ( uves ) of the eso very large telescope ( vlt ) .
this instrument / telescope combination is the best possible today for forest observations .
the main motivation of the program was to investigate the intervening absorption and compare its properties with the equivalent in qso forests .
the results of the analysis of the intergalactic forest distributed along the cb58 sight line are presented in this paper .
the second goal was the study of the metallicity and dust content in the interstellar medium of the galaxy itself .
these results will be presented in a future paper .
the galaxy ms1512cb58 ( @xmath27 @xmath28 , j2000 ) was observed with the high resolution uv visual echelle spectrograph ( uves , dekker et al . , 2000 )
mounted at the ut2 telescope of the vlt observatory , at paranal ( chile ) .
the program was run in service observing mode between march and august 2000 .
the log of the observations is shown in table [ t1 ] .
the red and blue arms of uves were used , centered at @xmath29 and 4370 , respectively . due to the high declination of the object with respect to the latitude of the vlt observatory ( @xmath30 ) , the observations were performed in chunks of @xmath31 hours around the meridian ( @xmath32 above the horizon ) to minimize atmospheric absorption .
indeed , the airmass ranged between 2.1 , the minimum value for this object observed from paranal , and 2.3 . in order to maximize the signal to noise ratio ( s / n ) , the ccd was rebinned @xmath33 pixels during observations . due to good seeing conditions ,
the slit aperture was always set to @xmath34 and @xmath35 for the red and blue arm set
ups , respectively , and its inclination was along the galaxy apparent major axis .
the data reduction has been performed using a midas package specifically implemented for uves data .
the final useful spectral coverage was @xmath36 for the blue arm , and @xmath37 and @xmath38 for the red arm .
the pixel size was set to a constant value of 0.2 in the final combined spectrum .
the final resolution element corresponding to two pixels then ranges from fwhm @xmath39 at 4150 , to 26 at the , to 19 at 6206 .
the noise spectrum , used to determine the errors on the line parameters , has been calculated through photon statistics propagation of object and sky spectra and the detector read out noise .
the final s / n ratio per resolution element is s / n@xmath40 at @xmath41 , s / n@xmath42 at @xmath43 , s / n@xmath44 at @xmath45 , and s / n @xmath46 at @xmath47 .
the fit to the continuum blueward of the galaxy was performed interpolating between points of the observed flux free of apparent absorption .
although this is not particularly easy due to the non excellent s / n ratio of the spectrum , at these redshifts the line density of clouds allows the detection of several regions that are most likely unabsorbed .
the final normalized galaxy and noise per resolution element spectra in the interval @xmath48 are shown in fig .
the magnification of cb58 and the excellent performance of uves make this galaxy the only suitable candidate today to study the intergalactic matter distributed along a galaxy sight line , similarly to what is normally done for qso sight lines .
the study of qso forests has led in the past to a broad understanding of the formation and evolution of the universe . very briefly
we mention successful experiments to test the intergalactic medium ( igm ) reionization through the gunn
peterson effect for hi and heii , the early metal pollution of the igm , and the intensity of the uv background radiation , as discussed by songaila ( 1998 ) , songaila et al .
( 1999 ) , heap et al .
( 2000 ) , ellison et al .
( 2000 ) , just to mention the most recent publications on these subjects . for the first time we can test some of these results considering a galaxy line of sight .
any deviation from the expected properties of the igm might be strongly connected to a deviation of the galaxy surrounding environment with respect to the qso ones . to study the cloud distribution in cb58 ,
we have performed line fitting with voigt profiles using the midas package fitlyman ( fontana & ballester , 1985 ) .
the wavelength range considered here is @xmath49 .
the upper limit is set by the wavelength of the galaxy line .
[ f1 ] shows the forest of cb58 , together with the fitted absorption profiles . in this range
some metal lines associated with cb58 itself have been identified and fitted for completeness .
no other metal lines associated with intervening metal systems at @xmath50 and 1.339 reported by pettini et al .
( 2000 ) , and at @xmath51 , 2.117 and 2.660 identified by us , have been found in this interval . in table [ t2 ]
we report absorption line parameters for the lines only , namely the redshift @xmath52 , column density @xmath53 , and the doppler parameter @xmath54 .
the intrinsic spectrum of the galaxy shows a very large absorption that extends almost 100 in the range @xmath55 .
this large trough has been considered by pettini et al .
( 2000 ) as a single interstellar cloud at @xmath56 with an hi column density of @xmath57 2 , assuming that the peak around 4540 is due to emission of cb58 redshifted to @xmath58 ( @xmath59 ) . making the same assumption
we find a slightly lower hi content , but still consistent with pettini et al.s result , with @xmath60 2 and redshift @xmath61 .
however we also notice that in our spectrum the core of this absorption ( @xmath62 ) is not completely black , and a small residual mean flux of about 1@xmath63 above the zero level is present .
this might indicate that the interstellar clouds are not fully covering the star forming regions of the galaxy and/or that resonant scattering fills up part of the absorption .
other metal lines associated with cb58 are listed in table [ t3 ] . due to the large hi absorption and metal lines associated with the galaxy , we have considered in our analysis the forest shortward of @xmath64 , namely , the redshift range @xmath65 ( @xmath66 ) . since the galaxy intrinsic brightness is much lower than a typical distant qso , the classical proximity effect , that in a qso decreases the cloud density within @xmath67 mpc from the source due to the high qso uv ionizing flux , is in the case of cb58 negligible . indeed , taking the observed magnitude and magnification , the absolute magnitude of the galaxy is @xmath68 , that is several hundreds times fainter than a high redshift qso typically used for proximity effect studies .
for instance , if we compare cb58 with a qso 400 times brighter , the ionization region is 20 times smaller .
this is even smaller if we consider that the spectrum of a galaxy is much softer than that of a qso and that the intensity of uv ionizing spectrum is much weaker .
we conclude that the proximity distance for cb58 is confined to within 1 mpc from the galaxy center .
the upper redshift limit of @xmath69 corresponds to a comoving distance from the galaxy of about 90 @xmath70 mpc ( we adopt throughout the paper a flat universe with non - zero cosmological constant : @xmath8 and @xmath71 ) . in the following sections we discuss how an excess of clouds can be present if the galaxy is embedded in a rich environment which might be typical for a star forming high redshift galaxy . to make a comparison with mean absorption in qso sight lines , we have considered two approaches : @xmath72
we have measured the mean flux depression blueward of the cb58 line ( @xmath5 ) and @xmath73 we calculated the number density of clouds .
the mean flux depression @xmath5 overcomes the problem of moderate signal to noise ratio in the spectrum .
it is defined by the equation : @xmath74 where @xmath75 and @xmath76 are the observed and intrinsic galaxy fluxes , respectively , in the forest between the rest frame and wavelengths .
this quantity is also expressed in terms of the effective optical depth in the same region : @xmath77 and has been measured in the past in qso sight lines when the data quality did not allow an accurate column density determination of absorption lines ( e.g , giallongo & cristiani , 1990 ) .
more recently fardal et al .
( 1998 ) have computed @xmath5 using different models of intergalactic absorption and compared results with observations .
the observed points are obtained not directly from the spectra , but from line lists of keck data and other lower s / n data .
their fig .
3 gives @xmath78 at @xmath79 and @xmath80 at @xmath81 . in an extensive study of the mean hi opacity in qsos , kim et al .
( 2001a ) report @xmath82 values in 13 qsos observed with vlt / uves and keck / hires and give a best fit to the data in the interval @xmath83 as @xmath84 . in our analysis , we considered the few very recent high resolution spectra of qso sight lines observed at similar redshifts , e.g. the keck / hires spectra of gb1759 + 7539 ( @xmath85 ; outram et al . , 1999 ) , and hs1946 + 7658 ( @xmath86 ; kirkman & tytler , 1997 ) , and the vlt / uves spectra of he13472457 ( @xmath87 ; kim et al . ,
2001b ) and j2233606 ( @xmath88 ; cristiani & dodorico , 2001 ) obtained during the science verification of the instrument .
details of the qso spectra used are given in table [ t4 ] , while in fig .
[ f2 ] we show the forest of cb58 and the comparison with qsos in the region of overlapping ( except for j2233606 that is too blue for cb58 ) .
for all the four qso spectra , we have excluded the proximity effect region . for gb1759 + 7539 , we have calculated the mean depression redward of the qso emission and far from the proximity effect and from a large hi absorption due to an intervening damped system at @xmath89 .
for he13472457 we have considered the region 4050 @xmath90 4250 .
for j2233606 we have excluded the region with two strong hi absorptions ( @xmath91 atoms 2 ) associated with two metal systems at @xmath92 . for cb58
we have calculate @xmath5 in the range 4150 @xmath93 4430 , corresponding to @xmath94 . for the uncertainties on the mean depression ,
we have re calculated @xmath5 considering as a new continuum the original estimate @xmath95% for cb58 , gb1759 + 7539 and j2233606 , and @xmath96% for he13472457 and hs1946 + 7658 , respectively , according to s / n .
we take as our error the uncertainty derived according to this procedure .
the value of @xmath5 in the whole cb58 interval at @xmath97 is @xmath98 .
this is consistent with the value calculated from the model of the fitted voigt profiles listed in table [ t2 ] , i.e. @xmath99 . the total mean depression @xmath5 in the qsos is given in table [ t4 ] , while in table [ t5 ] we report @xmath5 in different redshift intervals for cb58 and the qsos , with the error range between brackets .
these values of @xmath5 have been combined with results for other qsos given by kim et al .
( 2001a ) and reported , as a function of redshift , in fig .
[ f3 ] . considering the uncertainty and the cosmic variance that we expect in different lines of sight
, we can conclude that the @xmath5 values for cb58 are consistent with those for qsos in the first two redshift bins .
however , we notice that the mean depression in cb58 is appreciably larger than expected from qsos in the last redshift bin . in the last redshift bin of cb58 we have @xmath10 at @xmath100 , while the best fit to qso sight lines gives @xmath101 at the same redshift ( kim et al .
, 2001a ) , a @xmath14 effect .
we notice that metal lines associated with cb58 could contribute to @xmath5 .
this effect is marginally important in qso sight lines , because the powerful uv emission of qsos highly ionize the foreground gas of the host galaxy . to verify what this contribution is in cb58 ,
we have compared its uv spectrum with the uv spectrum of a nearby galaxy .
thanks to the advent of fuse , it is now possible to get high resolution uv spectra of galaxies .
one example is presented by heckman et al .
( 2001 ) , who discuss the ovi outflow in the dwarf starburst galaxy ngc1705 . in fig .
[ f4 ] we compare their fwhm @xmath102 resolution spectrum with that of cb58 in the region of overlapping .
the rest frame overlapping range @xmath103 corresponds for cb58 to the observed interval @xmath104 .
the rest frame of cb58 is calculated assuming a redshift of @xmath105 , that is the redshift of the strongest central absorbing cloud in its interstellar medium ( ism ) .
the absorption lines in ngc1705 show different contributions coming from the ism and high velocity clouds of the milky way , and the ism of ngc1705 itself , spanning a total velocity range of about 600 ( heckman et al . , 2001 ) .
the strongest interstellar lines in ngc1705 , feii@xmath106 or ni@xmath107 , are also identified in cb58 .
however , the involved column densities are about an order of magnitudes higher in cb58 .
the strong absorption on the blue side of the feii complex , on the other hand , is due to the galactic ism , where metallicity and column densities are much larger .
the strong siiv absorption doublet at @xmath108 and the ciii absorption line at @xmath109 in the spectrum ngc1705 are due to ob stars ( walborn & bohlin , 1996 ) . in cb58
the absorption at @xmath110 can be identified with ciii absorption redshifted by about 300 with respect to the ism . the fitted profile shown in fig .
[ f1 ] has been obtained assuming a width of 250 for this line , very similar to what observed for ngc1705 .
the contribution of the siiv doublet in cb58 seems to be not very significant ( the first component is basically absent , while the second one is very weak ) . from unpublished hst / stis high resolution data of ngc1705 ( heckman , private communication ) we have seen that there are no other strong absorptions longward of the ciii feature in ngc1705 ,
up to @xmath111 , corresponding to @xmath112 in the cb58 observed frame , so other further contributions to @xmath5 caused by the cb58 ism can be considered negligible .
we have then recalculated @xmath5 in the last wavelength bin of the cb58 forest and excluded ciii absorption around @xmath113 and found @xmath114 .
this is still significantly larger than what found in qsos and supports the idea that the excess of absorption at @xmath115 ( 150 @xmath15 mpc comoving from the galaxy ) is not due to the galaxy itself but is associated with the intervening intergalactic matter .
the next comparison between lbg and qso forests is through the number density per unit redshift of clouds with hi column density higher than a threshold @xmath116 .
this eliminates the problem of the metal lines associated with cb58 .
the number density per unit redshift is parameterized as @xmath117 . for a non evolving population in the standard friedmann universe with null cosmological constant , @xmath118 and 0.5 for @xmath119 and 0.5 , respectively , whereas a non evolving slope can be approximated by @xmath120 for a non - zero cosmological constant universe ( kim et al .
. recent work on qso sight lines at different redshifts has shown that strong clouds ( @xmath121 atoms 2 ) evolve at high redshift ( @xmath122 for @xmath123 , kim et al . , 2001a ) followed by a flat distribution at redshifts @xmath124 ( @xmath125 , hst key program , weymann et al .
, 1998 ; penton et al . , 2000 )
consistent with no evolution .
tracking the number density evolution has implied a comparison of low resolution with high resolution data .
this complicates the interpretation ; however , the results are quite striking .
the number density evolution is different for different @xmath116 , being flatter at smaller @xmath116 for the large @xmath52 range ( penton et al . ,
2000 , kim et al .
, 2001a ) .
in cb58 we have calculated @xmath6 dividing the redshift interval in three different redshift bins as done in the previous section to calculate @xmath5 .
we also consider two different @xmath116 , namely , @xmath126 and @xmath127 atoms 2 .
this is because the completeness of the sample varies along the spectrum of cb58 .
indeed , the @xmath128 completeness of hi column densities obtained from lines with doppler parameter in the range 20 @xmath129 , is @xmath130 , and @xmath127 atoms 2 for @xmath131 , and 4150 , respectively .
the limited resolution and s / n have the consequence of underestimating the number of components observed in the absorption profile and overestimating the hi column density .
in other words , for low or high @xmath116 , the number of detected clouds is underestimated or overestimated , respectively . in tables
[ t6 ] and [ t7 ] we report the number of clouds with @xmath132 and 14.0 , while in fig . [ f5 ] we show the number density evolution as a function of redshift for clouds with @xmath133 2 .
when considering results for @xmath134 one should take into account that the number of clouds might be overestimated if blending of lines is important .
however , this is unlikely to have a major effect on our sight line , given the relative sparseness of lines at these redshifts . in the same tables
, we also make a comparison with the forests of gb1759 + 7539 , he13472457 and hs1946 + 7658 . between brackets
we give the 1@xmath63 range obtained using poisson statistics ( gehrels 1986 ) .
as typically done for qso forests , the identified metal lines in cb58 and in the qsos associated with the galaxy itself or other metal systems along the line of sight have been excluded from the counting .
tables [ t6 ] and [ t7 ] show that there is consistency in the number of clouds between cb58 and qso forests in the first two redshift bins , whereas we find again a significant discrepancy when the last redshift bin is considered .
this deviation appears more clearly in fig .
[ f5 ] for clouds with hi column density larger than @xmath13 atoms 2 and is at the @xmath135 level with respect to the best fit to qso forests in the interval @xmath136 .
as already found in the previous section where the mean depression is considered , the number density of clouds gives a hint of an excess of gas absorption in cb58 with respect to qsos at @xmath137 , or in the region within 150 @xmath15 mpc from the galaxy .
this excess appears to be more concentrated in the wavelength range @xmath138 , at a mean redshift of @xmath139 .
if this structure is isolated and not associated with the galaxy itself , it is @xmath140 mpc comoving wide ( velocity width of @xmath141 ) .
observations of the forest have been fundamental in helping to unveil the nature of the intergalactic gas .
these studies have been possible using large collecting areas pointed on very bright sources .
the only suitable candidates so far have been bright ( @xmath142 ) , distant ( @xmath143 ) qsos , observed with 4 m and 10 m class ground based telescopes .
similar studies are still precluded if normal lyman break galaxies or radio galaxies are considered .
this is because no high - redshift galaxy brighter than @xmath144 is known .
a unique chance has been offered to us by vlt / uves observations of the lbg galaxy cb58 .
this galaxy at @xmath0 is gravitationally magnified @xmath25 times by an intervening galaxy cluster at @xmath22 and its optical apparent magnitude is @xmath1 .
this is by far the brightest lbg known .
we have for the first time explored the forest of a high redshift galaxy and compared the observed properties with those shown by qso sight lines .
we found in cb58 an excess of absorption at a distance from cb58 ( @xmath0 ) of about 150 @xmath15 mpc comoving ( @xmath137 ) .
this has been quantified by both measuring the mean flux depression @xmath5 due to hi absorption and the number density of clouds with @xmath145 atoms 2 .
the deviation is in both cases significant at the @xmath146 level .
the large absorption and some metal lines associated with cb58 prevent us to detect intervening clouds for redshift larger than @xmath147 , so we can not exclude the possibility that this high density region is more extended .
we now have two possible scenarios , depending on the size of the region associated with this excess .
\1 ) the excess is due to the presence of a super cluster of clouds not associated with cb58 .
this super cluster of lines appears to be concentrated at @xmath148 ( @xmath149 ) , corresponding to a region @xmath150 mpc comoving wide .
this is an interesting possibility in view of the fact that a very large structure on scales @xmath151 or @xmath152 mpc has been found at @xmath153 in the lbg galaxy distribution ( steidel et al . , 1998 ) .
steidel et al . also suggest that similar clustering is common in lbg surveys and that lbgs represent the progenitors of local massive galaxies .
\2 ) the excess is due to high - density gas associated with the environment of cb58 . in this case
, this high density region could extend all the way down to the galaxy .
this has interesting implications for the proximity effect .
the proximity effect is used to measure the intensity of the uv background radiation , assuming that the gas distribution of the igm is independent of the distance from qsos .
our findings seem to undermine this assumption , which has also been questioned by pascarelle et al .
( 2001 ) , using a completely different approach .
these authors have analyzed the clustering properties of galaxies and qso absorption line pairs at low redshifts ( @xmath154 ) and found that galaxies and clouds cluster around qsos .
this effect extends up to a velocity range from the qso of @xmath155 . in cb58
the clustering would be extended down to @xmath115 ( @xmath156 ) , corresponding to a velocity range of @xmath157 .
we can imagine that a clustering effect similar to that found by pascarelle et al .
( 2001 ) may occur in star forming lyman break galaxies , such as cb58 .
indeed , its rest frame optical emission lines give an amplified star formation rate ( sfr ) of 620 m@xmath158 per year ( teplitz et al . , 2000 ) , which becomes @xmath159 m@xmath158 per year if corrected by the amplification factor of @xmath160 .
if cb58 is a starburst galaxy , as suggested by its sfr , then a large density of starforming galaxies is expected in its environment , and therefore a large number of absorbing clouds as well , consistent with our findings .
further evidence of the peculiarity of cb58 is given by the large velocity range shown by its components : the ism absorption ( @xmath105 ) and the emission lines ( @xmath161 ) are spread over almost 1000 .
if the excess found around cb58 is typical in collapsed objects , the qso proximity effect has been underestimated so far and therefore the intensity of the uv background is overestimated .
this , in turn , has important implications for estimates of the contribution of the lyman@xmath4 forest to the barionic content of the universe . to assess the general validity of our results in other lbg forests ,
it is necessary to analyze other good quality lbg spectra .
this relies on the chance that other gravitationally lensed lbgs are discovered , hopefully with a magnification factor larger than that of cb58 .
an alternative solution to study the intrinsic properties of high redshift galaxies and the intervening igm clouds can be provided by grb afterglow observations .
grb afterglows have been found up to redshifts of 4.5 and their optical magnitude can be as bright as 15@xmath162 magnitude .
these are perfect candidates for high resolution spectroscopy , provided that they are promptly targeted soon after the burst .
the authors wish to thank tim heckman , tea
sun kim , david kirkman , phil outram and david tytler for providing spectra and forest lists .
we are also grateful to our referee , david kirkman , for very insightful comments , and tim heckman for helpful discussions .
ss acknowledges the support of the eso and stsci visitor programs . ss would like to express her gratitude for warm hospitality at eso garching where the data reduction of uves spectra has been performed .
special thanks to vanessa hill , andrea modigliani , stefano cristiani and sandro dodorico for helping with the uves pipeline .
the paranal and uves teams are acknowledged for their effective support with observations .
cristiani s. , dodorico v. , 2000 , aj , 120 , 1648 dekker h. , dodorico s. , kaufer a. , delabre b. , kotzlowski h , 2000 , the international society for optical engineering , 4008 , 534 ellingson e. , yee h. k. c. , bechtold j. , elston r. , 1996 , apj , 466 , l71 ellison s. l. , songaila a. , schaye j. , pettini m , 2000 , 120 , 117 fiore f. , nicastro f. , savaglio s. , stella l. , vietri m. , 2000 , apj , 533 , l7 fontana a. , ballester p. , 1995
, the eso messenger , 80 , 37 gehrels n. , 1986 , apj , 303 , 336 giallongo , e. , cristiani , s. , dodorico , s. , fontana , a. , savaglio , s. 1996 , apj , 466 , 46 heap s. r. , williger g. m. , smette a. , hubeny i. , sahu m. s. , jenkins e. b. , tripp t. m. , winkler j. n. , 2000 , apj , 534 , 69 heckman t. m. , sembach k. r. , meurer g. r. , strickland d. k. , martin c. l. , calzetti d. , leitherer c. , 2001 , apj , 554 , 1021 hu e. m. , kim t .- s . ,
cowie l. l. , songaila a. , rauch , m. , 1995 , aj , 110 , 1526 kim t.s .
, carswell r. f. , cristiani s. , dodorico s. , giallongo e. , 2001b , mnras , submitted kim t.s .
, cristiani s. , dodorico s. , 2001a , a&a , 373 , 757 kirkman , d. , tytler , d. , 1997 , aj , 484 , 672 lu l. , sargent w. l. w. , womble d. s. , takada hidai m. , 1996 , apj , 472 , 509 outram p. j. , chafee f. h. , carswell r. f. , 1999 , mnras , 310 , 289 pascarelle s. m. , lanzetta k. m. , chen h w , webb j. k. , 2001 , apj , in press , astro ph/0107295 penton p. j. , shull j. m. , stocke j. t. , 2000 , apj , 544 , 150 pettini m. , steidel c. c. , adelberger k. l. , dickinson m. , giavalisco m. , 2000 , apj , 528 , 96 savaglio s. , et al . , 1999 , 515 , l5 steidel c. c. , adelberger k. l. , dickinson m. , giavalisco m. , pettini m. , kellogg m. , 1998 , apj , 492 , 428 steidel c. c. , giavalisco m. , pettini m. , dickinson m. , adelberger k. l. , 1996 , apj , 426 , l17 seitz s. , saglia r. p. , bender r. , hopp u. , belloni p. , ziegler b. , 1998 , mnras , 298 , 945 songaila a. , 1998 , aj , 115 , 2184 songaila a. , hu e. m. , cowie l. l. , mcmahon r. g. , apj , 525 , l5 teplitz h. i. , mclean i. s. , becklin e. e. , figer d. f. , gilbert a. m. , graham j , r. , larkin j. e. , levenson n. a. , wilcox m. k. , 2000 , apj , 533 , l65 walborn n. r. , bohlin r. c. , 1996 , pasp , 108 , 477 weymann r. j. , et al .
1998 , apj , 506 , 1 yee h. k. , ellingson e. , bechtold j. , carlberg r. g. , cuillandre j.c . , 1996 ,
aj , 111 , 1783 cccccc & & & & & + & & + + @xmath167 range & @xmath168 & cb58 & gb1759 + 7539 & he13472457 & hs1946 + 7458 + & & & & & + 40504150 & 2.3315 & & & 0.184 ( 0.1590.208 ) & + 41504250 & 2.4549 & 0.215 ( 0.1740.253 ) & 0.234 ( 0.1930.270 ) & 0.165 ( 0.1380.189 ) & 0.219 ( 0.1950.242)@xmath163 + 42504348 & 2.5363 & 0.232 ( 0.1920.269 ) & 0.189 ( 0.1460.228 ) & & 0.240 ( 0.2160.262 ) + 43484430 & 2.6104 & 0.358 ( 0.3250.389 ) & & & 0.206 ( 0.1820.229 ) + [ 2pt ] ccccccc & & & & & & + & & & + + @xmath167 range&@xmath168 & @xmath170 & cb58 & gb1759 + 7539 & he13472457 & hs1946 + 7658 + & & & & & & + 40504150 & 2.3315 & 0.0823 & & & 10 ( 6.914.3 ) & + 41504250 & 2.4549 & 0.0823 & 11 ( 7.715.4 ) & 12 ( 8.616.6 ) & 11 ( 7.715.4 ) & 10 ( 6.914.3)@xmath163 + 42504348 & 2.5363 & 0.0806 & 15 ( 11.220.0 ) & 7 ( 4.410.8 ) & & 12 ( 8.616.6 ) + 43484430 & 2.6104 & 0.0675 & 16 ( 12.121.1 ) & & & 9 ( 6.113.1 ) + [ 2pt ] ccccccc & & & & & & + & & & + + @xmath167 range&@xmath168 & @xmath170 & cb58 & gb1759 + 7539 & he13472457 & hs1946 + 7658 + & & & & & & + 40504150 & 2.3315 & 0.0823 & & & 8 ( 5.211.9 ) & + 41504250 & 2.4549 & 0.0823 & 7 ( 4.410.8 ) & 9 ( 6.113.1 ) & 5 ( 2.98.4 ) & 6 ( 3.69.6)@xmath163 + 42504348 & 2.5363 & 0.0806 & 9 ( 6.113.1 ) & 4 ( 2.17.2 ) & & 6 ( 3.69.6 ) + 43484430 & 2.6104 & 0.0675 & 15 ( 11.220.0 ) & & & 2 ( 0.74.6 ) + [ 2pt ] | the high redshift galaxy ms1512cb58 ( @xmath0 , @xmath1 ) has been observed with the very efficient high resolution echelle spectrograph vlt / uves .
although this is a very challenging observational program for a southern hemisphere telescope ( the galaxy is located at @xmath2 declination ) , high resolution spectra ( fwhm @xmath3 ) have revealed , with unprecedented detail along a galaxy sight line , the lyman@xmath4 forest due to intervening clouds in the intergalactic medium ( igm ) .
the mean depression @xmath5 due to igm absorption blueward of the galaxy wavelength and the number density @xmath6 of clouds have been compared with equivalent results obtained for qso sight lines at similar redshifts .
our results indicate a possible excess of absorption close to the galaxy .
the mean depression at @xmath7 mpc comoving ( @xmath8 , @xmath9 ) from the galaxy is @xmath10 , to be compared with @xmath11 , expected from a best fit to qso sight lines . in the same region ( @xmath12 ) , the number density of lines with hi column density in excess of @xmath13 atoms 2 is also @xmath14 larger than expected .
this high density region is at least 60 @xmath15 mpc comoving wide , but the large absorption of the galaxy itself prevents us from detecting a possible structure extending down to the galaxy .
this excess of clouds is suggestive of two possible scenarios .
one is the presence of a super cluster of clouds not associated with cb58 .
the other is a high density of gas associated with the environment of cb58 .
indeed , a hint of the complexity of cb58 and possibly its environment is given by the huge velocity range ( almost 1000 ) between the optical emission of star forming regions and uv absorption of its interstellar medium .
2@xmath16 epsf |
let @xmath12 denote the class of analytic functions @xmath13 defined in the open unit disk @xmath14 normalized by the conditions @xmath15 . if @xmath13 and @xmath16 are analytic in @xmath17 , then @xmath13 is subordinate to @xmath16 , written @xmath18 , if there is an analytic self - map @xmath19 of @xmath17 satisfying @xmath20 and @xmath21 . for @xmath22 ,
let @xmath23 $ ] be the class consisting of normalized analytic functions @xmath24 in @xmath17 satisfying @xmath25 for instance , if @xmath26 , then @xmath27 $ ] is the class of functions @xmath24 satisfying @xmath28 in @xmath17 . the class @xmath29 $ ] of janowski starlike functions @xcite consists of @xmath30 satisfying @xmath31.\ ] ] for @xmath26 , @xmath32:= \mathcal{s}^\ast(\beta)$ ] is the usual class of starlike functions of order @xmath33 ; @xmath34:= \mathcal{s}^\ast_{\beta } = \{f \in \mathcal{a } : @xmath35:=\mathcal{s}^\ast[\beta]= \{f \in \mathcal{a } : been studied , for example , in @xcite .
a function @xmath36 is said to be close - to - convex of order @xmath33 @xcite if @xmath37 for some @xmath38 .
this article studies the generalized beesel function @xmath39 given by the power series @xmath40 where @xmath41 .
the function @xmath42 is analytic in @xmath17 and solution of the differential equation @xmath43 if @xmath3,@xmath2,@xmath4 in @xmath44 , such that @xmath45 and @xmath46 .
this normalized and generalized bessel function of the first kind of order @xmath2 , also satisfy the following recurrence relation @xmath47 which is an useful tool to study several geometric properties of @xmath48 .
there has been several works @xcite studying geometric properties of the function @xmath42 , such as on its close - to - convexity , starlikeness , and convexity , radius of starlikeness and convexity . in section @xmath49 of this paper ,
sufficient conditions on @xmath0 , @xmath1 , @xmath4 , @xmath50 are determined that will ensure @xmath48 satisfies the subordination @xmath51 .
it is to be understood that a computationally - intensive methodology with shrewd manipulations is required to obtain the results in this general framework .
the benefits of such general results are that by judicious choices of the parameters @xmath0 and @xmath1 , they give rise to several interesting applications , which include extending the results of previous works . using this subordination result , sufficient conditions are obtained for @xmath52 $ ] , which next readily gives conditions for @xmath53 to be close - to - convex .
section @xmath54 gives emphasis to the investigation of @xmath42 to be janowski convex as well as of @xmath55 to be janowski starlike .
the following lemma is needed in the sequel .
@xcite[lem : miller - mocanu-1 ] let @xmath56 , and @xmath57 satisfy @xmath58 whenever @xmath59 , @xmath60 real , @xmath61 . if @xmath2 is analytic in @xmath17 with @xmath62 , and @xmath63 for @xmath64 , then @xmath65 in @xmath17 . in the case
@xmath66 , then the condition in lemma @xmath67 generalized to @xmath68 @xmath60 real , @xmath69 and @xmath70 .
in this section , one main result on the close - to - convexity of the generalized bessel function with several consequences are discussed in details .
[ thm :- janw - cc ] let @xmath71 .
suppose @xmath72 , and @xmath4 , @xmath73 satisfy @xmath74 further let @xmath0 , @xmath1 , @xmath50 and @xmath4 satisfy either the inequality @xmath75 whenever @xmath76 or the inequality @xmath77 whenever @xmath78 if @xmath79 , then @xmath80 $ ] .
define the analytic function @xmath81 by @xmath82 then , a computation yields @xmath83 and @xmath84 thus , using the identities @xmath85@xmath86 , the bessel differential equation @xmath87 can be rewrite as @xmath88 assume @xmath89 , and define @xmath90 by @xmath91 it follows from @xmath92 that @xmath93 to ensure @xmath94 for @xmath59 , from lemma @xmath67 , it is enough to establish @xmath95 in @xmath17 for any real @xmath60 , @xmath61 , and @xmath69 .
with @xmath96 in @xmath97 , a computation yields @xmath98 since @xmath61 , and @xmath99 $ ] , @xmath100 thus @xmath101 where @xmath102 condition @xmath103 shows that @xmath104 since @xmath105 for @xmath106 , it is clear that @xmath107 when @xmath108 with @xmath109 .
as @xmath110 , the above condition holds whenever @xmath111 that is , when @xmath112 to establish inequality @xmath113 , consider the polynomial @xmath114 given by @xmath115 where @xmath116 the constraint @xmath117 yields @xmath118 , and thus @xmath119 . now inequality @xmath120 readily implies that @xmath121 now considers the case of the constraint @xmath122 , which is equivalent to @xmath123 .
then the minimum of @xmath114 occurs at @xmath124 , and @xmath125 yields @xmath126 evidently @xmath127 satisfies the hypothesis of lemma @xmath67 , and thus @xmath128 , that is , @xmath129 hence there exists an analytic self - map @xmath19 of @xmath17 with @xmath20 such that @xmath130 which implies that @xmath131 theorem @xmath132 gives rise to simple conditions on @xmath4 and @xmath50 to ensure @xmath42 maps @xmath17 into a half - plane .
[ cor :- janw - cc-1 ] let @xmath133 and @xmath134
. then @xmath135 choose @xmath136 , and @xmath137 in theorem @xmath132 .
then both the conditions @xmath103 and @xmath117 are equivalent to @xmath138 which clearly holds for @xmath139 .
the proof will complete if the hypothesis @xmath120 holds , i.e. , @xmath140 since @xmath141 , it follows that @xmath142 which establishes @xmath143 .
[ cor :- janw - cc-3 ] let @xmath4 , @xmath50 be real such that @xmath144 then @xmath145
. put @xmath146 and @xmath137 in theorem @xmath132 .
the condition @xmath103 reduces to @xmath138 , which holds in all cases .
it is sufficient to establish conditions @xmath117 and @xmath120 , or equivalently , @xmath147 and @xmath148 for the case when @xmath133 , both the inequality @xmath149 and @xmath150 hold as @xmath138 .
finally it is readily established for @xmath151 and @xmath152 that @xmath153 and @xmath154 it is known that for @xmath155 and @xmath156 , the generalized bessel functions @xmath157 and @xmath158 respectively gives the spherical bessel and modified spherical bessel functions .
this specific choice of @xmath3 and @xmath4 , corollary @xmath159 yield @xmath160 for @xmath161 , and @xmath162 , for @xmath163 . since @xmath164 for @xmath165 , following inequalities
can be obtain with the aid of results in @xcite . for @xmath161 ,
the modified spherical bessel functions @xmath166 satisfy the following inequalities .
@xmath167 [ thm : janw - cc-2 ]
let @xmath168 and @xmath4 , @xmath73 satisfy @xmath169 suppose @xmath0 , @xmath1 , @xmath50 and @xmath4 satisfy either the inequality @xmath170 whenever @xmath171 or the inequality @xmath172 whenever @xmath173 if @xmath79 , then @xmath80 $ ] .
first , proceed similar to the proof of theorem @xmath132 and derive the expression of @xmath174 as given in @xmath175 .
now for @xmath61 , @xmath176 , and @xmath177 , @xmath178 and then with @xmath179 , and @xmath180 , it follows that @xmath181 where @xmath182 observe that the inequality @xmath183 implies that @xmath184 .
thus @xmath185 for all @xmath176 provided @xmath186 , that is , for @xmath187 , @xmath188 with @xmath110 , it is enough to show for @xmath189 , @xmath190 which is equivalent to @xmath191 where @xmath192 if @xmath193 holds , then @xmath194 .
since @xmath195 is increasing , then @xmath196 , which is nonnegative from @xmath197 .
on the other hand , if @xmath198 holds , then @xmath199 , @xmath200 , and @xmath201 implies @xmath202 .
either case establishes @xmath203 .
[ thm :- janw - cc-3 ] let @xmath71 .
suppose @xmath72 , @xmath204 with @xmath8 and satisfying @xmath205 further let @xmath0 , @xmath1 , @xmath50 and @xmath4 satisfy either @xmath206 whenever @xmath207 or the inequality @xmath208 when @xmath209 if @xmath210 , then @xmath211 $ ] .
[ thm :- janw - cc-4 ] let @xmath212 .
suppose @xmath204 , @xmath213 , such that @xmath214 suppose @xmath0 , @xmath1 , @xmath50 and @xmath4 satisfy either @xmath215 whenever @xmath216 or the inequality @xmath217 when @xmath218 if @xmath210 , then @xmath211 $ ] .
let @xmath219 , and @xmath220 then @xmath221 is close - to - convex of order @xmath222 with respect to the identity function .
let @xmath4 be a nonzero real number , and @xmath223 .
this section contributes to find conditions to ensure a normalized and generalized bessel functions @xmath225 in the class of janowski starlike functions . for this purpose ,
first sufficient conditions for @xmath42 to be janowski convex is determined , and then an application of relation @xmath226 yields conditions for @xmath227.$ ] define an analytic function @xmath235 by @xmath236 then @xmath237 and @xmath238 a rearrangement of @xmath239 yields @xmath240 thus , @xmath241 now a differentiation of @xmath87 leads to @xmath242 which give @xmath243 using @xmath244 and @xmath245 , @xmath246 yields @xmath247 equivalently @xmath248 for @xmath255 , @xmath176 , @xmath256 note that condition @xmath257 implies @xmath258 . in this case
, @xmath259 has a maximum at @xmath260 .
thus @xmath261 for all real @xmath60 provided @xmath262 since @xmath110 , it is left to show that @xmath263 @xmath189 . the above inequality is equivalent to @xmath264 where @xmath265 thus @xmath127 satisfies the hypothesis of lemma @xmath67 , and hence @xmath128 , or equivalently @xmath270 by definition of subordination , there exists an analytic self - map @xmath19 of @xmath17 with @xmath20 and @xmath271 a simple computation shows that @xmath272 and hence @xmath273 | sufficient conditions on @xmath0 , @xmath1 , @xmath2 , @xmath3 and @xmath4 are determined that will ensure the generalized bessel functions @xmath5 satisfies the subordination @xmath6 . in particular this gives conditions for @xmath7 , @xmath8 to be close - to - convex . also
, conditions for which @xmath9 to be janowski convex , and @xmath10 to be janowski starlike in the unit disk @xmath11 are obtained . |
pentaquark is the name for baryons made up of four quarks and one antiquark .
recent advances in theoretical and experimental work led to the observation of pentaquark candidates by some experiments , however , other experiments reported no observation from their search . for more details , see the review of the experimental evidence on pentaquarks and critical discussion @xcite .
the question of the existence of pentaquarks is raised again , due to the non - observation in so many experiments . to answer the question , it is important to assess the cross sections in different processes .
the candidate @xmath0 peak is seen in different reactions namely of @xmath5+a , @xmath6+a , p+p , k+xe , e+d , e+p , k+xe@xcite@xcite .
all of those reactions involved at least one baryon in the initial state . but other experiments , e.g. @xmath7 : babar , belle , bes , lep experiments@xcite , @xmath8 : cdf , d0 @xmath9:e690 , @xmath10 : focus , @xmath9 : hera - b , @xmath11 : zeus ( for the @xmath12 ) @xmath13 : compass , hadronic z decays : lep , @xmath14 , @xmath15 , p on a : hypercp , @xmath16 : l3 , @xmath17 on p : selex , @xmath9 : sphinx
, @xmath18 : wa89 @xmath19 : lass@xcite , did not observe those candidates . because the experiments such as @xmath7 did not observe those candidates , it is generally believed a non - zero initial baryon number is essential to the pentaquark production , i.e. pp collisions or collisions involved with nuclei , as we can see the collision types of the positive reports . to check this ,
we calculate in this paper the pentaquark production in @xmath1 annihilations .
a theoretical comparison of pentaquark production between p+p@xcite and @xmath1 processes may be helpful to the question if pentaquarks exist .
why we choose the microcanonical approach to calculate ?
traditionally , the hadron production in @xmath1 scattering at high energy is treated as a two - stage process .
first @xmath20 is calculated using perturbative theory
. then the @xmath21 system produces hadrons phenomenologically based on string fragmentation or cluster fragmentation .
pentaquark states are exotic , hard to treated within the frame of conventional string models . in this situation statistical approaches may be of great help@xcite,@xcite . it was hagedorn who introduced statistical methods into the strong interaction physics in order to calculate the momentum spectra of the produced particles and the production of strange particles@xcite . later , after statistical models have been successfully applied to relativistic heavy ion collisions @xcite , becattini and heinz @xcite came back to the statistical description of elementary reactions . according to the situations of obeying conservation laws ,
statistical models are classified into four ensembles : * microcanonical : both , material conservation laws ( q , b , s , c , @xmath22 ) and motional conservation laws ( e , @xmath23 , @xmath24 , @xmath22 ) , hold exactly . *
canonical : material conservation laws hold exactly , but motional conservation laws hold on average ( a temperature is introduced ) . * grand - canonical : both material conservation laws and motional conservation laws hold on the average ( temperature and chemical potentials introduced ) . *
mixed ensemble : for example , to estimate particle production in heavy ion collisions , so - called partially canonical and partially grand - canonical models are employed where strangeness conserves strictly , but the conservation of net charge hold on the average .
what one expects is that the microcanonical ensemble must be used for very small volumes , i.e. the systems created by the collisions between elementary particles . for intermediate volumes
the canonical ensemble should be a good approximation , while for very large volumes , i.e. the systems created by heavy ion collisions , the grand - canonical ensemble can be employed .
therefore we take a microcanonical approach to calculate the production of pentaquark states from electron - positron annihilations .
the idea was originally proposed by fermi in its microcanonical form@xcite and realized with markov chain technique@xcite .
we calculate the hadron production in @xmath1 annihilations at a given energy as a statistical decay of a cluster which carries net quark contents @xmath25 .
the cluster is charactered by three parameters , cluster energy ( mass ) @xmath26 , volume @xmath27 and strangeness suppression factor @xmath28 .
we assume that hadron production from the cluster is dominated by the n - body phase space .
more precisely , the probability of the cluster hadronization into a configuration @xmath29 of hadrons @xmath30 with four momenta @xmath31 is given by the microcanonical partition function @xmath32 of an ideal , relativistic gas of the @xmath33 hadrons @xcite , @xmath34 with @xmath35 being the energy , and @xmath36 the 3-momentum of particle @xmath37 .
the term @xmath38 ensures flavour conservation ; @xmath39 is the flavour vector of hadron @xmath37 .
the symbol @xmath40 represents the set of hadron species considered : we take @xmath40 to contain the pseudoscalar and vector mesons @xmath41 and the lowest spin-@xmath42 and spin-@xmath43 baryons @xmath44 and the corresponding antibaryons .
@xmath45 is the number of hadrons of species @xmath46 , and @xmath47 is the degeneracy of particle @xmath37 .
it s well known that strangeness will be overpopulated if the hadron production is purely determined by the @xmath48-body phase space .
the common treatment is to introduce so - called strangeness suppression factor @xmath28 with @xmath49 .
so we employ this factor as well in eq .
( 1 ) , and the index @xmath50 is the number of ( anti)strange components in the final - state particle @xmath51 , i.e. for kaons , @xmath52 , @xmath53 and their antibaryons , @xmath54 ; for @xmath55-meson , @xmath56 and @xmath57 , @xmath58 ; for @xmath59 and @xmath60 , @xmath61 . similar to the previous work@xcite , we add the pentaquark states @xmath2 and @xmath62 into @xmath63 .
the @xmath64 has quark contents @xmath65 $ ] .
the @xmath62 can be @xmath66 $ ] , @xmath67 $ ] , @xmath68 $ ] or @xmath69 $ ] .
the spin of pentaquark states can not be determined by experiments yet , and it is generally accepted pentaquark states are spin-@xmath70 particles , so we take a degeneracy factor @xmath71 . as for strangeness suppression , @xmath54 for @xmath0 and @xmath58 for @xmath56(1860 ) .
the high dimensional phase space integral is verified via constructing markov chains of hadron configurations @xmath15 .
the metropolis algorithm provides random configurations @xmath15 according to the weight , the corresponding microcanonical partition function @xmath72 .
all possible random configurations are included . working with markov chains one has to worry about two kinds of convergences :
the number of iterations per chain must be sufficiently big ( essentially a multiple of the so - called auto - correlation time ) , otherwise the method is simply wrong .
secondly , the number of simulated chains must be sufficient large , to obtain the desired statistical accuracy .
questions related to the auto - correlation time have been studied in detail in earlier publications@xcite , so that the error due to auto - correlations can be neglected .
the statistical error will be treated carefully at the result section .
in addition to checking auto - correlations and statistical errors , also physics cross checks have been performed : a comparison @xcite of our monte carlo method with canonical method gives a good agreement when the systems have big volumes , i.e. 50@xmath73 and high energies , i.e. 10 times the mass of observed particles .
we generate randomly configurations @xmath74 according to the probability distribution @xmath32 .
the monte carlo technique allows to calculate mean values of an observable as @xmath75 where @xmath76 means summation over all possible configurations and integration over the @xmath31 variables .
@xmath77 is some observable assigned to each configuration , as for example the @xmath78 multiplicity @xmath79 of hadrons of species @xmath80 present in @xmath74 .
since @xmath81 depends on @xmath82 , @xmath83 and @xmath84 , we usually write @xmath85 . the cluster parameters energy @xmath26 , volume @xmath27 and strange factor @xmath28
are chosen so that the cluster decay reproduces best the measured multiplicity of the selected hadrons in @xmath1 annihilations at a given energy of @xmath86 .
this is achieved by minimizing @xmath87 : @xmath88^{2}}{\sigma _ { j}^{2 } } \label{eq : ccc}\end{aligned}\ ] ] where @xmath89 , and @xmath90 are the experimentally measured multiplicity and its error of some selected hadron species @xmath91 in @xmath1 collisions at an energy of @xmath86 .
with the cluster parameters determined by the selected hadrons , we can predict the multiplicity of any particles included in the hadron species list , i.e. pentaquark states from @xmath1 collisions at the energy of @xmath86 .
we calculate pentaquark production from @xmath1 collisions at @xmath92 gev .
this is realized by two steps : first , we have to determine the cluster parameters energy @xmath26 , volume @xmath27 and strange factor @xmath28 for each collision energy @xmath93 ; then , we use the obtained cluster parameters to calculate pentaquark production . for the first step : with data taken from @xcite , we select @xmath94 and their antiparticles to determine the cluster parameters energy @xmath26 , volume @xmath27 and strange factor @xmath28
. the available experimental yields of other particles are used to check the reliability of our results . for each set of parameters ( @xmath26 , @xmath27 , @xmath28 ) ,
the multiplicities of selected hadrons are calculated from 10,000 random hadron configurations so that the relative statistical errors of the selected hadrons are within 0.03 .
then the best - chosen set of parameters is obtained by minimizing @xmath87 as in eq.(3 ) .
fig.[fig : eefit3p ] displays the results of our fit procedure in comparison with the experimental data at different energies . here
the error bars of theoretical yields are statistical .
( color online ) a comparison between hadron production from cluster decay and from @xmath1 annihilation experiments at 29 gev , 35 gev , and 91.2 gev . ]
fig.1 tells that the best - chosen cluster parameters with the experimental multiplicities of @xmath95 from @xmath1 collisions at the energy of @xmath96 gev can also reproduce very well the yields of other particles such as @xmath97 .
the less well reproduced particles are @xmath98 , @xmath99 and @xmath100 , spin-1 meson and spin-@xmath101 baryons . at the energy of @xmath102 gev , the best - chosen cluster parameters with the experimental yields of @xmath95 can also reproduce very well the yields of other particles such as @xmath103 . at the energy of @xmath104 gev , the best - chosen cluster parameters with the experimental yields of @xmath95 from @xmath1 collisions can well reproduce the yields of other particles such as @xmath105 .
the less well reproduced particles are @xmath98 , @xmath106 and @xmath100 , spin-1 meson and spin-@xmath101 baryons .
the large list of reproduced particle proves that the cluster parameters , energy @xmath26 , volume @xmath27 and strange factor @xmath28 , determined by 4 particle yields of @xmath95 can be reliably used to estimate the yields of particles , especially for spin-0 mesons and spin-@xmath70 baryons . as it is accepted that the spin of pentaquark states is @xmath70 , our estimation on pentaquark yields does not suffer from the problem with spin-3/2 baryon yields . in tab .
[ tab : table1 ] the best - chosen set of parameters ( @xmath26 , @xmath27 @xmath28 ) , for @xmath1 collisions at the energy of @xmath10729 , 35 and 91.2 gev are collected .
the small @xmath108/dof indicates a good fit quality .
we find that with the increase of @xmath1 energy @xmath4 , the created cluster has a higher energy @xmath26 and a bigger volume @xmath27 .
the strange factor @xmath28 does not change much with the increase of @xmath1 energy @xmath4 .
one may feel hard to understand why @xmath109 does not hold , instead , the created cluster has an energy @xmath26 much smaller than the @xmath1 energy @xmath4 .
let s imagine an expanding fireball ( mainly longitudinal / thrust ) created in @xmath1 annihilation . in this case
, there is a lot of collective kinetic energy .
the parameter @xmath26 means the sum of the energies of volume elements in their proper frames , in other words their invariant masses .
so this effective mass is much smaller than the mass of the total system , which is of course @xmath1 energy @xmath4 .
if we only consider total yields , we do not need to specify the details of the collective expansion .
the price is that this simple 3-parameter model can not make any statements about transverse momentum spectra or rapidity spectra .
the volume we find here is much bigger than that in fermi s model @xcite , where the volume is estimated according to the colliding particles with the lorentz contact . in our case ,
the volume @xmath27 is the size of the collision system at the moment of hadronization .
after @xmath1 annihilates , the collision system gets many quark pair production and expanding with time .
so the size is much much bigger than that of an electron and a positron .
the strangeness suppression factor @xmath28 changes very little with the collision energy @xmath4 .
the value is around 0.7 , bigger than what we found in pp collisions .
this is also consistent with the result from canonical fitting by becattini@xcite .
.[tab : table1 ] cluster parameters for @xmath1 annihilations at different energies determined by the yields of @xmath95 and their antiparticles . [ cols="<,<,<,<,<",options="header " , ] with the cluster parameters obtained above , we can calculate the production of all hadrons in the hadron list .
pentaquark states have been included in the hadron list , so we can get the @xmath78 yields of pentaquark states from @xmath1 annihilations at the energy of 29 , 35 , and 91.2 gev . since the yields of pentaquark states are much lower than ordinary hadrons , we calculate their yields from the average of 1,000,000 random configurations with eq.(2 ) . in fig .
[ fig:1penta ] we plot the result as @xmath78 yields of pentaquark states versus the center mass system energy @xmath4 , and compare with that from pp collisions at sps energy e@xmath110=158 gev and rhic energy e@xmath111=200 gev .
the error bar is a combination of the statistical monte carlo evaluation error with the influence of the fit error of the parameters .
the two errors are at the same magnitude , and the statistical one is smaller . a comparison of the @xmath78 yields of @xmath0 ( left ) and @xmath112(right ) from @xmath1 annihilations and from pp collisions at@xcite . ] the production of pentaquark states , i.e. @xmath0 and @xmath112 , from @xmath1 annihilations increase with the collision energy . in pp collisions
, we can see the similar behavior for all pentaquark states except @xmath0 .
the production of @xmath0 in pp collisions is relatively high at a few gev , decreases slightly , then increases at teens gev and keep increasing for higher collision energies@xcite .
the special behavior of @xmath0 in pp collisions at low energies is due to the proton excitation , which does not apply in @xmath1 collisions .
the yields of @xmath0 from @xmath1 collisions at the energies we study are at the same magnitude of that from pp collisions .
the @xmath0 production from @xmath1 at lep energy @xmath10791.2 gev is even higher than the yields from pp collisions at both sps and rhic energy . for @xmath112
, we get obviously higher production from @xmath1 collisions than from pp collisions .
this can be understood : + 1 .
the average mass of clusters created in @xmath1 collisions is much bigger than that in pp collisions , i.e. the clusters have average mass 17.6 gev for @xmath1 at @xmath10791.2 gev , and mass 7.3 gev for pp collisions at sps and mass 16.15 gev at rhic energy .
the cluster mass is most sensitive parameter for particle production .
certainly cluster with big mass can produce more particles .
the strangeness suppression factor is bigger in @xmath1 collisions .
@xmath112 contains two strange quarks , and gets squared strangeness suppression , which is about 0.1 in pp collisions and about 0.4 in @xmath1 collisions .
the particle ratios @xmath113 and @xmath114 are of interest and have been discussed by several different approaches in heavy collisions and pp collisions .
here we study the ratios in @xmath1 collisions at different energies .
we get the @xmath115 about 0.03 in @xmath1 annihilations , while it is much smaller , about 0.007 , in pp collisions with the microcanonical approach@xcite .
we recognize the big difference can be caused by the strangeness suppression factor , 0.6 @xmath116 0.7 in @xmath1 annihilations , but 0.33 in pp collisions .
the grandcanonical ensemble@xcite gives about 0.06 for this ratio in heavy ion collisions , which is even higher .
this occurs when the strange chemical potential @xmath117 , which corresponds to @xmath118 , an even bigger strangeness suppression factor .
we get higher @xmath114 ratio , about 0.1 , in @xmath1 annihilations , which is 0.02 in pp collisions with the microcanonical approach@xcite , and 0.01 in heavy ion collisions obtained by the grandcanonical results@xcite .
we estimate the pentaquark production , i.e. @xmath0 and @xmath112 , from @xmath1 annihilations at the energy of @xmath4=29 , 35 and 91.2 gev using fermi statistical model as originally proposed in its microcanonical form .
we obtain increasing production of pentaquark states with the increase of collision energy . comparing with the previous work for pp collisions , we find that the yields of @xmath0 , from @xmath1 at the above mentioned energies are at the same magnitude as those from pp collisions at sps and rhic energies . the yields of @xmath0 from @xmath1 at lep energy @xmath4=91.2 gev is higher than the yields from pp collisions at sps and rhic energies . from our estimation ,
the production of @xmath112 from @xmath1 collisions is obviously higher than that from pp collisions at sps and rhic energies . that s very different from the experimental report the observation of @xmath112 was reported for the first time in the sps experiment , by na49 collaboration@xcite ; but none of the four collaborations in lep experiment get the observation , though they did search .
theoretically we conclude that fermi statistical model pentaquark production is quite high from @xmath1 annihilations if pentaquark states do exist due to a very high energy clusters are created when the @xmath1 annihilate . the energy parameter @xmath26=17.6 gev for @xmath1 at @xmath10791.2 gev , while @xmath26=16.2 gev for pp collisions at @xmath107200 gev .
the average proper mass of clusters created in pp collisions at rhic energy is lower , indicates a lot of energy is taken away collectively , due to the existence of the leading particles .
initial baryons are not necessary for pentaquark production
. the clusters with big masses created from any kind of high energy collisions can provide a rather high yield .
this work was sponsored by national natural science foundation of china ( nsfc @xmath11910447110 and @xmath11910505010 ) and the scientific research foundation for the returned overseas chinese scholars , state education ministry ( srf for rocs , sem , ( 2005)@xmath119383 ) .
fml thanks prof .
ma for fruitful discussions .
10 sonia kabana , j. phys .
g * 31 * , ( 2005 ) , s1155-s1164 .
clas collaboration , ( v. kubarovsky et al . ) , _ phys .
lett . _ * 92 * , ( 2004 ) 032001 , hep - ex/0311046 .
+ diana collaboration , ( v. barmin et al ) , _ phys . atom .
nucl . _ * 66 * , ( 2003 ) 1715 , and yad .
66 , ( 2003 ) , 1763 , hep - ex/0304040 .
+ hermes collaboration , phys .
b 585 , ( 2004 ) , 213 , [ hep - ex/0312044 ] . + a. e. asratyan et al . , phys .
67 , ( 2004 ) , 682 , yad .
67 , ( 2004 ) 704 , hep - ex/0309042 .
+ zeus collaboration , phys .
b 591 , ( 2004 ) , 7 , [ hep - ex/0403051 ] . + cosy - tof collaboration , phys . lett .
b 595 , ( 2004 ) , 127 , hep - ex/0403011 .
+ l. camilleri , nomad collaboration , contribution to the 21st international conference on neutrino physics and astrophysics , paris , 14 - 19 june , 2004 . c. alt et al , na49 collaboration , phys .
lett . * 92 * , 042003(2004 ) .
armstrong , nucl .
. suppl . *
142 * , 364 - 369 ( 2005 ) .
babar collaboration , hep - ex/0408037 and hep - ex/0408064 .
+ i. abt et al . , hera - b collaboration , hep - ex/0408048 . + t. wengler et al . , lep results , hep - ex/0405080 .
+ i. gorelov et al . , cdf collaboration , hep - ex/0408025 .
+ b. t. huffman et al , d0 collaboration , fermilab - conf-04/074-e , june 2004 .
+ r. mizuk , belle collaboration , contribution to the conference pentaquarks 2004 , spring-8 , july 20 , 2004 .
+ e. gottschalk et al , e690 collaboration , contribution to the conference pentaquarks 2004 , spring-8 , july 20 , 2004 .
+ g. brona et al . ,
compass coll .
, 2004 wwwcompass.cern.ch/compass/notes/2004-5 .
+ k. stenson et al . , focus coll . , 2004 , hep - ex/0412021 .
+ m. longo et al , hypercp coll .
, hep - ex/0410027 .
+ j napolitano et al , hep - ex/0412031 .
+ l3 coll .
, ichep 2004 conference , ichep04.ihep.ac.cn + j engelfried et al . ,
selex coll .
, these proceedings .
+ y. antipov et al . ,
sphinx coll .
j. a21 2004 455 , hep - ex/0407026 .
+ m antamovich et al . ,
wa89 coll .
, hep - ex/0405042 .
r. hagedorn , nucl .
b * 24 * ( 1970 ) 93 .
+ p. siemens , j. kapusta , phys .
* 43 * ( 1979 ) 1486 .
+ a. z. mekjian , nucl .
phys . a**384 * * ( 1982 ) 492 .
+ l. csernai , j. kapusta phys .
* 131 * ( 1986 ) 223 .
+ h. stoecker , w. greiner , phys . rep .
* 137 * ( 1986 ) 279 .
+ j. cleymanns and h. satz , z. phys . *
c57 * ( 1993 ) 135 ; + j. cleymanns , d. elliott , h. satz , and r.l .
thews , cern - th-95 - 298 , ( 1996 ) .
+ j. rafelski and j. letessier , j. phys .
g : nucl . part .
* 28 * ( 2002 ) 1819 - 1832 .
braun - munzinger , k. redlich and j. stachel , arxiv : nucl - th/0304013 .
f. becattini and u. heinz z.phys .
* c76 * ( 1997 ) 269 .
+ f. becattini , hep - ph/9701275 k. werner and j. aichelin , phys .
c * 52 * ( 1995 ) 1584 - 1603 .
f.m.liu , k.werner and j.aichelin , phys .
c * 68 * , ( 2003 ) 024905 .
+ @xmath120 : tpc coll .
_ b * 184 * , 299 ( 1987 ) + @xmath121 : tpc coll . ,
_ z. phys .
_ c * 27 * , 187 ( 1985 ) + @xmath122 : mark - ii coll .
lett . _ * 61 * , 1057 ( 1988 ) ; hrs coll .
_ b * 205 * , 111 ( 1987 ) + @xmath123 : hrs coll .
_ d * 40 * , 706 ( 1989 ) + @xmath124 : mark - ii coll . , _ phys .
lett . _ * 61 * , 1057 ( 1988 ) + @xmath55 : tpc coll .
_ * 52 * , 2201 ( 1984 ) + k@xmath125 : tpc coll .
_ b * 184 * , 299 ( 1987 ) + k@xmath126 : hrs coll . , _ phys .
_ b * 199 * , 151 ( 1987 ) + k@xmath127 : tpc coll .
, _ phys . rev .
* 53 * , 2378 ( 1984 ) ; hrs coll .
_ d * 40 * , 706 ( 1989 ) + p : tpc coll . ,
b * 184 * , 299 ( 1987 ) + @xmath52 : tpc coll .
lett . _ * 54 * , 274 ( 1985 ) ; mark - ii coll .
_ * 54 * , 2071 ( 1986 ) ; hrs coll . , _ phys . rev .
_ d * 45 * , 3949 ( 1992 ) + @xmath128 : hrs coll .
_ * 58 * , 2627 ( 1987 ) + @xmath129 : mark - ii coll . , _ phys . rev .
* 58 * , 644 ( 1987 ) ; hrs coll . ,
rev . lett .
_ * 58 * , 2627 ( 1987 ) + @xmath59 : mark - ii coll .
lett . _ * 59 * , 2412 ( 1987 ) + + @xmath120 : tasso coll . , _
_ c * 42 * , 189 ( 1989 ) + @xmath121 : tasso coll . ,
_ z. phys .
_ c * 33 * , 13 ( 1985 ) ; cello coll . ,
_ z. phys .
_ c * 47 * , 1 ( 1990 ) ; jade coll .
_ z. phys .
_ c * 46 * , 1 ( 1990 ) + @xmath122 : cello coll .
_ z. phys .
_ c * 47 * , 1 ( 1990 ) ; jade coll . , _
_ c * 46 * , 1 ( 1990 ) + k@xmath125 : tasso coll . , _
_ c * 42 * , 189 ( 1989 ) + k@xmath126 : cello coll . , _
_ c * 46 * , 397 ( 1990 ) ; jade coll .
_ b * 145 * , 441 ( 1984 ) ; tasso coll . ,
_ z. phys .
_ c * 47 * , 167 ( 1990 ) + p : jade coll .
_ b * 104 * , 325 ( 1981 ) ; tasso coll . ,
_ z. phys .
_ c * 42 * , 189 ( 1989 ) + @xmath52 : tasso coll . ,
_ z. phys .
_ c * 27 * , 27 ( 1985 ) ; tasso coll . , _
_ c * 45 * , 209 ( 1989 ) ; cello coll . ,
_ z. phys .
_ c * 46 * , 397 ( 1990 ) + @xmath129 : tasso coll . ,
_ z. phys .
_ c * 27 * , 27 ( 1985 ) ; tasso coll . ,
_ z. phys .
_ c * 45 * , 209 ( 1989 ) ; + + @xmath120 : opal coll .
_ z. phys .
_ c * 63 * , 181 ( 199 ) + @xmath121 : l3 coll . , _ phys .
_ b * 328 * , 223 ( 1994 ) ; delphi coll . , _
_ c * 69 * , 561 ( 1996 ) ; aleph coll . , cern / ppe 96 - 168 ( 1996 ) + @xmath122 : aleph coll .
_ b * 292 * , 210 ( 1992 ) , extrapolated by p. chliapnikov , v. a. uvarov , _ phys .
_ b * 345 * , 313 ( 1995 ) ; l3 coll .
_ b * 328 * , 223 ( 1994 ) + @xmath123 : aleph coll . ,
_ z. phys .
_ c * 69 * , 379 ( 1996 ) ; delphi coll . , _
_ c * 65 * , 587 ( 1994 ) + @xmath130 : aleph coll . , _
_ c * 69 * , 379 ( 1996 ) + @xmath131 : aleph coll .
_ b * 292 * , 210 ( 1992 ) , extrapolated by p. chliapnikov , v. a. uvarov , _ phys .
_ b * 345 * , 313 ( 1995 ) + @xmath55 : aleph coll . ,
_ z. phys .
_ c * 69 * , 379 ( 1996 ) ; opal coll . , _
_ c * 68 * , 1 ( 1995 ) ; delphi coll .
, cern / ppe 96 - 77 ( 1996 ) + k@xmath125 : delphi coll . , _ nucl .
_ b * 444 * , 3 ( 1995 ) ; opal coll . , _ z. phys .
_ c * 63 * , 181 ( 1994 ) + k@xmath126 : delphi coll . , _ z. phys .
_ c * 65 * , 587 ( 1995 ) ; opal coll . , _ phys .
_ b * 305 * , 407 ( 1993 ) + k@xmath127 : aleph coll . , _ z. phys .
_ c * 69 * , 379 ( 1996 ) ; opal coll . ,
_ z. phys .
_ c * 68 * , 1 ( 1995 ) ; delphi coll .
, cern / ppe 96 - 77 ( 1996 ) + p : delphi coll .
_ b * 444 * , 3 ( 1995 ) ; opal coll . ,
_ z. phys .
_ c * 63 * , 181 ( 1994 ) + @xmath132 : delphi coll . , _ phys .
_ b * 361 * , 207 ( 1995 ) ; opal coll .
_ b * 358 * , 162 ( 1995 ) + @xmath52 : delphi coll . , _ phys .
_ b * 318 * , 249 ( 1993 ) ; l3 coll .
_ b * 328 * , 223 ( 1994 ) ; aleph coll . ,
_ z. phys .
_ c * 64 * , 361 ( 1994 ) ; opal coll . , cern / ppe 96 - 099 ( 1996 ) + @xmath128 : aleph coll . , contributed paper eps0419 to the 1995 european physical society conference ( brussels ) ; delphi coll . , _
_ c * 67 * , 543 ( 1995 ) ; opal coll .
, cern / ppe 96 - 099 ( 1996 ) + @xmath133 : delphi coll .
, contributed paper pa01 - 40 to the 1996 i.c.h.e.p .
conference ( warsaw ) ; opal coll . , cern / ppe 96 - 099 ( 1996 ) + @xmath134 : opal coll . , cern / ppe 96 - 099 ( 1996 ) + @xmath129 : delphi coll . , contributed paper pa01 - 109 to the 1996 i.c.h.e.p .
conference ( warsaw ) ; aleph coll . , contributed paper eps0419 to the 1995 european physical society conference ( brussels ) ; opal coll .
, cern / ppe 96 - 099 ( 1996 ) + @xmath135 : delphi coll . , _
_ c * 67 * , 543 ( 1995 ) ; aleph coll . , contributed paper eps0419 to the 1995 european physical society conference ( brussels ) ; opal coll . , cern / ppe 96 - 099 ( 1996 ) + @xmath59 : delphi coll . , _
_ c * 70 * , 371 ( 1996 ) ; aleph coll . , contributed paper eps0419 to the 1995 european physical society conference ( brussels ) ; opal coll . , cern / ppe 96 - 099 ( 1996 ) j. randrup , phys .
c * 68 * ( 2003 ) 031903 .
j. letessier , g. torrieri , s. steinke and j. rafelski , phys .
c * 68*:061901,2003 . | the existence of pentaquarks , namely baryonic states made up of four quarks and one antiquark , became questionable , because the candidates , i.e. the @xmath0 peak , are seen in certain reactions , i.e. p+p collisions , but not in others , i.e. @xmath1 annihilations . in this paper , we estimate the production of @xmath2 and @xmath3 in @xmath1 annihilations at different energies using fermi statistical model as originally proposed in its microcanonical form .
the results is compared with that from pp collisions at sps and rhic energies .
we find that , if pentaquark states exist , the production is highly possible in @xmath1 annihilations .
for example , at lep energy @xmath4=91.2 gev , both @xmath2 and @xmath3 yield more than in pp collisions at sps and rhic energy . |
knowledge of the physical mechanism of diffusion is essential to an understanding of the dynamical complexities of supercooled liquids .
a mechanism should include the collective motions composing the elementary diffusive step , any correlations among the steps , and a statement about whether barrier crossing is important .
if so , distributions of barrier heights , reaction coordinates for barrier crossing , and details of the relevant saddle - barriers are significant .
valuable information is encoded in the @xmath0-dependence of the self diffusion coefficient @xmath12 and of the viscosity @xmath16 ( the stokes - einstein law , @xmath17 , holds until very close to the glass transition at @xmath18 ) . for most normal liquids
@xmath12 exhibits arrhenius @xmath0-dependence . in @xcite
strong liquids this continues all the way down to @xmath18 , while super - arrhenius , with the apparent activation energy @xmath19 increasing with decreasing @xmath0 , sets in before @xmath18 in @xcite fragile liquids .
some exceptions classified as weak by kivelson et .
al @xcite , including simple atomic models such as lennard - jones ( lj ) , show a normal or upper - supercooled range of power - law @xmath20 .
clearly the mechanism of diffusion is @xmath0-dependent , and a straightforward conclusion is that activated potential - energy barrier crossing , or _ hopping _ , plays an important role .
in contrast to chemical reactions where the arrhenius law was first introduced , however , the barriers and reaction coordinates for diffusion have proved elusive . to further complicate matters , free energy or entropic barriers , associated with the difficulty of finding a low - energy pathway ,
may also be important .
it is not straightforward to infer a mechanism from @xmath20 , and thus there has been considerable speculation .
adam and gibbs suggested @xcite that liquids may be decomposed into roughly independent _ cooperatively rearranging regions _
( crr ) containing @xmath21 particles , with super - arrhenius arising from the growth of the crr , and thus the activation free energy , with decreasing @xmath0 .
an influential paper by goldstein @xcite formulated dynamics in terms of the @xmath22-dimensional potential energy @xmath23-surface , or _
landscape_. he proposed the existence of distinct high and low-@xmath0 dynamical regimes , separated by a crossover temperature . at low @xmath0
the system stays for times long compared to a vibrational period in the basins associated with the local minima of the landscape , and diffusion is governed by infrequent hopping of the saddle - barriers connecting adjoining basins .
perhaps the rearrangements of the crr correspond to reaction coordinates on the landscape . at high-@xmath0 a different , freer motion is presumed to occur .
the onset of the low-@xmath0 mechanism is sometimes described as a transition to hopping , but if the high-@xmath0 diffusion constant is arrhenius , implying activation , the change must be more subtle .
the idea of a crossover is central to current thinking about supercooled liquids .
we will pursue the landscape point of view in this paper , so the mechanism of diffusion is naturally discused in terms of the topology .
goldstein s picture has been extended to involve two characteristic temperatures , meaning that the mechanism changes gradually over a crossover range @xmath1 .
interesting features of supercooled dynamics such as super - arrhenius and stretched exponential decay of time correlations first appear at @xmath24 @xcite and @xmath20 extrapolates to zero from above , via power - law fit , at the mode - coupling @xcite temperature @xmath25 .
note that @xmath9 , defined by the extrapolation , seems a useful quantity regardless of the validity of mode - coupling theory .
an alternative formulation , the @xcite frustration limited domain theory of kivelson and co - workers , introduces an upper crossover temperature @xmath26 at which collective dynamics first becomes important .
still lower @xcite lie @xmath18 , the kauzmann temperature @xmath27 and the vtf @xmath28 , but this paper is concerned with temperatures from @xmath24 to somewhat under @xmath9 .
it is difficult to obtain well - equilibrated computer simulation data below @xmath9 , and impossible to approach @xmath18 .
stillinger and weber showed @xcite how to obtain an explicit description of motion over the landscape in a molecular dynamics simulation , by mapping a @xmath22-dimensional liquid configuration @xmath29 to the associated minimum @xmath30 , or _ inherent structure _ ( is ) , with frequent steepest - descent minimizations ( faster conjugate - gradient minimizations are equivalent @xcite at supercooled @xmath0 ) . the configuration space is thus partitioned into the basins of attraction of the is and the rate of _ inherent structure transitions _
( ist ) may be measured .
the is - mapping is a powerful technique for the investigation of supercooled dynamics but it has not yet been exploited to anything like its full potential .
sastry et al .
demonstrated @xcite in the lj mixture that the averaged is energy , @xmath31 , undergoes a sharp drop ( the @xcite slope ) from an upper high-@xmath0 to a lower low-@xmath0 plateau .
they identify the beginning of the drop with @xmath24 , and @xmath9 lies near the apparent bottom .
the bottom energy is cooling - rate dependent and the true value , requiring cooling slow enough to maintain equilibration , remains unknown .
it is appealing , within the landscape point of view , to relate the change in dynamical mechanism to occupation by the system of lower - lying is .
more direct investigations of the influence of the landscape on dynamics are possible .
the minima of the squared potential gradient @xmath32 include all the extrema , or critical points , of @xmath23 , including minima of order @xmath33 and saddles of order @xmath34 .
minimizing @xmath35 defines a mapping to a configuration which need not be a minimum of @xmath23 . with the _ critical point mapping _ ,
cavagna , angelani et al .
@xcite showed that minima dominate below @xmath9 , saddles above .
the is - mapping assigns the system to a minimum even if it is in the upper reaches of the basin near the saddle - barriers ; the critical point mapping does so only if it is actually close to the is .
they suggest @xcite that the crossover is from motion among the minima ( is dynamics ) at @xmath36 to motion among the saddles ( saddle dynamics ) at @xmath37 .
either mapping may be used at any @xmath0 .
reference to saddle dynamics or is dynamics below indicates that it gives the simpler , more physical description .
the _ saddles ruled _ @xcite regime is also one of _
border dynamics_. reaction coordinates leading to several different is intersect at a multidimensional or @xcite monkey saddle , so is boundaries are closely spaced in the vicinity of a saddle .
the system is always near a border , crossings are facile and frequent , and ist are not hops but merely unphysical bookkeeping @xcite events .
a much better understanding of the @xmath37 region is being achieved with the new ideas about saddles and border dynamics .
recently @xcite we gave the first quantitative expression for @xmath12 in terms of the ist rate @xmath38 and the mean - square separation of successive is .
we argued that the long - time slope of the mean - square displacement per degree of freedom ( msd ) of the is - configuration , @xmath39 , is equal to @xmath12 , as is so for the ordinary msd , @xmath40 .
the is - displacement is the sum of the ist - vectors ( separations of successive minima ) , @xmath41 after @xmath42 transitions in time @xmath7 ; for any quantity @xmath43 , the displacement @xmath44 .
a key quantity in this approach is @xcite the ist - vector correlation , @xmath45 . in unit - density lj ,
@xmath46 = 32 atoms , where @xmath47 ( lj units ) , an ist - markov approximation of random walking among the is , predicts @xmath12 accurately for @xmath48 and substantially overestimates @xmath12 for @xmath49 .
the markov approximation is equivalent to the statement that the ist - vectors are uncorrelated , @xmath50 .
our result supports the idea of long sojourns in the basins below @xmath9 , since the system then has time to lose correlation or memory between ist , leading to a random walk . because of the uniquely weak @xmath0-dependence of @xmath12 at @xmath37 in lj , transition to hopping may actually be an accurate description of the crossover in this case .
the overestimate of @xmath12 by the markov approximation at @xmath37 is associated @xcite with a long ranged anticorrelation of successive transition vectors , negative @xmath51 at large @xmath52 , necessary for frequent ist to result in relatively little displacement .
_ anticorrelated ist - vectors are a signature of the bookkeeping ist occuring in border dynamics . _
a problem arises in trying to learn about landscape dynamics with the is - mapping . in the absence of long - range correlations
the configuration space of the liquid should be regarded as a composite of the spaces of @xmath53 independent local regions of @xmath54 particles .
a simple model @xcite uses 3@xmath46 sinusoidal potentials . local regions and adam - gibbs @xcite crr need not be equivalent , but we will treat them as such . with independent ist occurring in each region , and with a transition for the whole system recorded when
any region changes , @xmath55 and the distribution of waiting times between ist , @xmath4 , is correspondingly @xcite , and unphysically , skewed towards short time @xmath56 .
computational constraints then indicate that one should simulate the smallest realistic system ; with sufficiently large @xmath46 an ist will be observed on every time step and no meaningful calculation will be possible .
more fundamentally , the desired properties are those of the crr , the building blocks of the macroscopic liquid , but straightforward simulation yields those of the homogenized composite . with increasing @xmath46
the former become more difficult to disentangle from the latter .
this is also true for static quantities .
for example , whatever the distribution of @xmath57 in a local region , it will be gaussian for the entire system at large @xmath46 .
even for the best choice of @xmath46 some new theory is required to interpret the simulation data . in this article
we further examine the @xmath0-dependent mechanism of diffusion in unit - density lj , @xmath46 = 32 , via three new landscape dynamical entities , and with an initial attempt to derive local relaxation from simulated composite dynamics .
the first is @xmath58 , the msd _ with no contribution from ist_. starting in a thermal ensemble of basins , @xmath2 is calculated from only those trajectories which remain in the initial basin at time @xmath7 .
the second is @xmath3 and the third is the distribution @xmath5 of waiting times in an is _ for a local region or crr_. we present a simple method to obtain @xmath5 from @xmath4 and @xmath53 , estimated from the averaged participation ratio , @xmath59 , of the ist - vectors .
the number of degrees of freedom ( not particles ) participating in an ist is roughly @xmath59 ; thus @xmath60 , @xmath61 .
comparison of @xmath2 , @xmath3 , @xmath40 and @xmath5 casts considerable light upon the relation of diffusion to the landscape . for @xmath62 , where @xmath13 is a characteristic plateau time
, @xmath2 attains a constant value , due to @xcite the finite volume of the basin .
this is true even at high @xmath0 where the system moves freely among the is , since only trajectories which remain in the original is contribute ( thus averaging is difficult ) . at @xmath0
= 1.10 the plateau is reached quickly but a small interval of linear @xmath7-dependence is visible at intermediate times , from which an _ intrabasin diffusion coefficient _ , @xmath8 , may be calculated .
below @xmath63 the diffusive interval in @xmath2 expands ; @xmath24 is estimated as the top of the slope in @xmath31 .
more cooling reveals that the @xmath15 regime is _ intrabasin dominated _ with @xmath11 - the true diffusion constant may be calculated in a single basin !
thus , as for high @xmath0 , diffusion can not consist of a sequence of activated ist hops , but the mechanism is quite different from the high-@xmath0 mechanism .
intrabasin diffusion arises from the roughness of the landscape at the upper elevations of a basin and provides a mechanism for confinement different from inter - basin barriers but nonetheless capable of producing @xcite a markov chain of ist under the right conditions .
we suggest that the elementary steps in this process are transitions among the domains of the saddles connected to the is , ie , intrabasin diffusion is saddle dynamics .
dominated _ diffusion is also is dynamics , since the motion among the is is then a simple random walk ; at intermediate @xmath0 both descriptions are appropriate .
the system does not sense the finite size of the basin until the mean local waiting time , @xmath6 , reaches @xmath13 .
we thus have a new criterion for the onset of the low-@xmath0 is - dynamics mechanism and , indeed , it is found that @xmath6 crosses @xmath13 slightly below @xmath9 .
the analysis would make no sense with an @xmath46-dependent mean time taken from @xmath4 .
more information is obtained by comparison of @xmath3 with @xmath40 .
let @xmath64 , where @xcite @xmath65 is the return vector , the separation of the system from its is . if @xmath66 and @xmath67 are uncorrelated , @xmath40 must lie above @xmath3 .
this is so at low @xmath0 but as @xmath0 increases @xmath3 rises above @xmath40 .
the relation of @xmath3 to @xmath40 , indicating the correlation of @xmath66 and @xmath67 , yields a _ border index _
@xmath68 which complements the information contained in @xmath51 regarding is vs saddle dynamics . in sum
, with some novel applications of the is- mapping we achieve an extremely detailed description of the @xmath0-dependent mechanism of diffusion .
molecular dynamics simulations are performed on unit - density supercooled lj @xcite , @xmath46 = 32 , with the methods described in ref @xcite . as discussed there , and above , a small system should be used for ist dynamics . increasing @xmath46 beyond the number required to solvate a crr merely creates an intractably large ist rate and makes it difficult to extract the properties of a crr from those of the composite landscape .
is @xmath46=32 large enough ?
the diffusion constants are close to those for @xcite @xmath46=108 and @xcite @xmath46 = 256 , with slightly weaker @xmath0-dependence .
the crystal melts at @xmath69 and @xcite fitting @xmath20 yields @xmath47 , while at @xmath46=256 @xcite @xmath70 and ( for modified lj ) @xcite @xmath71 .
the saddle order @xcite @xmath72 extrapolates to zero at @xcite @xmath9 ( from @xmath20 by definition ) , as is so for @xcite @xmath46 = 108 and @xcite for modified lj , @xmath46 = 256 .
the plot @xcite of @xmath31 is similar to that of ref @xcite , reaching the bottom of the landscape at the same @xmath73 , and with the gentler drop from the high-@xmath0 plateau expected for small @xmath46 .
clearly , the essential physical features of the crossover - we make no claim about deeply supercooled liquids - are present at @xmath46=32 and using a larger system would simply fine - tune various numerical estimates at enormous @xcite computational cost . natural lj units will be used throughout . rather than making a single long
md run in the supercooled liquid , we average all calculated quantities over an ensemble of quenches , starting from different high @xmath0 = 5.00 configurations , to avoid @xcite broken ergodicity .
the hot liquid is cooled in one step to a temperature in the 1.20 - 0.60 range .
the system is equilibrated for 2.5 @xmath74 , data are gathered for 62.5 @xmath74 , @xmath0 is decreased by 0.02 and the process is repeated 10 - 25 times , generating a single quench run ; most quenches sampled 16 @xmath0 .
the cooling rate is @xmath75 .
sastry et al .
@xcite found in an lj mixture that quenches with a cooling rate of @xmath76 , close to ours , exhibited @xmath77 of the fall in @xmath31 attained by quenching almost 100 times slower and reached an apparent bottom somewhat below @xmath9 ; this should be adequate for probing the crossover . at @xmath46
= 32 the abrupt decreases in @xmath23 signaling solidification , common at @xmath46 = 256 , do not occur but some quenches develop solid - like pair distributions and these are discarded .
results are averaged over 23 quenches at the lowest @xmath0 and 30 at the highest .
quench - to - quench fluctuations are much larger than any systematic changes over @xmath0 = 0.02 , so we also average results at each @xmath0 with those from the next higher and lower @xmath0 . even so our data are somewhat noisy but the trends are clear for @xmath78 , encompassing the crossover range @xmath79 .
we believe that our results are well averaged and equilibrated down to @xmath80 , and are out of equilibrium ( but well averaged over the set of non - equilibrium configurations allowed by the quench ) at the lowest @xmath0 .
conjugate gradient minimizations are performed every 5 time steps ( @xmath81 = .00125 ) , or 160 minimizations/@xmath74 .
since the range of @xmath82 is from 9.0 ist/@xmath74 at @xmath0 = 1.10 to 0.23 at @xmath0 = 0.34 , this is sufficient .
the equivalence of conjugate - gradient and steepest - descent for the relevant @xmath0 was verified .
the distribution of ist - vector ( logarithmic ) lengths @xmath83 is bimodal , and @xcite to avoid counting contributions of two - level systems and anharmonicities irrelevant to diffusion we record a transition for @xmath83 in the large - displacement lobe only .
similar considerations have @xcite entered the efforts to relate @xmath12 to @xmath84 instantaneous normal modes , where _ non - diffusive _ modes must be discarded . at each @xmath0
we obtain the three msd , @xmath2 , @xmath3 , and @xmath40 ( yielding @xmath12 ) , the distribution of waiting times , @xmath4 , and the averaged participation ratio @xmath59 of the ist - vectors . with the @xmath7-dependent is in hand evaluation of @xmath3 and @xmath59 is straightforward .
the intrabasin msd @xmath2 is found from an ordinary msd algorithm with the condition that , when an ist is detected , the current run is terminated and a new run is begun ( new origin of coordinates , msd=0 ) .
binning the times spent in an is , instead of simply calculating the averaged ist rate , is all that is required for @xmath4 .
distributions of is waiting times @xmath4 are described very well by a kww ( stretched exponential ) function @xmath85 over the entire temperature range .
the kww fit is superior to a sum of two exponentials , even though the latter has one more adjustable parameter .
the average @xmath52 = 0.50 , and deviations appear to be noise with no systematic @xmath0-dependence . to investigate trends in @xmath52 over a broader @xmath0-range we obtained data at normal liquid @xmath0 = 2.00 and also found , to our surprise , @xmath52 = 0.50 .
one hesitates to read very much into the intriguing half - integral value of @xmath52 because @xmath4 is a composite property , dependent upon @xmath53 .
an estimate of @xmath5 may be obtained as follows .
the probability that there is no ist in time @xmath7 is @xmath86 . in a composite ,
no transition means that no crr has a transition , and @xmath87 . equating the two expressions , solving for @xmath88 , and taking @xmath89 yields @xmath90
the averaged participation ratio of the ist - vectors is nearly constant , fluctuating between 19 - 22 over the @xmath0-range with an average of 21 for @xmath21=7.0 atoms in a crr and @xmath53 = 4.6 crr in the simulation box . using the average @xmath53 at all @xmath0 we calculate @xmath91 which are also well described by a kww form , with @xmath0-independent @xmath52 = 0.36 .
some representative @xmath5 from eq [ gloc ] , simulated @xmath4 and fits with @xmath52 = 0.36 and @xmath52 = 0.50 , respectively , are presented in fig [ ggloc ] .
the shift of @xmath5 to longer times is evident .
processes with kww distributions are not well characterized by a single time , but the best compromise is the correlation time , @xmath92 .
fig [ tau ] provides further evidence @xcite that the low-@xmath0 diffusive mechanism sets in at @xmath93 , where @xmath94 begins to increase strongly .
also shown is @xmath95 , the obvious intensive composite corrected time available from the average ist rate , in reasonable agreement with @xmath94 .
let us now view the changes in the mechanism of diffusion with decreasing @xmath0 through the paired conventional and intra - is msd , @xmath40 and @xmath2 , combined with our knowledge of @xmath94 .
the @xmath96 scenario is found at @xmath0 = 1.10 , fig [ msd110 ] , a relatively high ( although supercooled ) temperature @xmath97 where thermal energy is sufficient for the system to move freely among closely - spaced is borders .
the unconstrained msd diverges from @xmath2 at @xmath98 , before it has even fully established its linear diffusive form . for the same reason the trajectories which stay in the basin explore it quickly
; the short time rise in @xmath2 bends over to form the plateau at about @xmath99 with only a glimpse of intermediate-@xmath7 behavior .
even so , almost all trajectories leave the basin before its finite size is felt , with @xmath100 ( thus @xmath2 is noisy at long time ) .
this is illustrated with the inclusion of @xmath91 , scaled to intersect @xmath2 at @xmath6 .
the ist are bookkeeping events as border dynamics prevails , @xmath12 is not @xcite proportional to @xmath38 , there is no effective barrier to leaving a basin and hopping is surely not an apt description .
one might say either that intrabasin dynamics are irrelevant to diffusion , or that no distinction exists between inter- and intra- basin dynamics . at @xmath101 , @xmath2 looks rather different - the difusive interval has grown to the point that it resembles a conventional msd and easily allows estimation of an intrabasin @xmath8 .
the plateau must be reached eventually , but in the current simulation all we can say is @xmath102 .
comparison with @xmath40 shows ( fig [ ddis ] ) that @xmath103 , with @xmath8=0.0045 , @xmath12=0.00714 .
the two msd have a brief interval of overlap at the beginning of their linear regimes but soon diverge , so the mechanism is not governed by intrabasin diffusion and is probably unchanged from @xmath0 = 1.10 .
nonetheless we believe that the emergence of intrabasin diffusion is significant and characteristic of the upper crossover regime .
basins are not harmonic away from the minima and apparently below @xmath24 the higher elevations are rough enough to produce diffusion .
an equivalent statement is that the process is best viewed as a random walk among the saddles @xcite .
configurations in the upper part of a basin are closer to the saddles than to the is and will be assigned accordingly by the critical point mapping . a random walk among saddles connected to a basin
will appear as intrabasin diffusion . at this @xmath0 ordinary diffusion
is also saddle dynamics , so with @xmath103 it follows that unconstrained saddle dynamics is faster than intrabasin saddle dynamics .
the need to utilize an inefficient random walk to escape the basin , and of course the lower thermal energy , has caused the plateau time to increase significantly from its @xmath0 = 1.10 value and @xmath104 holds even more strongly despite an increase in @xmath6 . in our picture of intrabasin diffusion ,
saddle transitions occur while the is does not change .
we previously discussed bookkeeping ist , where the is changes but there is little actual motion . with several
is available from a saddle , motion about the basin of attraction of a single saddle naturally generates bookkeeping ist ; the is changes but the saddle does not . both possibilities , which coexist at intermediate @xmath0 , may be observed with a time - dependent critical point mapping , made possible by @xcite a new , very efficient algorithm .
fig [ sadis ] shows is and saddle energies vs @xmath7 at @xmath105 . in the time interval @xmath106 the system undergoes seven is transitions while assigned to the same saddle ( bookkeeping ist ) . for @xmath107
the is is constant and there are two saddle transitions at @xmath108 , followed by rapid exchange between a pair of saddles starting just after @xmath109 ( intrabasin diffusion ) . cooling to @xmath0 = 0.50
@xmath80 produces a remarkable result .
the period of overlapping linear @xmath7-dependence of @xmath110 and @xmath111 is now substantial , and ( fig [ ddis ] ) * the true diffusion constant may be calculated from trajectories with no ist .
* the system now stays in a basin long enough that on average intrabasin diffusion is fully developed before an ist occurs , and before the plateau is reached ( although @xmath6 is now approaching @xmath13 , fig [ ttpl ] ) . under these circumstances diffusion
is _ intrabasin dominated _ , and the ist simply act to keep the process going , ie , to avoid the influence of finite basin size .
intrabasin and unconstrained saddle dynamics ( ordinary diffusion ) are indistinguishable .
we thus have some new perspective on the changes around @xmath9 . there
the chain of ist becomes @xcite a markov process , and an obvious explanation is that sufficiently long confinement of the system in a basin allows successive ist to lose correlation . while conventionally confinement arises from energy barriers , we find that the system diffuses to the border and crosses with ease .
confinement is produced by diffusion itself ; the time required to escape by the random walk is long even absent a barrier at the border . a sufficiently slow walking about the saddles connected to a basin can produce a markov process among the is themselves , and is dynamics is also a good description of the intrabasin dominated regime . upon close examination ,
the crossover is not simply from saddle dynamics at high @xmath0 to is dynamics at low @xmath0 .
there is a third , intermediate-@xmath0 region where both descriptions are correct , defined by the condition @xmath112 .
arrhenius plots of @xmath8 and @xmath12 are displayed in fig [ ddis ] .
estimates of @xmath8 are possible for @xmath113 .
linear behavior of @xmath2 is not cleanly visible below @xmath0 = 0.38 and is marginal at @xmath0 = 1.10 ; we expect it disappears at slightly higher @xmath0 , but collecting data where trajectories leave a basin so quickly is difficult .
the intrabasin dominated @xmath112 range , marked by arrows , is approximately @xmath114 .
starting from high @xmath0 , the upper limit of intrabasin dominance is reached when two conditions hold : @xmath0 is low enough for @xmath2 to have a diffusive @xmath7-interval , and @xmath6 adequately exceeds the time at which that interval begins . for the lower limit , either of two conditions may be true : @xmath6 substantially exceeds @xmath13 , so a barrier to leaving the basin exists and activation becomes important , or the linear portion of @xmath2 shrinks so that it no longer represents the intrabasin motion .
in fact , both appear to set in gradually and simultaneously . at our lowest temperature ,
@xmath0 = 0.34 ( fig [ msd034 ] ) , @xmath2 rises quickly to the plateau just as at the high @xmath0 = 1.10 ; a linear region of this curve can not be identified and @xmath13 is now back within the 6.25 @xmath74 window . quantitative estimates of @xmath115 , along with @xmath94 , are shown in fig [ ttpl ] ; at intermediate @xmath0 all we can say is @xmath116 . while @xmath13 is short at high @xmath0 because of the abundant thermal energy , it is short at low @xmath0 because the system occupies the lower regions of the basin only .
thus a ) the plateau is reached after a shorter displacement , and b ) the intra - basin motion is mostly fast harmonic oscillations , not diffusion .
the @xmath13 and @xmath6 curves cross at @xmath117 , and we propose @xmath118 as a new way to define a crossover temperature . below @xmath0 = 0.51
the system explores the accessible portion of the basin more quickly but exits more slowly , indicating barriers at the border .
gradually saddle dynamics loses out to is dynamics as the physical description , @xmath119 below @xmath0 = 0.46 , and barrier crossings assume the rate - limiting role for diffusion .
in addition to the standard analysis of @xmath20 and the results @xcite of very recent work , we now have two new indicators pointing to a change in the mechanism of diffusion at @xmath93 .
the local waiting time @xmath6 crosses @xmath13 at @xmath0 = 0.51 , and intrabasin dominance sets in at @xmath0 = 0.52 with @xmath11 for @xmath120 .
however the crossover develops over a wide temperature range , and involves much more than the presence or absence of hopping , as explained in the last few paragraphs .
we hope it is clear that the is - mapping is a powerful tool for studying the dynamical crossover .
ist - vectors are anticorrelated @xcite at high @xmath0 , and correlation decreases through the crossover range until the ist - markov approximation becomes accurate below @xmath9 .
a substantial negative value of @xmath51 is a sign of saddle or border dynamics , while @xmath121 indicates is or hopping dynamics . is borders are closely spaced near a multidimensional saddle .
anticorrelation is required for the bookkeeping ist arising from a small motion through such a region to not , wrongly , predict a large motion .
it implies that saddle dynamics is a more physical description than is dynamics .
a simple process appears complicated when formulated with inappropriate elementary steps , and loss of correlation signals the onset of the low-@xmath0 mechanism , which is truly is dynamics .
analysis of the msd within the basin of attraction of an is , @xmath2 , proved quite fruitful .
we will now see that such is also the case for the msd of the is itself , @xmath3 , leading to another informative correlation , that of the return vector and the is configuration . with the definitions from the introduction ,
@xmath122 it follows that @xmath123 unless the is displacement , @xmath124 , and the return vector displacement , @xmath125 , are anticorrelated .
fig [ msds ] shows that at @xmath0 = 1.00 , @xmath3 lies _ above _ @xmath40 , they nearly sumerimpose at @xmath0 = 0.80 , and the obvious @xmath123 holds at @xmath126 . thus the pattern of anticorrelation of ist - vectors @xmath127 at high @xmath0 , decreasing with decreasing @xmath0 ,
is repeated with the @xmath128 correlation . for a more quantitative analysis
we have calculated @xmath129 .
one anticipates a rapid decay of @xmath130 leading to @xmath131 and this expectation is correct .
the correlation @xmath132 , determined from @xmath40 , @xmath3 , and @xmath129 via eq [ msdeq ] , decays quickly ( fig [ cru ] ) to an asymptotic negative value which decreases in amplitude with decreasing @xmath0 .
we suggest that @xmath133 is a direct measure of the degree of _ borderism _ - the extent to which the system lives on the is - borders @xcite . expressing @xmath124 as the sum of the ist - vectors the most important contributions are evident ,
@xmath134 the first term on the rhs may be understood with fig [ anticor ] , a one - dimensional illustration of the situation at time @xmath7 , just after the @xmath135th ist .
the system has moved from the left- to the right - hand basin and it follows that @xmath136 and @xmath65 point in opposite directions , with a negative contribution to @xmath133 .
for the second term on the rhs , set @xmath7 = 0 in fig [ anticor ] .
the first ist will occur from right to left in the future so change @xmath136 to @xmath137 and reverse its direction .
the vectors now point in the same direction so with the minus sign in eq [ crq ] this contribution to @xmath133 is also negative . in the @xmath22 dimensional configuration space ( or @xmath138 dimensional local region space )
the vectors will be distributed about the @xmath83 = 1 relative orientations , reducing the dot products .
the geometrical arguments apply when @xmath139 is evaluated just before , and @xmath65 just after , an ist . for border dynamics these conditions always hold and @xmath140 . on the other hand
if the system oscillates about the is for long periods between ist , the low-@xmath0 scenario , the return vector has no special relation to any ist vector at any time and @xmath141 .
the correlation of @xmath124 and @xmath125 is a maximum for pure border dynamics and vanishes in a hopping model .
the interesting part of @xmath133 is its asymptotic value and we define @xmath142 as the _ border index_. the @xmath0-dependence of the border index is shown in fig [ bordr ] .
it extrapolates to zero via power - law fit for @xmath143 at @xmath0 = 0.53 , yielding another physically appealing route to a crossover temperature .
the fit range differs from that used for @xmath9 but @xmath68 is close enough ( power = 0.74 ) to linear in @xmath0 that we expect only a small effect .
in the foregoing we have examined the @xmath0-dependent physical mechanism of diffusion in unit - density lj . for some perspective , it is useful to ask why this is a nontrivial problem .
lj is classified @xcite as weak / nonfragile , indicating weak @xmath0-dependence at high @xmath0 and a constant activation energy at low @xmath0 .
the simulated diffusion constant is indeed well represented by @xmath144 for @xmath78 ; @xmath145 = 1.16 .
the factor of @xmath0 is appropriate because @xmath146 at high @xmath0 and constant density .
linear or other multiplicative powers are not very important when fitting over several decades but we have only @xmath147 and they significantly influence the goodness of fit and the value obtained for the activation energy . despite the simplicity of eq [ fit ] , fitting @xcite the same data to a power law for @xmath148 yields a reasonable @xcite @xmath9 .
one might well wonder if anything is going on beyond a gradual change from @xmath149 to @xmath150 . that is a bare - bones prescription for a transition to hopping , consistent with the ist approaching a markov chain @xcite and the saddle order extrapolating to zero @xcite .
however it then seems odd that @xmath145 = 1.16 does not appear as a characteristic dynamical temperature instead of various @xmath93 , and the sharp drop in @xmath31 around @xmath9 suggests that the crossover is more interesting . to investigate this crucial point
we have begun @xcite to catalog the saddles and barrier heights in unit - density lj . starting from the thermal configuration the is is determined and the connected first - order saddles are found @xcite by eigenmode following .
steepest descent from the saddle yields barrier heights @xmath151 , which may be collected as a function of is energy @xcite or of the original temperature .
preliminary results @xcite are that the barrier heights increase through the slope region @xcite of @xmath31 from @xmath152 = 2.8 to @xmath153 = 3.7 ; @xmath0 is less than the averaged barrier height for the entire range of this study .
the same result has recently been obtained @xcite in a soft - sphere mixture .
comparisons of barrier heights to @xmath0 must be made carefully .
most discussions of barrier crossing assume that @xmath151 is much greater then @xmath0 , but that does not apply here and one might argue that the relevant height is @xmath154 . even with such a modification , average barrier heights remain larger than @xmath145 , indicating - along with the new results found above - that the mechanism of diffusion in lj is highly non - trivial .
there are two explanations for the clear presence of a high-@xmath0 mechanism with free motion among the basins at @xmath49 despite an apparent requirement for activated hopping .
one is the broad distribution of barrier heights , reflected in the low @xmath155 for the kww distribution of local waiting times .
even if @xmath0 falls below @xmath156 , there will be no transition to low-@xmath0 dynamics if enough low barriers remain .
the average alone does not adequately describe the physics .
furthermore there are several ways to perform the average @xcite and a physically motivated choice must be made .
for example a calculation of @xmath156 including saddles of all orders would be nonsense , since high barriers which are never visited in thermal motion would dominate .
averaging over first order saddles eliminates many irrelevant high barriers , but not necessarily all .
other topological features such as the connectivity of the is and saddles should be important as well .
the landscape determines @xmath9 and more generally @xmath20 through the interplay of several physically significant effects .
note that one important landscape quantity , @xmath31 , is a thermodynamic property of the basins .
one can imagine different liquids with the same @xmath31 and different t - dependent distributions of barrier heights , giving rise to quite different activation energies and other aspects of dynamics .
another is found in cavagna s @xcite argument that barrriers are irrelevant while the saddle mechanism dominates , so the action of a high-@xmath0 mechanism while @xmath0 is less than @xmath157 need present no conundrum .
then , when the saddle mechanism shuts down and activation becomes required at @xmath93 , the barrier heights have already completed most of their growth .
the system will exhibit approximate arrhenius behavior upon further cooling .
cavagna gives a fragile scenario where the barriers are already @xmath158 at @xmath9 , for a very strong increase in relaxation time .
however although we have @xmath159 , @xmath145 is only @xmath97 , presumable due to the importance of low - barrier paths . relatively weak arrhenius behavior at @xmath160 results , consistent with lj being @xcite a weak / nonfragile liquid .
even if @xmath161 is not meaningful at higher @xmath0 , eq [ fit ] can still fit the data because the arrhenius factor is approaching its @xmath145-independent high-@xmath0 limit .
saddle dynamics , in which @xmath20 is @xcite proportional to the number of @xmath84 instantaneous normal modes or @xcite the saddle order @xmath162 , apparently is consistent with @xmath146 .
thus eq [ fit ] can represent the full @xmath0-range , but the empirical activation energy is a compromise resulting from the fitting process with no simple relation to a barrier height . in either case
the crossover is seen to be a subtle , complex process .
application of the is - mapping has revealed far more then could be deduced from @xmath20 alone , including : 1 .
intrabasin and intrabasin _ dominated _ diffusion , associated with saddle dynamics 2 .
the distribution of local waiting times , well behaved as @xmath163 , calculated from the simulated distribution with the idea that the configuration space is a composite 3 . a new , physically appealing criterion for dynamical crossover , @xmath14 4 .
a quantitative indicator @xmath68 of the dominance of saddle or border dynamics ( borderism ) , which also yields a crossover temperature .
the value of @xmath164 , from a power - law fit to @xmath20 , is now just one of many pieces of evidence that a dynamical crossover occurs at @xmath93 . from recent prior work @xcite : 1 .
is energy @xmath31 is undergoing its steepest fall .
the true bottom is uncertain due to cooling - rate dependence and point 1 . only identifies the general vicinity of the crossover .
2 . saddle order @xmath72 also extrapolates to zero at @xmath0 = 0.52 3 .
ist - markov approximation becomes accurate at @xmath15 . in this paper
we have added : 4 .
@xmath11 , diffusion is intrabasin dominated , for @xmath120 .
local waiting time reaches time to explore basin , @xmath14 , at @xmath0 = 0.51 .
border index @xmath68 extrapolates to zero , @xmath0 = 0.53 .
we suggest that points 3 - 5 are particularly direct evidence that the mechanism of diffusion is changing .
they are based on ideas about the mechanism itself , while a fit to @xmath20 produces a characteristic temperature based upon the symptoms. although 1 - 6 focus attention on a narrow range @xmath165 the crossover is gradual and might be thought to take place over @xmath113 , the interval where intrabasin diffusion can be detected , encompassing the slope of @xmath31 .
this agrees roughly with having the crossover begin at @xmath24 .
we believe it significant that intrabasin diffusion can produce confinement for times long enough that the ist become a markov chain , _ without _ invoking barriers .
thus diffusive confinement and confinement by barriers become two distinct features of the mechanism .
since intrabasin diffusion can be explained as saddle dynamics , our work fits with the growing recognition that the saddles of the landscape need to be considered on an even footing with the is .
there is no contradiction in having the intrabasin diffusion and ist - markov regimes overlap . in the latter case @xmath166 , and in the former , since escape from a basin requires diffusion to the border , @xmath167 .
our simulation model exhibits most of the physical features currently being discussed for moderately supercooled dynamics , as well as some found here for the first time .
there is much to be learned from lj , @xmath46 = 32 , although of course we intend to apply our methods to other liquids .
it is difficult to achieve equilibrium below @xmath9 .
note that , even though we report data down to @xmath0 = 0.34 , only one significant @xmath0 appearing in the above estimates is below 0.50 , the end of intrabasin dominance at @xmath0 = 0.46 .
our interest is the crossover and the lowest , probably non - equilibrated , @xmath0 are not required for the principal conslusions .
we would like to thank frank stillinger for valuable discussions .
this work was supported by nsf grants che9708055 and che0090975 . c. a. angell _ science _ * 267 * , 1924 ( 1995 ) .
m. ferrer , c. lawrence , b. demirjian , d. kivelson , g. tarjus and c. alba - simionesco , j. chem .
phys . * 108 * , 8010 ( 1998 ) .
g. adam and j. h. gibbs , j. chem .
phys . * 43 * , 139 ( 1965 ) .
m. goldstein , j. chem .
phys . * 51 * , 3728 ( 1969 ) . c. a. angell , b. richards and v. velikov , j. phys .
matter * 11 * , a75 ( 1999 ) .
w. goetze and l. sjogren , rep .
. phys . * 55 * , 241 ( 1992 ) .
p. viot , g. tarjus and d. kivelson , j. chem . phys . * 112 * , 10368 ( 2000 ) . f. h. stillinger and t. a. weber , phys .
a * 28 * , 2408 ( 1983 ) ; _ science _ * 225 * , 983 ( 1984 ) .
f. h. stillinger , _ science _ * 267 * , 1935 ( 1995 ) . t. keyes and j. chowdhary , phys .
e * 64 * , 032201 ( 2001 ) .
s. sastry , p. debenedetti and f. h. stillinger , _ nature _ * 393 * , 554 ( 1998 ) .
a. cavagna , europhys .
lett . * 53 * , 490 ( 2001 ) l. angelani , r. di leonardo , g. ruocco , a. scala and f. sciortino , phys .
* 85 * , 5356 ( 2000 ) .
b. madan and t. keyes , j. chem .
phys . * 98 * , 3342 ( 1993 ) .
r. laviolette and f. h. stillinger , j. chem . phys . *
83 * , 4079 ( 1985 ) . t. keyes , j. chem . phys . * 101 * , 5081 ( 1994 ) ; phys . rev .
e * 59 * , 3207 ( 1990 ) .
j. chowdhary and t. keyes , phys .
e , submitted ( 2001 ) .
j. chowdhary and t. keyes , unpublished ( 2001 ) .
d. frantz , d. freeman and j. doll , j. chem .
phys . * 93 * , 2769 ( 1999 ) ; i. andricioaei and j. e. straub , j. chem . phys . * 107 * , 9117 ( 1997 ) . c. donati , f. sciortino and p. tartaglia , phys .
* 85 * , 1464 ( 2000 ) .
wu - xiong li and t. keyes , j. chem .
* 111 * , 5503 ( 1999 ) ; e. la nave , a. scala , f. starr , f. sciortino and h. e. stanley , phys .
lett , * 84 * , 4605 ( 2000 ) .
j. kurchan and l. laloux , j. phys .
a * 29 * , 1929 ( 1996 ) .
t. grigera , a. cavagna , i. giardina and g. parisi , preprint ( 2001 ) . | the mechanism of diffusion in supercooled liquids is investigated from the potential energy _
landscape _
point of view , with emphasis on the crossover from high- to low-@xmath0 dynamics over the range @xmath1 .
molecular dynamics simulations with a time dependent mapping to the associated local mininum or _ inherent structure _ ( is ) are performed on unit - density lennard - jones ( lj ) .
new dynamical quantities introduced include @xmath2 , the mean - square displacement ( msd ) within a basin of attraction of an is , @xmath3 , the msd of the is itself , and @xmath4 , the distribution of is waiting times .
the configuration space is treated as a composite of the contributions of cooperative local regions , and a method is given to obtain the physically meaningful @xmath5 and mean waiting time @xmath6 from @xmath4 .
a new understanding of the crossover is obtained in terms of @xmath2 and @xmath6 . at intermediate @xmath0 ,
@xmath2 posesses an interval of linear @xmath7-dependence allowing calculation of an intrabasin diffusion constant @xmath8 .
near @xmath9 , where intrabasin diffusion is well established for @xmath10 , diffusion is _ intrabasin dominated _ with @xmath11 ; _ @xmath12 may be calculated within a basin_. below @xmath9 , @xmath6 exceeds the time , @xmath13 , needed for the system to explore the basin , indicating the action of barriers at the border ; @xmath14 is a criterion for a transition to activated hopping .
intrabasin diffusion provides a means of confinement not involving barriers and plays a key role in dynamics above @xmath9 .
the distinction between motion among the is ( is dynamics ) below @xmath9 and _ saddle , or border dynamics _ above @xmath9 , where the system is always close to one of the saddle - barriers connecting the basins and is boundaries are closely spaced and easily crossed , is discussed .
a _
border index _ is introduced based upon the relation of @xmath3 to the conventional msd , and shown to vanish at @xmath15 .
it is proposed that intrabasin diffusion is a manifestation of saddle dynamics . |
a two - generator subgroup @xmath1 of @xmath2 is determined up to conjugacy by its parameters @xmath3 , @xmath4 , and @xmath5 - 2 $ ] whenever @xmath6 @xcite .
so the conjugacy class of an ordered pair @xmath7 can be identified with a point in the parameter space @xmath8 whenever @xmath6 .
the subspace @xmath9 of @xmath10 that corresponds to the discrete non - elementary groups @xmath1 is called the _ parameter space of two - generator kleinian groups_.
note that a two - generator kleinian group @xmath11 can be represented by several points in @xmath9 , since the same group can have different generating pairs . among all two - generator subgroups of @xmath2
, we distinguish the class of @xmath12 _ groups _ ( two - generator groups with real parameters ) : @xmath13 the aim of this paper is to completely determine all points in @xmath10 that are parameters for the discrete non - elementary @xmath14 groups with one generator parabolic : @xmath15 where @xmath16 denotes the class of all discrete non - elementary @xmath14 groups .
geometrically , @xmath17 is a two - dimensional slice through the six - dimensional parameter space @xmath9 .
the slice @xmath17 intersects the well - known riley slice @xmath18 , @xmath19 , which consists of all kleinian groups generated by two parabolics .
consider the sequence of slices @xmath20 , where @xmath21 the first slice @xmath22 of this sequence is of great interest in the theory of discrete groups .
this slice consists of all parameters for discrete @xmath14 groups with an elliptic generator of order 2 and was investigated in @xcite .
it was shown that if @xmath23 has parameters @xmath24 , then there exists a group @xmath25 with parameters @xmath26 such that if @xmath27 , then @xmath25 is discrete whenever @xmath23 is .
hence , the slice @xmath22 gives necessary discreteness conditions for a group with parameters @xmath24 , where @xmath28 and @xmath29 are real .
it follows that every @xmath30 with @xmath31 , including @xmath17 , is a subset of @xmath22 .
since a parabolic element can be viewed as the limit of a sequence of primitive elliptic elements of order @xmath32 as @xmath33 , the following two questions for @xmath34 and @xmath17 naturally arise .
* is it true that for every point @xmath35 there exists a sequence @xmath36 with @xmath37 that converges to @xmath38 ? *
is it true that for each @xmath39 there exists @xmath40 such that the @xmath41-neighbourhood of @xmath17 contains @xmath30 for all @xmath42 ?
note that the structure of @xmath30 for @xmath31 is unknown .
we work out @xmath17 by splitting the plane @xmath43 into several parts .
it turns out that @xmath1 has an invariant plane in one of the following cases : ( 1 ) @xmath44 and @xmath45 ; ( 2 ) @xmath46 and @xmath47 .
such discrete groups were investigated , for example , in @xcite and @xcite , respectively . if @xmath44 and @xmath48
, then @xmath11 is truly spatial ( non - elementary and without invariant plane ) and this case is treated in @xcite .
we get these dicreteness criteria together and transform them into conditions on @xmath28 and @xmath29 if it was not done before .
so the last case to consider is when @xmath46 and @xmath49 . in this case
@xmath11 is truly spatial with @xmath50 @xmath0-loxodromic .
we complete the study of the slice @xmath17 by giving discreteness criteria for such groups .
the paper is organised as follows . in section 2 ,
discreteness criteria are given for truly spatial @xmath14 groups @xmath11 generated by a @xmath0-loxodromic and a parabolic elements ( theorems [ criterion_psl ] and [ criterion_par ] ) . in section 3 , for each such discrete @xmath11 we obtain a presentation and the kleinian orbifold @xmath51 ( theorem [ groups ] ) .
section 4 is devoted to the analysis of the parameter space .
we completely describe the slice @xmath17 by giving explicit formulas for the parameters @xmath28 and @xmath29 .
we also program the obtained formulas in the package maple 7.0 and plot a part of @xmath17 on the @xmath43-plane to give an idea of how it looks like .
recall that an element @xmath52 with real @xmath53 is _ elliptic _ , _ parabolic _ , _ hyperbolic _ , or _ @xmath0-loxodromic _ according to whether @xmath54 , @xmath55 , @xmath56 , or @xmath57 .
if @xmath58 , then @xmath50 is called _ strictly loxodromic_. an elliptic element @xmath50 of order @xmath32 is said to be _ non - primitive _ if it is a rotation through @xmath59 , where @xmath60 and @xmath32 are coprime ( @xmath61 ) .
if @xmath50 is a rotation through @xmath62 , then it is called _
primitive_. [ criterion_psl ]
let @xmath52 be a @xmath0-loxodromic element , @xmath63 be a parabolic element , and let @xmath1 be a non - elementary @xmath14 group without invariant plane
. then * there exist unique elements @xmath64 such that @xmath65 and @xmath66 , @xmath67 and @xmath68 . *
the group @xmath11 is discrete if and only if one of the following conditions holds : * * @xmath69 is either a hyperbolic , or parabolic , or primitive elliptic element of even order @xmath70 , and @xmath71 is either a hyperbolic , or parabolic , or primitive elliptic element of order @xmath72 ; * * @xmath69 is a primitive elliptic element of odd order @xmath73 , and @xmath74 is either a hyperbolic , or parabolic , or primitive elliptic element of order @xmath75
. we will construct a group @xmath76 that contains @xmath1 as a subgroup of finite index .
the idea is to find @xmath76 so that a fundamental polyhedron for a discrete @xmath76 can be easily constructed
. it will be clear from the construction that @xmath11 is commensurable with a reflection group which either coincides with @xmath76 or is an index 2 subgroup of @xmath76 .
the construction presented below will be used throughout sections 2 and 3 and we shall use the notation introduced here .
let @xmath50 and @xmath77 be as in the statement of theorem [ criterion_psl ] . since @xmath11 is a non - elementary @xmath14 group without invariant plane , there exists an invariant plane of @xmath77 , say @xmath78 , which is orthogonal to the axis of @xmath50 ( * ? ? ?
* theorem 2 ) .
denote by @xmath79 the fixed point of @xmath77 and by @xmath80 the plane that passes through @xmath79 and @xmath50 ( we denote elements and their axes by the same letters when it does not lead to any confusion ) .
note that @xmath50 keeps @xmath80 invariant . since @xmath50 is orthogonal to @xmath78 , @xmath80 is also orthogonal to @xmath78 .
let @xmath81 be the half - turn with the axis @xmath82 .
then @xmath81 passes through @xmath79 and is orthogonal to @xmath50 .
let @xmath83 and @xmath84 be half - turns such that @xmath85 then @xmath83 is orthogonal to @xmath80 and @xmath84 lies in @xmath78 .
let @xmath86 be the plane passing through @xmath84 orthogonally to @xmath78 and let @xmath87 .
the planes @xmath86 and @xmath80 are parallel and @xmath79 is their common point on the boundary @xmath88 . since @xmath83 is orthogonal to @xmath80 , the planes @xmath89 and @xmath80 are also parallel with the common point @xmath90 on @xmath88 . since @xmath91 , the planes @xmath80 , @xmath89 , and @xmath86 do not have a common point in @xmath92 .
therefore , there exists a unique plane @xmath93 orthogonal to all @xmath80 , @xmath89 , and @xmath86 .
it is clear that @xmath94 .
consider two extensions of @xmath11 : @xmath95 and @xmath96 .
( we denote the reflection in a plane @xmath97 by @xmath98 . )
one can show that @xmath99 and @xmath100 . from ( [ efeg ] )
, it follows that @xmath101 contains @xmath11 as a subgroup of index at most 2 .
moreover , @xmath101 is the orientation preserving subgroup of @xmath76 and , hence , @xmath76 contains @xmath11 as a subgroup of finite index .
therefore , @xmath11 , @xmath101 , and @xmath76 are either all discrete , or all non - discrete .
we then concentrate on the group @xmath76 .
, width=10 ] let @xmath102 be the infinite volume polyhedron bounded by @xmath78 , @xmath80 , @xmath86 , @xmath89 , and @xmath93 . @xmath102 has five right dihedral angles ( between faces lying in @xmath78 and @xmath80 , @xmath78 and @xmath86 , @xmath93 and @xmath80 , @xmath93 and @xmath86 , and @xmath93 and @xmath89 ) .
the plane @xmath89 may either intersect with , or be parallel to , or be disjoint from each of @xmath86 and @xmath78 . if @xmath89 and @xmath86 intersect , then we denote the dihedral angle of @xmath102 between them by @xmath103 , where @xmath104 , @xmath105 is not necessary an integer .
we keep the notation @xmath103 taking @xmath106 and @xmath107 for parallel or disjoint @xmath89 and @xmath86 , respectively .
similarly , we denote the `` dihedral angle '' between @xmath78 and @xmath89 by @xmath108 , where @xmath109 is real , @xmath110 , or @xmath111 .
( we regard @xmath112 , @xmath113 , @xmath114 , @xmath115 for any positive real @xmath38 . )
@xmath102 exists in @xmath116 for all @xmath104 and @xmath109 by @xcite . in figure [ fund_poly1 ] ,
@xmath102 is drawn under assumption that @xmath117 , @xmath118 , and @xmath119 .
the shaded triangle shows the hyperbolic plane orthogonal to @xmath78 , @xmath89 , and @xmath80 .
note that this plane is not a face of @xmath102 and is shown only to underline the combinatorial structure of @xmath102 . in figures
, we do not label dihedral angles of @xmath120 in order to not overload the picture .
suppose now that @xmath117 , that is @xmath89 and @xmath86 intersect .
let @xmath121 be the plane passing through @xmath83 orthogonally to @xmath93 .
then @xmath121 is orthogonal to @xmath80 .
one can see that @xmath122 and @xmath121 is the bisector of the dihedral angle of @xmath102 made by @xmath86 and @xmath89 .
let @xmath123 be the polyhedron bounded by @xmath78 , @xmath86 , @xmath80 , @xmath93 , and @xmath121 .
@xmath123 has six dihedral angles of @xmath120 ; the dihedral angle between @xmath86 and @xmath121 is equal to @xmath124 with @xmath125 . denote the `` dihedral angle '' between @xmath78 and @xmath121 by @xmath126 , where @xmath127 is real , @xmath128 , or @xmath129 .
@xmath123 exists in @xmath116 for all @xmath104 and @xmath127 by @xcite .
note that @xmath130 is not necessary in @xmath76 , but if it is and if @xmath76 is discrete , then we will see that @xmath123 is a fundamental polyhedron for @xmath76 . in figure [ fund_poly2 ] , @xmath123 is drawn under assumption that @xmath131 .
, width=6 ] [ hs ] let @xmath52 be a @xmath0-loxodromic element , @xmath63 be a parabolic element , and let @xmath1 be a non - elementary @xmath14 group without invariant plane . then there exist unique elements
@xmath64 such that * @xmath65 and @xmath66 , * @xmath67 and @xmath68 . moreover , the elements @xmath69 and @xmath71 are not strictly loxodromic . first , note that @xmath132 and @xmath133 . therefore , @xmath134 let us show that if we take @xmath135 , then the assertion ( 1 ) of the lemma hold .
indeed , @xmath136 moreover , @xmath137 . since @xmath121 and @xmath80 are orthogonal , @xmath138 .
hence , @xmath66 .
note also that since @xmath69 is a product of two reflections , @xmath69 is not strictly loxodromic .
now let us show that @xmath69 is unique .
the element @xmath139 is uniquely determined as an element of @xmath2 .
if @xmath139 is parabolic , it has only one square root @xmath69 .
suppose that @xmath139 is hyperbolic .
then it has exactly two square roots , one of which is @xmath69 defined above and the other , denoted @xmath140 , is a @xmath0-loxodromic element with the same axis and translation length as @xmath69 .
clearly , @xmath141 .
if @xmath139 is elliptic , then it also has two square roots @xmath69 and @xmath140 , both are elliptic elements .
the element @xmath140 is elliptic with the same axis as @xmath69 and with rotation angle @xmath142 , while @xmath69 is a rotation through @xmath103 in the opposite direction .
again , @xmath141 .
now we take @xmath143 then @xmath144 these two conditions determine @xmath71 uniquely .
note that the elements @xmath145 defined in lemma [ hs ] determine combinatorial and metric structures of @xmath102 .
for example , if @xmath69 is elliptic , then its rotation angle is equal to the dihedral angle of @xmath102 between @xmath89 and @xmath86 . if @xmath71 is elliptic , then its rotation angle is equal to the doubled dihedral angle of @xmath102 between @xmath78 and @xmath89 .
vice versa , if the metric structure of @xmath102 is fixed , then the types of elements @xmath69 and @xmath71 can be determined .
the same can be said about @xmath123 and the elements @xmath69 and @xmath74 .
the element @xmath74 is responsible for the mutual position of the planes @xmath78 and @xmath121 ( see the proof of lemma [ h1h2 ] ) .
lemmas [ h1][h1h2 ] below give some necessary conditions for discreteness of @xmath11 via conditions on elements @xmath69 and @xmath71 .
one needs to keep in mind the connection between these elements and the polyhedra @xmath102 and @xmath123 .
[ h1 ] if @xmath11 is discrete , then @xmath69 is either a hyperbolic , or parabolic , or primitive elliptic element of order @xmath73 . the subgroup @xmath146 of @xmath11 keeps @xmath93 invariant and is conjugate to a subgroup of @xmath147 .
since @xmath11 is discrete , @xmath148 must be discrete . by @xcite or @xcite ,
the group @xmath148 is discrete if and only if either \(1 ) @xmath149 is a hyperbolic , or a parabolic , or a primitive elliptic element , or \(2 ) @xmath69 is a primitive elliptic element of odd order @xmath105 , where @xmath73 . if @xmath150 is parabolic of hyperbolic , then @xmath69 is parabolic or hyperbolic , respectively .
if @xmath150 is a primitive elliptic element , then @xmath69 is a primitive elliptic of even order @xmath70 .
[ h2 ] if @xmath11 is discrete , then @xmath71 is either a hyperbolic , or parabolic , or primitive elliptic element of order @xmath72 .
let @xmath97 be the plane orthogonal to @xmath78 , @xmath89 , and @xmath80 .
the subgroup @xmath151 of @xmath101 keeps the plane @xmath97 invariant and is conjugate to a subgroup of @xmath147 . by @xcite ,
@xmath148 is discrete if and only if @xmath152 is either a hyperbolic , or parabolic , or primitive elliptic element of order @xmath72 .
[ h1h2 ] if @xmath11 is discrete and @xmath69 is a primitive elliptic element of odd order , then @xmath74 is either a hyperbolic , or parabolic , or primitive elliptic element of order @xmath75 .
recall that @xmath153 . since @xmath69 has odd order and @xmath154 , @xmath155 . since , moreover , @xmath156 , @xmath157 , and @xmath158 , both @xmath130 and @xmath159 are also in @xmath76 .
further , since the plane @xmath121 is orthogonal to @xmath80 , the group @xmath160 keeps @xmath80 invariant and is conjugate to a subgroup of @xmath147 .
it is clear that @xmath160 is discrete if and only if @xmath161 is a hyperbolic , parabolic , or primitive elliptic element of order @xmath75 @xcite .
_ proof of theorem [ criterion_psl ] .
_ lemma [ hs ] proves existence and uniqueness of elements @xmath69 and @xmath71 .
now we prove part ( 2 ) of the theorem .
if @xmath11 is discrete then @xmath69 is either a hyperbolic , or parabolic , or primitive elliptic element of order @xmath73 by lemma [ h1 ] .
we split the discrete groups @xmath11 into two families .
the first family consists of those groups for which @xmath69 is hyperbolic , parabolic , or primitive elliptic of even order .
by lemma [ h2 ] , for these groups @xmath71 is a hyperbolic , parabolic , or primitive elliptic element .
the second family consists of the discrete groups with @xmath69 elliptic of odd order .
then by lemma [ h1h2 ] , @xmath74 is a hyperbolic , or parabolic , or primitive elliptic element of order @xmath75 .
( note that in this case @xmath71 is necessarily hyperbolic or primitive elliptic . )
so if @xmath11 is discrete , then either ( 2)(i ) or ( 2)(ii ) of theorem [ criterion_psl ] can occur .
clearly , if neither ( 2)(i ) nor ( 2)(ii ) holds , then @xmath11 is not discrete by lemmas [ h1][h1h2 ] .
now prove that each of ( 2)(i ) and ( 2)(ii ) is a sufficient condition for @xmath11 to be discrete .
in each of the two cases we will give a fundamental polyhedron for @xmath76 to show , by using the poincar polyhedron theorem @xcite , that @xmath76 is discrete .
suppose that ( 2)(i ) holds .
then since @xmath105 is even , the group @xmath162 generated by the side pairing transformations @xmath163 , @xmath164 , @xmath165 , @xmath166 , and @xmath83 and the polyhedron @xmath102 satisfy the poincar polyhedron theorem , @xmath162 is discrete and @xmath102 is its fundamental polyhedron .
obviously , @xmath167 .
suppose that ( 2)(ii ) holds .
then the group @xmath168 generated by the side pairing transformations @xmath163 , @xmath164 , @xmath130 , @xmath166 , and @xmath159 and the polyhedron @xmath123 satisfy the poincar theorem , @xmath168 is discrete , and @xmath123 is its fundamental polyhedron . in the proof of lemma [ h1h2 ]
it was shown that , for @xmath105 odd , @xmath169 and @xmath170 .
moreover , @xmath171 .
hence , @xmath172 , so @xmath76 is discrete .
theorem [ criterion_psl ] is proved .
our next goal is to compute parameters @xmath173 for both series of discrete groups listed in theorem [ criterion_psl ] .
if @xmath52 is a loxodromic element with translation length @xmath174 and rotation angle @xmath175 , then @xmath176 and @xmath177 is called the _ complex translation length _ of @xmath50 .
note that if @xmath50 is hyperbolic then @xmath178 and @xmath179 .
if @xmath50 is elliptic then @xmath180 and @xmath181 .
if @xmath50 is parabolic then @xmath182 ; by convention we set @xmath183 .
we define the set @xmath184 in other words , the set @xmath185 consists of all complex translation half - lengths @xmath186 for hyperbolic , parabolic , and primitive elliptic elements @xmath50 .
furthermore , we define a function @xmath187 as follows : @xmath188 given @xmath189 and @xmath50 with @xmath190 , @xmath191 determines the type of @xmath50 and , moreover , its order if @xmath50 is elliptic . note also that since we regard @xmath192 and @xmath193 , an expression of the form @xmath194 with @xmath195 means , in particular , that @xmath191 is finite .
[ criterion_par ] let @xmath196 with @xmath197 , @xmath198 , and @xmath199 .
then @xmath1 is discrete if and only if one of the following holds : 1 .
@xmath200 and @xmath201 , where @xmath202 with @xmath203 , @xmath204 , and @xmath205 ; 2 .
@xmath200 and @xmath206 , where @xmath202 with @xmath207 , @xmath208 , and @xmath205 . obviously , @xmath197 and @xmath198 if and only if @xmath50 is @xmath0-loxodromic and @xmath77 is parabolic . with this choice of @xmath53 and @xmath209 , @xmath199
if and only if the group @xmath1 is a non - elementary @xmath14 group without invariant plane @xcite .
this means that the hypotheses of theorem [ criterion_par ] are equivalent to the hypotheses of theorem [ criterion_psl ] .
therefore , in order to prove theorem [ criterion_par ] it is sufficient to calculate the parameters @xmath53 and @xmath210 for both families of the discrete groups listed in theorem [ criterion_psl ] .
let @xmath211 be the image of @xmath89 under @xmath164 , that is @xmath212 .
using the identity ( [ fprim ] ) and the fact that @xmath133 , we have @xmath213=fgf^{-1}g^{-1}=(r_\omega r_\sigma)(r_\omega r_\tau)= ( r_{\sigma ' } r_\omega)(r_\omega r_\tau)=r_{\sigma'}r_\tau.\ ] ] note that @xmath211 and @xmath86 are disjoint and @xmath93 is orthogonal to both of them .
therefore , @xmath214 $ ] is a hyperbolic element with the axis lying in @xmath93 and the translation length @xmath215 , where @xmath216 is the distance between @xmath211 and @xmath86 .
hence , since @xmath199 , @xmath217 - 2=+2\cosh d-2.\ ] ] now , using generalised triangles in the plane @xmath93 , it is not difficult to calculate that @xmath218 where @xmath219 is the distance between @xmath89 and @xmath86 if they are disjoint .
hence , @xmath220 where @xmath189 , @xmath221 .
let us calculate @xmath53 .
the element @xmath50 is @xmath0-loxodromic if and only if @xmath222 , where @xmath223 is the translation length of @xmath50 .
that is , @xmath224 note that @xmath225 is the distance between @xmath81 and @xmath83 .
it is measured in @xmath80 and equals @xmath226 ( see figure [ poly_par ] ) .
suppose that we are in case ( 2)(i ) of theorem [ criterion_psl ] , that is @xmath204 , and that @xmath89 and @xmath86 intersect .
recall that @xmath121 is the bisector of the dihedral angle of @xmath102 made by @xmath89 and @xmath86 .
let @xmath227 be the angle that @xmath121 makes with @xmath78 .
note that @xmath228 . from the link of @xmath229
, we have that @xmath230 and , therefore , @xmath231 further , from the link of @xmath229 , @xmath232 from the @xmath233 , @xmath234 and , from the quadrilateral @xmath235 , @xmath236 finally , from @xmath237 , @xmath238 combining ( [ psi])([sinht ] ) , we have that @xmath239 similar calculations can be done for parallel or disjoint @xmath89 and @xmath86 .
hence , @xmath240 , where @xmath241 , @xmath205 . now note that in case ( 2)(ii ) of theorem [ criterion_psl ] , the angle @xmath228 must be of the form @xmath126 , @xmath75 is an integer , @xmath110 , or @xmath111 .
then we need to recompute the formulas ( [ adc])([sinht ] ) with @xmath242 : @xmath243 then @xmath244 hence , @xmath206 , where @xmath241 , @xmath205 .
denote by @xmath245 the discontinuity set of a kleinian group @xmath11 .
the _ kleinian orbifold _
@xmath246 is said to be an orientable @xmath247-orbifold with a complete hyperbolic structure on its interior @xmath248 and a conformal structure on its boundary @xmath249 . * @xmath250=\langle x , y , s\,|\ , x^\infty = s^2=(xs)^2=(ys)^2=(xyxy^{-1})^m=(y^{-1}xys)^q=1\rangle$ ] , * @xmath251=\langle w , x , y , z\,|\,w^\infty = x^2=y^2=z^2=(wx)^2=(wy)^2=(yz)^2= ( zx)^q=(zw)^m=1\rangle$ ] , * @xmath252= \langle x , l\,|\,x^\infty=(xlxl^{-1})^m=(xl^2x^{-1}l^{-2})^q=1\rangle$ ] , * @xmath253=\langle x , y , z\,|\ , x^\infty = y^2=z^\infty=(xy)^m=(yzy^{-1}z^{-1})^q=[x , z]=1\rangle$ ] . here
@xmath105 and @xmath60 are integers greater than 1 , or @xmath110 or @xmath111 with the following convention .
if we have a relation of the form @xmath254 with @xmath255 , then we simply remove the relation @xmath254 from the presentation ( in fact , this means that the element @xmath256 is hyperbolic ) .
further , if @xmath257 and we keep the relation @xmath258 , we get a kleinian group presentation where parabolics are indicated . to get an abstract group presentation , we need to remove all relations of the form @xmath259 . 1 . if @xmath204 and @xmath201 , where @xmath261 , @xmath205 , @xmath262 , then @xmath11 is isomorphic to @xmath263 $ ] .
2 . if @xmath204 and @xmath201 , where @xmath261 , @xmath264 , @xmath265 , then @xmath11 is isomorphic to @xmath266 $ ] .
3 . if @xmath208 and @xmath206 , where @xmath261 , @xmath205 , @xmath262 , then @xmath11 is isomorphic to @xmath267 $ ] .
4 . if @xmath208 and @xmath206 , where @xmath261 , @xmath264 , @xmath265 , then @xmath11 is isomorphic to @xmath268 $ ] . suppose @xmath204 , that is the dihedral angle of @xmath102 between @xmath89 and @xmath86 is @xmath103 with @xmath105 even , @xmath110 , or @xmath111 . consider a polyhedron @xmath269 bounded by @xmath89 , @xmath86 , @xmath270 , @xmath271 , @xmath78 , and @xmath93 . applying the poincar theorem to @xmath269 and the side pairing transformations @xmath77 , @xmath272 , @xmath81 , and @xmath83
, one can see that @xmath273 is isomorphic to @xmath101 and has the presentation @xmath274 if @xmath275 is odd , then @xmath276 and @xmath277 $ ] .
if @xmath275 is even , @xmath110 , or @xmath111 , then @xmath101 contains @xmath11 as a subgroup of index @xmath278 and has presentation @xmath279 $ ] .
in order to see this , one can apply the poincar theorem to a polyhedron @xmath280 bounded by @xmath86 , @xmath89 , @xmath281 , @xmath211 , @xmath78 , and @xmath282 , and side - pairing transformations @xmath50 , @xmath77 , and @xmath283 .
assume that we are in case ( a ) or ( b ) .
then each @xmath284 , @xmath285 , is a @xmath0-loxodromic element with translation length @xmath286 and @xmath287 , where @xmath288 is the distance between @xmath289 and @xmath290 .
moreover , from the matrix representation , @xmath291 .
the inequalities ( [ kbounds ] ) enable us to determine the signs of @xmath292 and @xmath293 : @xmath294 suppose that @xmath118 .
simple calculations in the plane @xmath93 show that @xmath295 and , on the other hand , @xmath296 so , we obtain @xmath297 applying lemmas [ gamma ] and [ fgk ] and the facts that @xmath298 and @xmath299 , we get @xmath300\cdot[k(\tr(fg)-\tr f)+\tr f]\\ & = & k(k-1)(\tr(fg)-\tr f)^2+(2k-1)\cdot\tr f\cdot ( \tr(fg)-\tr f)+\tr^2 f\\ & = & k(k-1)\gamma+(2k-1)\sqrt{-\beta-4}\sqrt{-\gamma}+\beta+4.\end{aligned}\ ] ] hence , @xmath301 , where @xmath302 , @xmath303 is an integer .
analogous calculation can be done for @xmath304 and @xmath305 , and we obtain item ( 1 ) of the theorem . further , since @xmath316 and , from the plane @xmath93 , @xmath317 , we have that @xmath318 thus , @xmath319 , where @xmath72 is an integer .
analogous calculations can be done for @xmath304 and @xmath305 and we obtain item ( 3 ) of the theorem . when @xmath46 , the parameters were described in theorem [ criterion_par ] .
here we just note that for @xmath46 and @xmath323 , the group @xmath23 is free if and only if @xmath43 lies in the region @xmath324 finally , we are able to draw those subsets of @xmath17 that correspond to discrete free groups .
these subsets are shown in figure [ free ] .
the dashed lines @xmath315 are plotted to show a certain symmetry of @xmath17 .
the other discrete groups contain elliptic elements .
their parameters are represented by lines , parabolas , hyperbolas , and points accumulating , as orders of elliptic elements tend to @xmath110 , to the regions of free groups . in figure
[ whole ] , the whole picture for the slice @xmath17 is shown to give an idea of the structure of @xmath17 .
the formulas for @xmath28 and @xmath29 obtained in theorems [ criterion_par ] , [ ell_rp ] , [ hyp - par_rp ] , [ hyp_fuchsian ] , and [ lox_inv ] , were programmed with the package maple 7.0 for some ( sufficiently large ) values of independent variables like @xmath326 and @xmath202 and plotted on the plane @xmath43 .
the most interesting families of parameters appear when @xmath29 and @xmath28 are of the same sign .
for a fixed @xmath327 , the hyperbolas @xmath328 where @xmath303 is an integer , form a one - parameter family of curves converging to the boundary of @xmath320 as @xmath329 .
each hyperbola has the asymptotes @xmath330 and @xmath331 , which are obviously parallel to @xmath332 and @xmath315 , respectively . for @xmath46 and @xmath333 ,
consider a one - parameter family of parabolas @xmath334 .
let @xmath335 be the domain bounded by @xmath336 : @xmath337 within each @xmath335 , the parameters for discrete groups are given by @xmath338 where @xmath339 , @xmath340 , and @xmath189 .
note that for @xmath341 , we have @xmath315 and @xmath342 . as @xmath33 , the curves @xmath343 accumulate to the boundary of @xmath335 , i.e. , to the boundaries of @xmath344 and @xmath345 ( see figure [ sigma2 ] for an example of @xmath335 for @xmath346 )
. 99 a. f. beardon , _ the geometry of discrete groups _ , springer - verlag , new york heidelberg berlin , 1983 .
a. f. beardon , _ fuchsian groups and @xmath32th roots of parabolic generators _ , holomorphic functions and moduli , vol .
ii ( berkeley , ca , 1986 ) , 1322 , math .
publ . , 11 , 1988 .
d. b. a. epstein and c. petronio , _ an exposition of poincar s polyhedron theorem _
, lenseignement mathmatique * 40 * ( 1994 ) , 113170 . w. fenchel , _ elementary geometry in hyperbolic space _ , de gruyter studies in mathematics , * 11*. walter
de gruyter & co. , berlin , 1989 . f. w. gehring , j. p. gilman , and g. j. martin , _ kleinian groups with real parameters _ ,
contemp . math . * 3 * , no . 2 ( 2001 ) , 163186 . f. w. gehring and g. j. martin , _ stability and extremality in jrgensen s inequality _ , complex variables theory appl . * 12 * ( 1989 ) , no . 1 - 4 , 277282 . f. w. gehring and g. j. martin , _ chebyshev polynomials and discrete groups _ , proc .
of the conf . on complex analysis ( tianjin , 1992 ) , 114125 , conf .
lecture notes anal .
, i , internat . press , cambridge , ma , 1994 .
j. gilman and b. maskit , _ an algorithm for @xmath278-generator fuchsian groups _ , mich .
j. * 38 * ( 1991 ) , no . 1 , 1332 .
e. klimenko and n. kopteva , _ discreteness criteria for @xmath12 groups _ , israel j. math .
* 128 * ( 2002 ) , 247265 .
e. klimenko and n. kopteva , _ all discrete @xmath12 groups whose generators have real traces _ , int
. j. algebra comput . *
15 * ( 2005 ) , no . 3 , 577618 .
e. klimenko and n. kopteva , _
discrete @xmath12 groups with a parabolic generator _ , sib .
* 46 * ( 2005 ) , no . 6 , 10691076 .
e. klimenko and n. kopteva , _ two - generator kleinian orbifolds _ , 2005 , preprint .
e. klimenko and m. sakuma , _ two - generator discrete subgroups of @xmath347 containing orientation - reversing elements _ , geometriae dedicata * 72 * ( 1998 ) , 247282 .
a. w. knapp , _ doubly generated fuchsian groups _
j. * 15 * ( 1968 ) , no . 3 , 289304 . j. p. matelski , _ the classification of discrete 2-generator subgroups of @xmath348 _ , israel j. math .
* 42 * ( 1982 ) , no . 4 , 309317 .
e. b. vinberg , _ hyperbolic reflection groups _ , russian math .
surveys * 40 * ( 1985 ) , 3175 . | we describe all real points of the parameter space of two - generator kleinian groups with a parabolic generator , that is , we describe a certain two - dimensional slice through this space . in order to do this
we gather together known discreteness criteria for two - generator groups and present them in the form of conditions on parameters .
we complete the description by giving discreteness criteria for groups generated by a parabolic and a @xmath0-loxodromic elements whose commutator has real trace and present all orbifolds uniformized by such groups . |
defining complete samples of agn , both locally and in the distant cosmological fields , has been a major ongoing goal of agn research .
hard x - ray selection is generally considered the best way to detect both relatively uncontaminated and complete samples of agn . at high column densities
, however , the dust and gas surrounding the central engine ( in combination with that located in the host galaxy ) are capable of hiding virtually all accessible agn tracers . therefore , while the overall resolved fraction of the cosmic x - ray background ( cxrb ) is high , it drops with increasing energy to 60% at 6 - 8 kev and to 50% at @xmath7 8 kev @xcite . population synthesis models of the cxrb therefore predict a significant population of heavily obscured agn not detected in the deepest x - ray fields . for instance , @xcite predict that current x - ray catalogs are 25% incomplete at @xmath8 @xmath3 and 70% incomplete at @xmath9 @xmath3 .
numerous attempts have been made to detect this population of heavily obscured agn , many of which have focused on the mid - infrared ( mir ) emission where the obscured radiation is re - emitted or on combinations of mir and multi - wavelength data .
we focus here on two selection techniques independent of the optical and x - ray properties of the agn : radio - excess and mir power - law selection .
the former selects agn
whose radio emission far exceeds that of the radio - infrared correlation for star - forming galaxies , whereas the latter selects agn whose irac seds exhibit the characteristic power - law emission expected of luminous agn ( e.g. * ? ? ?
* ) . by focusing on the infrared and radio , wavebands where dust obscuration is minimal
, we can select both unobscured agn as well as heavily obscured agn likely to be missed in the uv , optical , and soft x - ray bands . to test the x - ray properties of the _ spitzer_-selected agn samples , we focus on the cdf - n , the deepest x - ray field observed to date .
_ spitzer _ mips and irac images with exposures of 1400 s and 500 s , respectively , cover the full area of the _ chandra _ 2 ms field , as does the deep 1.4 ghz radio data of @xcite .
optical and near - infrared imaging and photometry are available from the goods dataset @xcite as well as from the data of @xcite ( _ ubvrizhk _ ) .
we assume a cosmology of ( @xmath10,@xmath11)=(0.3 , 0.7 , 72 km s@xmath12 mpc@xmath12 ) .
we use the well - known radio - infrared relation of star - forming galaxies and radio - quiet agn , to select a sample of radio - excess agn . @xcite
define @xmath13 = log ( f@xmath14/f@xmath15 ) . for star - forming galaxies ,
@xmath13 is tightly constrained out to z @xmath16
1@xmath17 @xmath13 = 0.94 @xmath18 0.23 after k - correction @xcite . galaxies with values of @xmath13
well below this range ( strong radio with respect to 24 emission ) are unlikely to be dominated by star formation and instead are radio - emitting agn ( e.g. * ? ? ?
we therefore set a selection threshold of @xmath19 to classify a galaxy as probably having an agn and require a _
chandra _ exposure of @xmath7 1 ms to provide high x - ray sensitivity .
this selection identifies 27 radio - excess agn in the cdfn @xcite .
the radio - excess agn lie at the typical redshift of x - ray sources in the cdf - n , @xmath0 , and ( when detected in the x - ray ) tend to have x - ray luminosities typical of seyfert galaxies and seds dominated by the 1.6 stellar bump .
approximately 60% of these radio - excess agn , however , are x - ray undetected in the 2 ms _ chandra _ catalog , even at exposures of @xmath20 1 ms ; 25% lack even 2@xmath21 x - ray detections .
the absorbing columns to the faint x - ray - detected objects are @xmath22 @xmath23 n@xmath24 @xmath25 @xmath3 , i.e. , they are obscured but unlikely to be compton thick .
of the 9 agn that are either x - ray non - detected or are too close to a known x - ray source to search for weak x - ray emission , we estimate that 6 could be compton thick .
this corresponds to 22% of our sample and is consistent with predictions from the x - ray background .
in the mir , luminous agn can often be distinguished by their characteristic power - law emission , which extends from the infrared to the ultraviolet ( e.g. * ? ? ?
* ; * ? ? ?
this emission is not necessarily due to a single source , but can arise from the combination of non - thermal nuclear emission and thermal emission from various nuclear dust components ( e.g. * ? ? ?
@xcite selected a sample of 92 such sources in the cdf - s , 70% of which are hyper - luminous infrared galaxies ( hyperlirgs , log l@xmath26(l@xmath27)@xmath28 ) or ultra - luminous infrared galaxies ( ulirgs , log l@xmath26(l@xmath27)@xmath29 ) . nearly half ( 47% ) of their power - law sample were undetected in x - rays at exposures of up to 1 ms .
we use a selection similar to that of @xcite to identify a sample of 62 high @xmath30 , red ( @xmath31 , where @xmath32 ) irac power - law galaxies in the 2 ms cdf - n @xcite . of the 62 power - law galaxies , only 34 ( 55% ) have x - ray counterparts in the 2 ms cdf - n x - ray catalog @xcite , consistent with the findings of @xcite , although 85% show evidence for x - ray emission at the @xmath33 detection level .
while the power - law galaxies comprise only @xmath34% of the x - ray and mir - detected agn in the cdf - n , they make up a significant fraction of the high redshift ( @xmath1 ) and high luminosity x - ray agn sample , as is demonstrated in fig .
this is not surprising , as the power - law selection requires the agn to be energetically dominant , and therefore preferentially selects the most luminous agn .
to illustrate the effect of x - ray luminosity on the agn contribution to the optical through mir continuum , we also plot in fig .
1 the median rest - frame optical - mir seds of the x - ray detected members of the comparison sample , as a function of x - ray luminosity .
the comparison sample consists of the 1420 irac sources in the cdf - n that meet our @xmath30 and x - ray exposure cuts .
low - luminosity x - ray sources are dominated by the stellar bump in the optical - nir bands ( e.g. * ? ? ?
the strength of this feature decreases with increasing x - ray luminosity , and disappears almost entirely at luminosities of log @xmath35(ergs s@xmath12)@xmath36 , where the sed takes on the characteristic power - law shape . the obscured fraction ( @xmath2 @xmath3 ) of the x - ray detected power - law galaxies is @xmath16 68% , in agreement with that of @xcite for agn at similar redshifts .
if we consider only those power - law galaxies in the x - ray catalog , we therefore derive an obscured to unobscured ratio of @xmath16 2:1 . adding the 10 weakly - detected power - law galaxies , all of which are likely to be obscured , we calculate an obscured fraction of @xmath16 75% . if we further include the power - law galaxies not detected in x - rays , assuming that all are obscured , the maximum obscured ratio of power - law galaxies rises to @xmath37 4:1 - 5:1 ( 82% )
power - law galaxies detected to high @xmath30 in the irac bands account for at most @xmath38% of both the mir - detected agn and the mir - detected obscured agn predicted by the x - ray luminosity function synthesis models of @xcite down to 24 flux densities of @xmath39 @xmath40jy .
the majority of obscured agn should therefore have seds dominated by the host galaxy , or red power - law seds that fall below the irac detection limit .
we plot in fig .
2 the position of the power - law galaxies with respect to the irac agn color cuts of @xcite and @xcite , designed for relatively shallow surveys .
all of the power - law galaxies lie within both the lacy et al . and stern et al . color regions , and as such
, they comprise a subset of the color - selected mir sources . while the color - selected agn samples in the cdf - n comprise a higher fraction of high - luminosity agn than the power - law selected sample ( see fig .
1 ) , they also appear to suffer from a higher degree of contamination from star - forming galaxies ( see table 1 ) .
this is not surprising , given the behavior of the star - forming and agn templates shown in fig .
2 ( available in * ? ? ?
the star - forming templates enter the agn selection regions at both low and high redshifts , whereas they fall outside of the more stringent power - law region at @xmath41 .
lccc selection criteria & power - law & stern et al .
& lacy et al .
+ number & 62 & 216 & 393 + detected in x - rays & 55% & 38% & 25% + detected at 24 & @xmath792% & 83% & 72% + l@xmath42 w hz@xmath12 & 88% & 54% & 49% + quiescent optical / x - ray colors & @xmath43 & @xmath44 & @xmath45 +
we present the results of a _ spitzer _ search for agn in the _ chandra _ deep field - north .
we focus on two selection techniques independent of the optical and x - ray properties of the sources .
the first is radio - excess selection , in which we select sources whose radio emission far exceeds that of the radio - infrared correlation . in the cdf - n ,
we detect 27 radio - excess agn .
these sources tend to lie at @xmath0 and have seds dominated by the 1.6 stellar bump .
only 40% are detected in the x - ray at exposures of @xmath46 ms , although 75% show signs of x - ray emission at the @xmath47 detection level .
those with x - ray counterparts have seyfert - like x - ray luminosities .
the majority of the radio - excess agn are consistent with being obscured ( @xmath2 @xmath3 ) , and 20% are potentially compton - thick .
the second selection technique is mir power - law selection , in which we select sources with red ( @xmath48 ) irac power - law seds .
the sources selected in this way tend have higher redshifts ( @xmath1 ) and x - ray luminosities than the radio - excess sample .
the irac power - law selection recovers a significant fraction of the most x - ray luminous agn in the cdf - n . despite this , only 55% of the power - law galaxies are detected in the x - ray , with 15% remaining undetected down to the @xmath49 detection level .
these x - ray detection fractions are similar to those of the radio - excess agn , even though there is almost no overlap in the two samples .
this suggests that a relatively low x - ray detection rate may be a common feature for agn samples selected independently of their optical and x - ray properties , regardless of luminosity or redshift .
a study of the intrinsic obscuration of the power - law galaxies suggests that as many as 82% ( 4:1 - 5:1 ) are obscured .
the number densities of these sources indicate that the power - law sample comprises only 20 - 30% of the mips - detected obscured agn population .
the remaining agn are likely to have seds dominated by the host galaxy , like those of the radio - excess sample .
the power - law galaxies discussed here comprise a subset of agn selected via the irac color - selection criteria of @xcite and @xcite . while larger in number than the power - law sample , the color - selected samples in the cdf - n appear to suffer from a higher degree of contamination from star - forming galaxies . by applying additional sed constraints to a color - selected sample , or by requiring either a 24 detection or agn - like optical / x - ray colors
, one can create both a more complete and reliable sample of mir - selected agn , even in the deepest astronomical fields .
doing so is crucial if we are to constrain the fraction of heavily obscured agn in the distant universe and determine their contribution to the accretion history of the universe . | we present the results of a _ spitzer _ search for obscured agn in the _ chandra _ deep field - north , using both radio - excess and mid - infrared power - law selection .
agn selected via the former technique tend to lie at @xmath0 , have seds dominated by the 1.6 stellar bump , and have seyfert - like x - ray luminosities ( when detected in the x - ray ) . in contrast , the irac ( 3.6 - 8.0 ) power - law selected agn lie at higher redshifts of @xmath1 , and comprise a significant fraction of the most x - ray luminous agn in the cdf - n . while there is almost no overlap in the agn samples selected via these two methods , their x - ray detection fractions are very similar .
only 40% and 55% of the radio - excess and power - law samples are detected in the 2 ms x - ray catalog , respectively . the majority of the agn selected via both methods are consistent with being obscured ( @xmath2 @xmath3 ) , but not compton - thick ( @xmath4 @xmath3 ) , although compton - thick candidates exist in both samples .
we place an upper limit of @xmath5% ( or @xmath6 ) on the obscured fraction of the power - law sample , consistent with predictions from the cosmic x - ray background .
the sources selected via the power - law criteria comprise a subset of agn selected via other irac color - color cuts . while smaller in number than the color - selected samples in the deep fields , the power - law sample suffers from less contamination by star - forming galaxies . |
very dense gas ( n @xmath4 ) has an important effect upon star formation in molecular clouds .
the presence of very dense gas affects the jeans mass and other measures of stability .
in addition , the quantity of very dense gas has consequences for the calculated star formation efficiency , since it is this material that actively participates in star formation (; ) .
most stars ( even low - mass stars ) form in regions where high - mass stars are forming ( elmegreen 1985 ) . in high - mass star - forming regions , winds and radiation from nearby , newly - formed stars
can disrupt the local gas and effectively shut down further star formation .
cores of very dense gas , however , resist these disruptive forces and can help to maintain star formation in the hostile environments associated with young , massive stars (; ) .
how universal is very dense gas ( n @xmath4 ) ? and have shown that densities of order 10@xmath5 to @xmath6 are common in low - mass star - forming regions .
studies of a few selected regions that are forming massive stars ( e.g. , ; ; ; ; ; ; ; ; ; hofner et al .
1996 ) have demonstrated the presence very dense gas ( n @xmath7 to @xmath8 ) .
however , it is unclear whether such gas is common to all regions forming massive stars .
the overall sample of such regions is small , and the studies used a variety of selection criteria and density measurement techniques . to assess the prevalence of very dense gas , we need to determine densities using a consistent method in a large and representative sample of regions forming massive stars .
the small beam sizes at frequencies used to probe for very dense gas , along with limited amounts of available telescope time , make it impossible to map completely all the regions known to be forming massive stars .
therefore , we require a pointer to likely locations of active star formation within molecular clouds .
masers are ideal for this purpose since they contain at least small amounts of extremely dense gas ( n @xmath10 ) (; ) ; and , in well - studied regions , they are intimately associated with star formation ( e.g. , ; ; ; ) . our initial survey (; hereafter paper i ) searched for thermal emission from dense gas associated with h@xmath9o masers .
the sample consisted of 179 of the 181 h@xmath9o masers listed in the catalog of as `` hii region '' ( i.e. , not a late - type star ) masers which were north of @xmath11and had positions known to better than 8 .
we used the j = 7@xmath126 transition of cs , which has a high critical density ( n@xmath13 ) , selecting regions of very dense gas .
observations of a single transition of the cs molecule are not sufficient to determine gas density .
therefore , we have observed the j = 5@xmath124 , 3@xmath122 , and 2@xmath121 transitions of cs in 150 of the 179 regions from our initial cs j = 7@xmath126 survey ( paper i ) . of the regions sampled in the current survey , 85 ( 57% )
contained cs 7@xmath126 emission ; this percentage is similar to that of the original cs 76 survey ( 104 of 179 , or 58% ; paper i ) , indicating that this study was not biased towards the densest regions .
we also observed the same transitions of c@xmath0s in 49 of the strongest cs sources .
the c@xmath0s data yield an independent measurement of the densities .
we have also mapped 21 sources in the j = 54 line of cs to determine the size of the cores . in
3 , we present basic detection statistics and discuss the individual spectra and the maps . in 4 , we present densities and column densities based upon excitation analysis of the data , consider effects of opacity and temperature uncertainties on the results , and compute masses . in 5 , we consider issues like the star formation efficiency , compare these regions to other regions , and estimate the luminosity of the galaxy in the cs j = 5@xmath124 line .
we observed the j = 54 , 32 and 21 transitions of cs and c@xmath0s in 1990 june , 1991 april , and 1991 october at the iram 30-m telescope at pico veleta , spain , with the 3-mm , 2-mm , and 1.3-mm sis receivers tuned to single - sideband mode .
table 1 lists the line frequencies , main beam efficiencies ( @xmath14 ) , beamsizes , typical system temperatures , and velocity resolutions for each transition . to convert to the t@xmath15 scale , the data were scaled by @xmath14 ( i.e. , t@xmath15 = t@xmath16/@xmath14 ) . in iram notation , @xmath17 , the back spillover and scattering efficiency divided by the forward spillover and scattering efficiency . for all excitation calculations
we have assumed that the source is fully resolved , so that t@xmath15 = t@xmath18 , the rayleigh - jeans temperature of a spatially resolved source observed with a perfect telescope above the atmosphere .
the pointing was checked regularly . in 1990 june , we pointed on continuum emission from k3 - 50 , w3(oh ) , and jupiter and found the pointing to be very sensitive to changes in azimuth and elevation , with a maximum spread of @xmath19 15 .
consequently , we made nine - point maps on a 16 grid for each source , while defining a pointing curve .
the reliability of these observations will be discussed below .
the pointing curves were well - determined by the time we made the c@xmath0s observations . in 1991 april
, we checked the pointing with continuum observations of w3(oh ) , ngc 7027 , bl lac , and saturn . for this observing run ,
the telescope pointing was accurate to within 4 rms . in 1991 october , the pointing was good to 5 rms .
the data were calibrated to the t@xmath20 scale using the chopper wheel method ( ) . to include the effects of different sideband gains , we observed the line calibration sources w51 m , w51n , w44 , w3(oh ) , and dr21s in 1990 june , and irc+10216 , w3(oh ) , and orion irc2 in 1991 april .
these sources were observed by in single - sideband mode with an image - sideband rejection of @xmath21 8 db .
we compared our observed antenna temperatures with those tabulated by and adjusted our temperature scale to agree with theirs .
for the 1990 june run , we did not need to adjust the cs j = 21 data .
the june cs 32 and 54 transitions needed to be reduced in strength by @xmath19 10% .
the 1991 april cs j = 21 and 54 data both needed to be increased by 20% .
the 1991 april cs j = 32 data were multiplied by a factor of 2 and assigned a 50% calibration error . lacking a standard source for the c@xmath0s j = 32 line , we also scaled it up by a factor of 2 and assigned a 50% calibration error . in 1991 october ,
no scaling was necessary .
we have also compared the cs j=76 results of to the more recent single - sideband observations of .
for the five sources in common , the ratio of t@xmath16 values is [email protected] , indicating that the calibration of the j = 76 data in paper i was good .
we also used the cso for some auxiliary observations , with parameters shown in the last 4 lines of table 1 .
we made small maps of the cs j = 54 transition toward 21 of the sources in our sample in 1994 june .
the pointing was checked on planets and repeated to 3 .
we observed the j = 76 transition of c@xmath0s in 1991 june .
beam sizes and efficiencies are based on mangum ( 1993 ) .
for a few sources , we also observed the j = 109 transition of cs in 1993 december .
the pointing was checked on planets and found to be constant to 6 .
the beamsize was assumed to be 14 , based on scaling from lower frequencies , and the efficiency was measured on saturn .
finally , the j = 1413 line of cs was observed toward two sources in 1994 december .
the pointing was checked by observing planets and was constant to 6 .
the beamsize was assumed to be 11 .
the optical depth at the zenith was measured by tipping to be 0.35 at 685 ghz .
for all the cso observations , two aos spectrometers were used , with resolutions of about 140 khz and 1 mhz .
the choice of spectrometer for determining line parameters was made on the basis of getting sufficient resolution and signal - to - noise in the line .
table 2 lists the 150 h@xmath9o maser sites observed at iram in the cs j = 54 , 32 , and 21 transitions .
tables 1 and 2 of paper i list the positions of the masers and the cs j=76 line parameters or upper limits . table 2 of the present paper lists the source names in order of increasing galactic longitude , the radiation temperature ( t@xmath15 ) , integrated intensity ( @xmath23 ) , velocity centroid ( v@xmath24 ) , and full width at half maximum ( fwhm ) for the three transitions . for cs data obtained in 1990 june
, we list the line parameters at the position in the nine - point map with the strongest emission in the j=54 line .
this choice is based on the results of 3.2 , where we find that the j=54 emission almost always peaks at the maser position .
while the line parameters for 1990 june are useful in detection statistics and as a guide for follow - up work , we have found that the position correction was inadequate for them to be used together with the j=76 data to determine densities ; therefore we do not use the 1990 june data in 4 . for undetected lines ,
the upper limits to t@xmath15 are 3@xmath25 . for cs j = 32 and 21 ,
we have tabulated only the data with the highest spectral resolution .
we also observed the c@xmath0s lines in 49 of the strongest cs emitters .
the results for c@xmath0s are presented in table 3 .
transitions listed with dashes ( ) instead of values or upper limits to t@xmath15 were not observed .
table 4 has the results for j=109 and 1413 .
usually , we obtained the line parameters from gaussian fits to the lines but some sources listed in table 2 had spectra with more than one peak . to determine the line parameters in these cases
, we took the following approach .
first , if the profiles of the higher j ( i.e. , 76 or 54 ) lines or c@xmath0s lines ( where available ) matched one or more of the peaks seen in the lower j transitions , we assumed that the source was composed of distinct cloud components ( e.g. , figure 1a ) ; and we derived the line parameters by performing a multiple gaussian fit to the whole profile . each gaussian component is listed individually in table 2 .
three sources have 2 velocity components and one has 3 components ; these are identified in tables 2 and 3 by the notation `` c # '' ( where # is the component number ) . with the inclusion of all the separate components , table 2 displays results for 155 cloud components .
second , if comparison of cs data with c@xmath0s data indicated that the cs line was self - absorbed ( figure 1b shows an example of this situation ) , we calculated the line parameters ( @xmath26 , v@xmath24 , and fwhm ) from moment integrals over the profile .
then @xmath27 was calculated from @xmath26 /fwhm ( values given in parentheses in table 2 ) .
only 18 of the 150 spectra were obviously self - absorbed in cs 21 , with smaller numbers showing obvious self - absorption in the higher - j lines .
of course , self - absorption may exist at a less obvious level in other sources .
figure 2 illustrates the detection rate for the observed cs transitions . the distribution counts as detected only those sources with observed t@xmath28k . because the sensitivity achieved for the cs j = 76 line ( paper i ) was similar to that for the lower j transitions , the drop in the detection rate towards higher rotational levels reflects a real drop in the number of sources exhibiting emission at the same level in the higher j lines . to determine the effect that very dense gas has upon star formation
, we need to know the extent of the gas and its location within the star - forming regions .
we have observed 21 of our sources in the cs 54 line with the cso . for each source , we made a cross - scan in r.a . and dec .
, typically consisting of 9 points .
for most of the sources , the separation of the observed points was 30 . for a few of the smaller sources
, we made the observations at 15 intervals .
in addition , we have assembled from the literature data taken with the same equipment for four other sources from our survey .
table 5 lists the mapping results for all 25 sources .
the integrated intensities listed in table 5 are for the interpolated maximum along each cross scan . from the maps we derived diameters and beam correction factors , @xmath29)/@xmath30 .
the beam correction factors were calculated assuming that a gaussian was a good representation of both the beam shape and the source intensity distribution .
using the integrated intensity , the @xmath31 , and the distances , @xmath32(kpc ) , we calculated the luminosity in the cs j=54 line from @xmath33 table 5 also lists the offsets of the cs 54 peaks from the maser positions in arcseconds . with the exception of a few of the larger sources , differences in the peak position of the cs 54 distribution and the h@xmath9o maser position are smaller than the combined pointing uncertainties and maser positional uncertainties ( @xmath223 and @xmath348 , respectively ) .
jenness et al .
( 1995 ) have also found a very good correlation between the peak of the submillimeter emission and the maser position .
the mean diameter of the sources listed in table 3 is 1.0 pc .
the dispersion about this mean , however , is large ( 0.7 pc ) .
if one examines sources at @xmath35 pc , the mean diameter is 0.5 pc with a dispersion of 0.4 pc .
this difference , while significant , probably does not arise from observational biases in the cs data .
most of the more distant sources are well resolved and bright .
it is more likely that the differences arise from selection biases in the original samples used to search for h@xmath9o masers .
complete mapping of the cs 54 line in several sources gives similar sizes .
the emission in ngc2024 has a diameter of 0.4 pc , while s140 has a diameter of 0.8 pc ( snell et al .
the emission in m17 is more extensive : 2.3 pc in 54 ( snell et al . ) ; 2.1 pc in 76 ( wang et al . 1993 ) .
with the addition of the lower j transitions in the present study to the cs j = 76 data from paper i , we can determine densities in a large sample of star - forming regions . in 4.1 , we discuss the calculations and examine the effects of opacity and uncertainties in kinetic temperature on density and column density determinations . in
4.2 , we consider the effects of density inhomogeneities , and we compute masses in 4.3 . to determine densities and column densities , we used a large velocity gradient ( lvg ) code to solve the coupled equations of statistical equilibrium and radiative transfer , including the first 20 rotational levels of cs in the calculation .
we assume that the gas has a constant density and temperature and that it uniformly fills all the beams used in this study .
we calculated a 20@xmath3620 grid of radiation temperatures in column density per velocity interval
density space for a kinetic temperature of 50 k. the cs densities in the lvg model grid ran from @xmath37 to @xmath38 , and the column densities per velocity interval ( n/@xmath39v ) ranged from @xmath40 to @xmath41 @xmath42/km s@xmath43 .
these ranges span the parameter space of all solutions which fit our data .
all the models converged to a solution . using a @xmath44 minimization routine
, we fit the lvg models to the observed cs line intensities .
table 6 lists the densities for 71 sources .
we have not included fits for the cs data obtained in 1990 june , for reasons discussed below .
we have listed the log of the density and column density , along with the value of @xmath44 and a key to which transitions were used and whether the lines were self - absorbed .
the values of density and column density apply to the region selected by the typical beam used for the observations ( about 20 ) .
the @xmath44 values allow us to assess whether the models ( any of the points in the lvg grid ) are a good representation of the data .
the distribution of @xmath44 values for sources with 4 transitions ( 40 sources ) is similar to what is expected theoretically if the model is a reasonable fit to the data , as is the distribution for sources with only three transitions ( 31 sources ) .
these facts suggest that our estimates of the calibration uncertainties are reasonable .
we originally included the 1990 june data in the fits , but they had a very high percentage of bad fits , leading us to conclude that the uncertain pointing made them unsuitable for combining with the cso j=76 data .
the 8 self - absorbed sources with fits in table 6 ( marked by a flag ) do not have significantly worse than the other sources .
one source with 3 transitions ( 212.25 - 1.10 ) produced a very uncertain density , and we have excluded it from the statistics that follow . the mean logarithmic density for sources with detected emission from all 4 cs transitions is @xmath45 , where 0.23 represents the standard deviation of the distribution .
the mean logarithmic column density is @xmath46 .
the results for the sources undetected in j=76 are @xmath47 ; @xmath48 .
figure 3 shows histograms of the densities and column densities .
the solid line plots the densities determined from all 4 cs transitions and the dashed line is the density distribution for sources without j= 76 detections .
these results show that the difference between a cs 76 detection and non - detection is more related to column density than to volume density .
therefore , the detectability of lines of high critical density is more affected by the quantity of dense gas present than by its density . to check whether the difference was solely a result of
having a j=76 line to fit , we re - fit the sources with 76 detections , forcing the @xmath49 fitting routine to ignore the cs 76 line and to fit only the 3 lower transitions .
the resulting is @xmath50 , and is @xmath51 .
this result confirms our conclusion that the most significant difference between a j=76 detection and a non - detection is the column density . what effect would high opacity in the cs lines have on the derived densities and column densities ?
eighteen of the sources in this survey have noticeable self - absorption in at least one transition .
in addition , an lvg model run for the mean density , column density , and linewidth results in cs line opacities that are roughly unity .
thus , self - absorption may affect the fits , even if it is not apparent in the line profiles .
since the c@xmath0s transitions will usually be optically thin , we independently fit the c@xmath0s transitions to an lvg model grid , with a range of parameters identical to those used in the original cs grid .
table 6 lists the densities , column densities , and @xmath52 derived from fits to the c@xmath0s data .
problems with the receivers during the c@xmath0s observations meant that we have various combinations of lines to fit , as indicated by the key in table 6 .
there are few sources with both adequate cs and acceptable c@xmath0s data .
the fits to the sources with three transitions of c@xmath0s give @xmath53 , essentially identical to the derived from 4 transitions of cs . the mean difference between cs and c@xmath0s in log(n )
is @xmath54 , indicating no significant difference in the derived densities .
it is unlikely that the densities calculated for sources in our survey from the cs lines alone are seriously affected by cs optical depth .
the average isotope ratio , @xmath55 , is @xmath56 , clearly less than the terrestrial ratio , and lower than the isotope ratios of 917 found by mundy et al .
( 1986 ) and 13 ( wang et al . 1993 ) .
chin et al .
( 1996 ) have recently found evidence for low values of this ratio in the inner galaxy , but our values are lower still .
it is likely that our procedure has underestimated @xmath57 to some extent .
for this reason , and also because these ratios are not very well determined for individual sources , we have adopted an isotopic abundance ratio of 10 in what follows . by increasing the number of transitions , simultaneous fitting of the cs and c@xmath0s data
should , in principle , allow us to determine the densities and column densities more accurately . using the lvg model grid for cs and constraining the isotope ratio to be 10 , we fit cs and c@xmath0s transitions simultaneously .
the results are listed in table 6 . while neither the densities nor the column densities are significantly different from those determined from fits to the cs data alone , @xmath49 is considerably larger .
the poor fits probably result from assuming a fixed isotopic abundance ratio for all sources .
it is likely that many of the regions of massive star formation contained within this study have temperatures in excess of 50 k. at the densities implied by the cs observations , the gas kinetic temperature will be coupled to the dust temperature . for grains with opacity decreasing linearly with wavelength , one can write @xmath58^{0.2},\ ] ] where @xmath59 is the luminosity in solar units , @xmath32 is the distance in kpc , and @xmath60 is the angular separation from the heating source in arcseconds .
using these units , @xmath61 to 40 ( makinen et al . 1985 , butner et al . 1990 ) .
we can estimate the range of temperatures in our sources from the luminosities in table 7 and distances in table 5 ; @xmath62 . at a radius of 10@xmath63 , characteristic of the beams in table 1 and the beam of the j = 76 observations , @xmath64 to 100 k. to assess the effects of temperature uncertainties on the derived source properties
, we also fit the sources with 4 transitions to a grid of models run for a temperature of 100 k. the value of decreased by 0.3 and the value of was essentially unchanged . regardless of the assumed temperature , our data imply a thermal pressure , @xmath65 k @xmath66 , which is much higher than found in regions not forming massive stars . within the limitations of a single - density model , we conclude that the effects of opacity and temperature on the determinations of density are not severe ( about at the factor of 2 level ) .
typical densities in regions detected in the j=76 survey are @xmath8 . toward water masers
not detected in the j=76 survey , the densities are about a factor of 2 less , but the column densities of cs are about a factor of 7 less , on average , than the values found for regions detected in the j=76 line . the densities for both groups of sources are considerably less than the critical density of the cs j=76 line ( @xmath67 ) , reminding us that detection of emission from a hard - to - excite line does not imply the existence of gas at the critical density .
molecules can emit significantly in high - j transitions with critical densities considerably above the actual density because of trapping and multilevel effects ( see also evans 1989 ) .
for example , levels with j@xmath68 have many possible routes for excitation by collisions , but only one radiative decay path .
the high densities found in this survey of regions forming massive stars are similar to those obtained from other , more detailed , studies of individual , luminous , star - forming regions ( see ref . in 1 ) .
consequently , the results found from studies of a few clouds can be applied , in a statistical sense , to the broader sample of massive star - forming regions .
our lvg analysis assumes that the density is uniform and that the emitting gas fills the beam .
how good are these assumptions ?
figure 4 gives examples of lvg model fits to several of the sources : three with good fits and three with bad fits , as measured by the value .
while the lvg models generally fit the data within the uncertainties , a closer look reveals that the discrepancies between model and observation are very consistent , even for the good fits .
almost all fits overpredict the 32 and 54 lines and underpredict the 21 and 76 lines .
thus , the data have , on average , a smaller variation of intensity with j than do the best - fit lvg models , as would be expected for a source with a mixture of gas at different densities . in this section
, we examine models with varying densities to see how well they explain the intensity versus j distribution . and wang et al . ( 1993 )
have discussed the effects of fitting a single density to the cs emission from a mixture of gas at about @xmath8 and gas lower in density by about a factor of 10 .
they showed that , until the filling factor of the high density gas becomes very small ( i.e. , @xmath69 ) , the density derived from fitting a single density model matches that of the high density component to within a factor of two .
the cs transitions we have observed should behave in a similar way in that they are biased toward measuring gas with densities close to @xmath8 . we now ask a more radical question .
could the apparent density near @xmath8 be an artifact of fitting to a single density a mixture of ultra - dense gas ( n = @xmath38 ) and gas at a much lower ( n = @xmath37 ) density ?
in this picture , the histogram of densities ( figure 3 ) would be produced by varying the filling factor of the dense component .
we chose a value of @xmath38 for the density of the ultra - dense gas because the 76 transition becomes completely thermalized at that density .
thus , the component with n@xmath70 represents any gas with n@xmath71 .
we synthesized clouds from a mixture of these two components at 20 values of n/@xmath39v between @xmath72 and 10@xmath73 @xmath42/km s@xmath43 .
for each density and column density , we used the lvg code to calculate the expected emission .
we then varied the filling factor of the ultra - dense gas ( @xmath74 ) and the low - density gas ( @xmath75 ) , with @xmath76 in steps of 0.05 , and summed the contributions to each transition for each possible combination of @xmath74 , column density of the gas at n@xmath77 ( n@xmath78 ) , and column density of the gas at n@xmath70 ( n@xmath79 ) .
these results then formed a grid of models which could be fitted to the data , just as the single - density models had been fitted .
we found that the value worsened , despite the extra free parameter , for sources where the single - density fit had been good ( @xmath80 ) . on the other hand ,
the sources which were poorly fitted ( @xmath81 ) with the single - density model were better fitted with the two - density model .
the two - density fits typically required very high column densities ( @xmath82 ) of the low - density gas compared to those of the ultra - dense gas ( @xmath83 ) . to see if we could constrain the amount of ultra - dense gas in the sources with poor single - density fits
, we followed a similar , but less restrictive , procedure .
we started by assuming that the cs j = 21 and 32 transitions effectively probe the low density gas in the beam , and we used them to fit the density ( n@xmath78 ) and column density ( n@xmath78 ) of the low - density component .
we then used the lvg code to obtain the expected emission from each rotational transition for a gas at this density and column density at a temperature of 50k .
these intensities , multiplied by ( @xmath75 ) , were used to represent the lower density component .
we then searched a parameter space of @xmath74 and log(n/@xmath39v ) for the best values for the ultra - dense component ( density once again fixed at 10@xmath84 @xmath66 ) .
we summed ( @xmath75 ) times the lower density intensities and @xmath74 times the ultra - dense gas intensities and compared this sum to the observations .
this method has a large number of free parameters : @xmath74 , n@xmath78 , n@xmath85v , and n@xmath86v , which are constrained by only 4 transitions .
furthermore , it does not correct the properties of the lower density component for the contributions of the high density gas to the j = 21 and 32 emission .
we use it for illustrative purposes only .
we show the two - density fits as dashed lines in figure 4 , but we do not tabulate the results .
the mean properties of these solutions for the sources with single - density @xmath81 are as follows : @xmath87 , log(n@xmath88 , log(n@xmath89 , and log(n@xmath90 ( equal column densities in the two components ) .
thus , in general , the filling factor of ultra - dense gas is small ( less than 25% ) , and the data still favor a large amount of gas at @xmath91 .
another possible source model is a continous density gradient , such as a power law .
power - law density distributions have been proposed for regions of low - mass star formation on theoretical grounds ( shu 1977 ) and seem to fit the observations well in some cases ( e.g. , zhou et al .
they have also been applied to some regions forming stars of higher mass ( e.g. , zhou et al . 1994 ; carr et al .
the latter reference is particularly relevant here , as it included a more complete analysis of gl2591 ( called crl2591 in this paper ) , including data from this paper , but adding other data . while table 6 indicates a good fit to the data for that source with a single - density model , carr et al
. found that a single density can not fit all the data , when other data are included , particularly j = 54 and 109 data from the cso .
they developed models with power - law density and temperature gradients that fit all the data .
we can use the example of crl2591 to explore the meaning of the densities in table 6 if the actual density distribution is a power law . if @xmath92 , with @xmath93 ( the density at 1 pc ) set by matching the line profiles ( carr et al . 1995 ) , the density in table 6 is reached at radii of 18 to 7 for @xmath94 , corresponding to filling factors of 0.3 to 0.6 in our largest beam .
we conclude that , in this source , the densities derived in this study characterize gas on scales somewhat smaller than our beams , if the source has a density gradient .
similar studies of other sources are needed to see if this conclusion can be generalized .
further evidence for a range of densities is that j=109 emission has been seen in a number of sources ( hauschildt et al .
1993 and our table 4 ) .
the data do not warrant detailed source - by - source modeling , but we have predicted the expected j=109 emission from a source with the mean properties found in 4.1 : log(n ) = 5.93 and log(n ) = 14.42 .
we assumed a linewidth of 5.0 , about the mean for our sample , and = 50 k. the predicted t@xmath18 of the j=109 line is 0.2 k for this average cloud , weaker than any of the detections .
if we use the conditions for the cloud with properties at the high end of the 1 @xmath25 spread , we can produce t@xmath18 = 1.6 k , about the weakest detection .
increasing to 100 k raises the prediction to 7 k , similar to many of the detections .
detection of a j=109 line therefore implies a cloud with higher density , column density , and/or temperature than the average cloud in our sample of sources detected at j=76 .
table 7 contains mass estimates for the regions for which we have determined cloud sizes .
we have computed three different estimates .
the first estimate assumes that the volume density fills a spherical volume with the diameter of the j=54 emission : @xmath95 where r is the radius of the cloud and @xmath96 is the mean mass per particle .
the second estimate uses the cs column densities ( n ) and the formula : @xmath97 where x is the abundance of cs .
we have used @xmath98 , based on a more detailed analysis of one of the sources in this study ( carr et al .
1995 ) . finally , we estimated masses from the virial theorem : @xmath99 for a spherical , non - rotating cloud . assuming that the velocity profile is gaussian , @xmath100 is related to the fwhm ( @xmath101 ) of the line by @xmath102 .
we used the average @xmath101 of the cs lines .
the value of @xmath103 for gl490 is probably underestimated substantially because the maser position is quite far from the peak of a very compact source .
zhou et al . (
1996 ) have analyzed this source in more detail and found considerably higher densities from spectra on the peak .
consequently , we ignore this source in the following discussion .
the average ratio of @xmath104 is @xmath105 .
the agreement is gratifying , but the poorly known abundance of cs makes @xmath106 quite uncertain .
in contrast , the agreement between @xmath103 and @xmath107 is worse , with @xmath103 almost always considerably larger than @xmath107 .
a likely explanation is that the gas is distributed inhomogeneously within the beam , whereas the calculation of @xmath103 assumes that the density is uniformly distributed .
we have used the ratio of @xmath107 to @xmath103 to estimate the volume filling factor ( @xmath108 ) of the gas , also listed in table 7 .
the filling factors have a large range ( 0.02 to 2.3 ) and a mean value of @xmath109 .
the virial mass estimate is susceptible to error because the linewidth may be affected by unbound motions , such as outflows , and it ignores effects of external pressure .
least certain is @xmath103 , which depends on the cube of the size ( and hence distance ) .
each mass estimate depends on a different power of the size , making their ratio strongly dependent on uncertainties in the distance . in view of the problems
inherent in each of the different mass calculations , the masses agree reasonably well . because the virial mass estimates have the fewest potential problems
, we will use them in what follows .
the average @xmath110 m@xmath1 .
are the high densities seen in this survey peculiar to regions of massive star formation or are they a feature of star formation in general ?
lada , evans , & falgarone ( 1996 ) have found that the density in the most active star - forming cores in l1630 is about log(n ) = 5.8 , very similar to what we find .
we also compared the results of our study with surveys of regions forming low - mass stars . observed a sample of low - mass cores in cs transitions up to j=54 and derived densities of @xmath111 .
these densities are about a factor of 4 lower than the densities we find in this study ( and in other studies of regions of massive star formation ) . since zhou et al .
( 1989 ) did not have j=76 data , it may be more appropriate to compare with our fits to sources without j=76 detections ; in that case , our densities are larger by a factor of about 2 .
the net result is that regions forming massive stars do seem to have larger densities when similar techniques are used , but the difference is not an order of magnitude .
the ability to form low - mass stars in regions of massive star formation may depend on whether the jeans mass remains low as the cloud is heated .
we can calculate the jeans mass from @xmath112 using the mean logarithmic densities and the assumed temperatures ( 10 k for the low - mass cores , 50 k for our sample ) , we compute @xmath113 m@xmath1 for the clouds forming low - mass stars and @xmath114m@xmath1 for clouds in this study with j=76 emission .
the assumed temperatures make @xmath115 higher in regions forming massive stars even though they are denser .
however , the strong dependence of @xmath115 on temperature means that statements about average properties should not be taken too literally until the temperatures are known better .
in addition , the fragmentation spectrum may have been established early in the evolution of the core , before the temperatures were raised by the formation of massive stars .
most studies of the global properties of molecular clouds deal with the usual linewidth size density relations , as proposed by larson ( 1981 ) and confirmed by others ( e.g. , fuller & myers 1992 ; solomon et al .
1987 ; caselli & myers 1995 ) .
these relations were generally found by comparing properties of whole clouds ; similar relations were found within single clouds by comparing map sizes in transitions of different molecules .
a recent paper by caselli & myers ( 1995 ) includes information on both low mass cores and more massive cores within the orion molecular cloud .
they fit the non - thermal linewidth ( the observed linewidth after correction for the thermal contribution ) and cloud radius for these types of regions separately to this relation : @xmath116 they found a strong relationship ( correlation coefficient , @xmath117 ) in low - mass cores with @xmath118 and @xmath119 .
the relation was considerably weaker ( @xmath120 ) and flatter ( @xmath121 ) in the massive cores . in figure 5 , we plot log(@xmath122 ) versus log@xmath123 for the sources in table 5 , which are generally denser and more massive than the cores studied by caselli & myers .
no relationship is apparent ( the correlation coefficient is only @xmath124 ) , despite the fact that our sample covers a range of 30 in source size .
nevertheless , we fitted the data to equation 7 using least squares and considering uncertainties in both variables ( we assumed 20% uncertainties in size and used the standard deviation of the linewidths of the different lines for the uncertainty in @xmath122 ) .
the result was @xmath125 and @xmath126 , but the goodness of fit parameter , @xmath127 , whereas a decent fit should have @xmath128 .
alternatively , we minimized the mean absolute deviation ( robust estimation ; see press et al .
the result was @xmath129 and @xmath130 , indicating essentially no size linewidth relation .
thus our data confirm the trend discernable in the analysis of caselli & myers : the @xmath131 relation tends to break down in more massive cores .
we have plotted the caselli & myers relations in figure 5 , along with larson s original relation .
it is clear that our sources have systematically higher linewidths at a given radius than sources in other studies . for the radii we are probing
, most other studies were considering linewidths from co or its isotopes and may thus have included a larger contribution from low - density envelopes .
the usual relations would predict larger @xmath122 in these regions , which would make the discrepancy worse . however , our sources are regions of massive star formation , and larson ( 1981 ) noted that such regions ( orion and m17 in his study ) tended to have larger @xmath122 for given size and not to show a size
linewidth correlation .
most previous studies have found an inverse relation between @xmath132 density and size , corresponding to a constant column density . however , scalo ( 1990 ) and kegel ( 1989 ) have noted that selection effects and limited dynamic range may have produced this effect , and leisawitz ( 1990 ) found no relationship between density and size in his study of clouds around open clusters . in previous studies ,
the @xmath132 densities were found by dividing a column density by a size , which might be expected to introduce an inverse correlation if the column density tracer has a limited dynamic range . since our densities
were derived from an excitation analysis , it may be interesting to see if any correlation exists in our data .
we plot log(n ) versus log@xmath123 in figure 5 .
again , no correlation is evident ( @xmath133 ) , and our densities all lie well above ( factors of 100 ! ) predictions from previous relations ( e.g. , myers 1985 ) .
again , larson ( 1981 ) noted a similar , though much less dramatic , tendency for regions of massive star formation in his analysis .
for a recent theoretical discussion of these relations , see vzquez - semadeni , ballesteros - paredes , & rodrguez ( 1997 ) . to use data on sources without size information , we plot ( in the bottom panel of figure 5 ) log(@xmath122 ) versus log(n ) .
the previous relations would predict a negative slope ( typically @xmath134 ) in this relation .
in contrast to the predictions , our data show a positive , but small , correlation coefficient ( @xmath135 ) .
the slope from a least squares fit is quite steep ( @xmath136 ) , but robust estimation gives a slope of only 0.39 .
in addition , the linewidths are much larger than would have been predicted for these densities from previous relations .
these results suggests that an uncritical application of scaling relations based on _ mean _ densities to actual densities , especially in regions of massive star formation , is likely to lead to errors .
the fact that larson s laws are not apparent in our data indicate that conditions in these very dense cores with massive star formation are very different from those in more local regions of less massive star formation .
the linewidths may have been affected by star formation ( outflows , expanding hii regions , etc . ) ; the higher densities are probably caused by gravitational contraction , which will also increase the linewidths .
while the regions in this study may not be typical of most molecular gas , they are typical of regions forming most of the massive stars in the galaxy .
these conditions ( denser , more turbulent than usually assumed ) may be the ones relevant for considerations of initial mass functions .
we have collected data from the literature ( or our own unpublished data ) on the luminosity of the sources in table 7 .
the ratio of the luminosity to the virial mass ( @xmath2 ) , roughly proportional to the star formation rate per unit mass , ranges from 24 to 490 in solar units ( see table 7 ) with a mean of @xmath137 , where 43 is the standard deviation of the mean ( all other uncertainties quoted in the text are standard deviations of the distribution ) .
previous studies , using masses determined from co luminosity , have found much lower average values of @xmath2 : 4.0 for the inner galaxy ( mooney & solomon 1988 ) ; 1.7 for the outer galaxy ( mead , kutner , & evans 1990 ) .
in fact , the maximum values in those samples were 18 and 5 , respectively , smaller than any of our values .
the enormous difference is caused by the fact that we are calculating the mass of the dense gas , which is much less than the mass computed from the co luminosity . while we have also tried to use luminosities measured with small beams ,
the main difference is in the mass .
one way to interpret this result is that the star formation rate per unit mass rises dramatically ( a factor of 50 ) in the part of the cloud with dense gas .
the star formation rate per unit mass of very dense gas may be more relevant since stars do not seem to form randomly throughout molecular clouds ( lada et al .
1991 ) . instead
, the 4 most massive cs cores in l1630 , which cover only 18% of the surveyed area , contain 58% to 98% of all the forming stars , depending on background correction .
li et al . (
1996 ) have found that there is little evidence for any recent star formation outside the clusters , suggesting that the 98% number is closer to correct .
the star formation efficiency in the clusters can be quite high ( e.g. , 40% ) compared to that of the cloud as a whole ( 4% ) ( lada et al .
1991 ) .
the gas depletion time ( @xmath138 ) is the time to turn all the molecular gas into stars .
considering only stars of @xmath139 m@xmath1 , the star formation rate can be written as @xmath140 ( m@xmath1 yr@xmath43 ) = @xmath141 ( gallagher & hunter 1986 ; hunter et al .
the coefficient differs by only 20% if the lower mass cutoff is 10 m@xmath1 .
the gas depletion time can then be written as @xmath142 yr .
using our value of average @xmath143 , @xmath144 yr .
this time is comparable to that for dispersal of clouds surrounding open clusters ; clusters with ages in excess of @xmath145 yr do not have associated molecular clouds with masses as large as @xmath146 ( leisawitz , bash , & thaddeus 1989 ) .
cs j = 54 emission has been seen toward the centers of ngc 253 , m82 , ic 342 , maffei 2 , and ngc 6946 ( . for comparison to studies of other galaxies
, we will estimate the luminosity of the milky way in cs 54 [ @xmath147 from the mean per cloud in table 5 and an estimate of the number of such clouds ( @xmath148 ) in the galaxy . from table 5
we find @xmath149 l@xmath1 and @xmath150 k km s@xmath43 , whereas @xmath151 k km s@xmath43 for the whole sample in table 2 .
if we correct for the fact that the mean integrated intensity of the sources in table 5 is less than the mean of the whole sample , we would get @xmath152 l@xmath1 for the typical core .
we do not have a direct measurement of @xmath148 because our survey is incomplete .
the most recent update to the maser catalog ( brand et al .
1994 ) brings the total number of masers with iras colors characteristic of star formation regions ( see palagi et al .
1993 ) to 414 .
if we assume that our cs 54 detection rate of 75% applies equally to the other sources , we would expect 311 regions of cs j = 54 emission in a region which covers two thirds of the galaxy .
if we correct for the unsurveyed third of the galaxy , we would estimate the total number of cloud cores emitting cs j = 54 to be 466 .
consequently , we will assume @xmath153 , with the larger values probably being more likely .
using these numbers , we calculate @xmath154 l@xmath1 .
even though we have made some completeness corrections , we expect these estimates to be underestimates because of our limited sensitivity and the likelihood of cs emission from dense regions without masers .
these values can be compared with the luminosities of other galaxies in table 8 .
however , our estimate applies to the entire galaxy excluding the inner 400 pc , while the for other galaxies are derived from a single beam , centered on the nucleus , with a radius given in the table .
the inner 100 pc of m82 and ngc 253 emit more cs j = 54 than does our entire galaxy , excluding the inner 400 pc .
we can also compare our galaxy to others in terms of its star formation rate per unit mass . in
5.3 , we used @xmath2 , with @xmath155 as the virial mass , to measure this quantity .
because linewidths in galaxy observations are likely to reflect the total mass , rather than the gaseous mass , we will use @xmath59/ as a stand - in for the star formation rate per unit mass of dense gas .
we have tabulated the far - infrared luminosity of the galaxies in table 8 , using the data with the smallest available beam , to provide the best match to the cs j = 54 observations , which were mostly done with the iram 30-m telescope ( @xmath156 beam ) .
the resulting @xmath59/ values range from @xmath157 ( ngc 253 ) to @xmath158 ( m82 ) .
these numbers apply to regions typically 100 pc in radius . for our galaxy , we have only the total , so we compare to the total @xmath159 l@xmath1 ( wright et al .
the result is @xmath160 , nominally similar to m82 ; however , much of the far - infrared emission of our galaxy is likely to result from heating by older stars .
probably a more useful comparison is to the values of @xmath59/ in individual clouds ( table 7 ) .
no individual cloud approaches the value in m82 .
the highest value in table 7 is about twice that of ngc 253 and half that of ic 342 .
very dense gas is common in regions of massive star formation .
the gas density for the regions selected by having a water maser is @xmath161 and the cs column density is @xmath162 . for regions without cs j
= 76 emission the mean density is half as large and the mean column density is about 7 times smaller .
these results are relatively insensitive to both cs optical depth and to changes in the kinetic temperature of the region .
the mean density is an order of magnitude less than the critical density of the j = 76 line because of trapping and multilevel excitation effects . 2 . in many regions forming massive stars ,
the cs emission is well modeled by a single density gas component , but many sources also show evidence for a range of densities . from simulations of emission from gas composed of two different densities ( @xmath37 and @xmath38 ) , we conclude that there are few clouds with filling factors of ultra - dense gas ( n@xmath163 ) exceeding 0.25 . 3 .
the densities calculated for the sources in this survey are comparable to the densities seen from detailed studies of a few individual regions forming massive stars .
therefore , it is likely that very dense gas is a general property of such regions .
the average density of regions forming massive stars is at least twice the average in regions forming only low - mass stars .
4 . using a subsample of sources whose cs 54 emission was mapped at the cso , the average cloud diameter is 1.0 pc and the average virial mass is 3800 m@xmath1 .
we see no evidence for a correlation between linewidth and size or density and size in our sample .
our linewidths and densities are systematically larger at a given size than those predicted by previous relations .
there is , however , a positive correlation between linewidth and density , the opposite of predictions based on the usual arguments .
the ratio @xmath2 , which is a measure of star formation rate per unit mass for the dense gas probed by cs j=54 emission , ranges from 24 to 490 , with an average value of 190 . 7 . the dense gas depletion time , @xmath164 yr , comparable to the dispersal time of gas around clusters and ob associations .
the estimated galactic luminosity in the cs j = 54 line is @xmath165 l@xmath1 .
this range of values is considerably less than what is seen in the inner 100 pc of starburst galaxies . in addition
, those galaxies have a higher ratio of far - infrared luminosity to cs j = 54 luminosity than any cloud in our sample .
we are grateful to the staff of the iram 30-m telescope for assistance with the observations .
we also thank t. xie , c. m. walmsley , and j. scalo for helpful discussion .
this research was supported in part by nsf grant ast-9317567 to the university of texas at austin .
cs 21 & 97.980968 & iram & 0.60 & 25 & 675 & 0.31@xmath171 & 3.06@xmath172 cs 32 & 146.969049 & iram & 0.60 & 17 & 990 & 0.32@xmath173 & 2.04@xmath172 cs 54 & 244.935606 & iram & 0.45 & 10 & 2500 & 1.22@xmath174 & c@xmath0s 21 & 96.412982 & iram & 0.60 & 25 & 620 & 0.31@xmath175 & 3.11@xmath172 c@xmath0s 32 & 144.617147 & iram & 0.60 & 17 & 835 & 0.32@xmath176 & 2.07@xmath172 c@xmath0s 54 & 241.016176 & iram & 0.45 & 10 & 2700 & 1.24@xmath174 & cs 54 & 244.935606 & cso & 0.71 & 30 & 445 & 0.17@xmath177 & 1.2@xmath178 c@xmath0s 76 & 337.396602 & cso & 0.55 & 20 & 1000 & 0.12@xmath177 & 0.89@xmath178 cs 109 & 489.75104 & cso & 0.39 & 14 & 4300 & 0.09@xmath177 & 0.61@xmath178 cs 1413 & 685.434764 & cso & 0.31 & 11 & 2050 & 0.06@xmath177 & 0.44@xmath178
l c c c c c source & @xmath179t@xmath180dv & @xmath181 & fwhm & @xmath182 & @xmath27(14 - 13 ) & ( k km s@xmath43 ) & ( km s@xmath43 ) & ( km s@xmath43 ) & ( k ) & ( k ) + gl2591@xmath183 & 2.7 & -5.3 & 1.6 & 1.6 & s158a & 22.6 & -57.2 & 2.9 & 7.2 & w3(2 ) & 6.4 & -38.49 & 2.28 & 2.6 & w3(oh ) & & & & & @xmath184 s255 & 10.3 & 8.2 & 2.3 & 4.4 & @xmath185 l c c c c c c c source & ref & dist . &
@xmath179t@xmath180dv & & beam corr .
& diameter & offset & & ( kpc ) & ( k km s@xmath43 ) & ( 10@xmath186 l@xmath1 ) & & ( pc ) & ( arcsec ) + w43s & & 8.5 & 52.8 & 6.1 & 1.5 & 0.9 & ( 0,5 ) w43main1 & & 7.5 & 22.1 & 5.2 & 4.0 & 1.9 & ( 20,-36 ) w43main3 & & 6.8 & 32.4 & 4.6 & 2.9 & 1.4 & ( -8,2 ) 31.25 - 0.11 & & 13 & 9.0 & 5.7 & 3.6 & 3.0 & ( -12,-15 ) 31.44 - 0.26 & & 9.4 & 23.0 & 8.6 & 4.0 & 2.4 & ( -2,-4 ) 32.8 + 0.2a & & 15 & 64.1 & 15 & 1.0 & @xmath1871.1 & ( -5,-4 ) w44 & & 3.7 & 87.9 & 3.1 & 2.5 & 0.7 & ( -3,0 ) w51w & & 7 & 12.0 & 1.6 & 2.6 & 1.3 & ( 0,-7 ) w51n & & 7 & 79.3 & 17 & 4.2 & 1.8 & ( 0,-5 ) w51 m & & 7 & 152 & 19 & 2.4 & 1.2 & ( -3,-2 ) on1 & & 6 & 24.4 & 1.6 & 1.7 & 0.7 & ( 0,0 ) k3 - 50 & & 9 & 11.3 & 1.9 & 2.0 & 1.3 & ( -5,5 ) on3 & & 9 & 11.0 & 1.8 & 2.0 & 1.3 & ( 0,-4 ) on2s & & 5.5 & 22.3 & 1.5 & 2.2 & 0.9 & ( -6,0 ) on2n & & 5.5 & 15.4 & 1.0 & 2.1 & 0.8 & ( 6,5 ) s106 & & 0.6 & 5.4 & 0.004 & 2.2 & 0.1 & ( 20,0 ) crl 2591 & 1 & 1.0 & 7.9 & 0.024 & 3.3 & 0.22 & ( 0,0 ) dr21 s & & 3 & 44.8 & 1.0 & 2.3 & 0.5 & ( -6,5 ) w75(oh ) & & 3 & 47.6 & 1.1 & 2.4 & 0.5 & ( -6,-5 ) w75s1 & & 3 & 9.4 & 0.9 & 9.7 & 1.3 & ( -3,7 ) w75s3 & & 3 & 6.8 & 0.2 & 3.2 & 0.7 & ( 0,0 ) w75n & & 3 & 35.2 & 0.8 & 2.5 & 0.5 & ( -5,6 ) cepa & 2 & 0.73 & 30.0 & 0.1 & 5.5 & 0.2 & ( 10,12 ) w3(2 ) & 2 & 2.3 & 26.3 & 0.8 & 5.5 & 0.7 & ( 0,12 ) gl 490 & 2 & 0.9 & 7.5 & 0.01 & 1.8 & 0.12 & ( -14,-12 ) l c c c c c c c c c source & flag & @xmath103 & @xmath106 & @xmath107 & @xmath108 & @xmath59 & ref . & @xmath188 & @xmath59/ & & ( @xmath189 ) & ( @xmath189 ) & ( @xmath189 ) & & ( @xmath190 ) & & ( @xmath191 ) & ( @xmath192 ) + w43 s & c@xmath0s & @xmath193 & @xmath194 & @xmath195 & 0.08 & & & & [email protected] & c@xmath0s & @xmath197 & @xmath198 & @xmath199 & 0.02 & & & & 32.8 + 0.20 a & c@xmath0s & @xmath200 & @xmath201 & @xmath202 & 0.13 & & & & w44 & c@xmath0s & @xmath203 & @xmath204 & @xmath205 & 0.27 & @xmath206 & 4 & 67 & 1.0 w51 w & c@xmath0s & @xmath207 & @xmath208 & @xmath209 & 0.03 & & & & w51 n c2 & c@xmath0s & @xmath210 & @xmath198 & @xmath201 & 0.02 & @xmath211 & 4 & 310 & 2.4 w51 m & cs & @xmath212 & @xmath213 & @xmath203 & 0.23 & @xmath214 & 3 & 170 & 1.5 k3@xmath19650 & c@xmath0s & @xmath207 & @xmath215 & @xmath216 & 0.10 & @xmath217 & 5 & 340 & 11 on3 & c@xmath0s & @xmath218 & @xmath219 & @xmath220 & 0.13 & @xmath221 & 5 & 160 & 2.1 on2s&c@xmath0s & @xmath222 & @xmath223 & @xmath224 & 0.03 & & & & crl2591 & cs & @xmath225 & @xmath226 & @xmath227 & 1.1 & @xmath228 & 2 & 63 & 8.3 dr21 s & c@xmath0s & @xmath229 & @xmath230 & @xmath231 & 0.31 & @xmath232 & 6 & 460 & 5.0 w75(oh ) & c@xmath0s & @xmath233 & @xmath234 & @xmath235 & 0.27 & @xmath236 & 8 & 35 & 0.5 w75 n & c@xmath0s & @xmath237 & @xmath238 & @xmath239 & 0.22 & @xmath240 & 3 & 130 & 2.3 cep a & cs & @xmath241 & @xmath242 & @xmath243 & 2.3 & @xmath244 & 1 & 42 & 2.5 w3(2 ) & c@xmath0s & @xmath245 & @xmath246 & @xmath247 & 0.03 & @xmath206 & 7 & 490 & 3.8 gl490 & cs & @xmath248 & @xmath249 & @xmath250 & 15 & @xmath251 & 2 & 24 & 2.2 ngc 253 & 2.5@xmath183 & 67 & 23.5 & 154 & 1 & 8 & 3 & 5 maffei 2 & 5 & 133 & @xmath1872 & @xmath18753 & 2 & 9.5 & 2 & @xmath254 ic 342 & 1.8@xmath255 & 48 & 0.76 & 3 & 1 & 0.64 & 4 & 21 m82 & 3.2@xmath171 & 85 & 2.6 & 28 & 1 & 47 & 3 & 170 ngc 6946 & 5 & 133 & @xmath1872.8 & @xmath18774 & 2 & 1.2 & 3 & @xmath256 | we have observed 150 regions of massive star formation , selected originally by the presence of an maser , in the j = 54 , 32 , and 21 transitions of cs , and 49 regions in the same transitions of c@xmath0s . over 90% of the 150 regions
were detected in the j = 21 and 32 transitions of cs and 75% were detected in the j=54 transition .
we have combined the data with the j = 76 data from our original survey ( plume et al .
1992 ) to determine the density by analyzing the excitation of the rotational levels .
using large velocity gradient ( lvg ) models , we have determined densities and column densities for 71 of these regions .
the gas densities are very high ( = 5.9 ) , but much less than the critical density of the j=76 line .
small maps of 25 of the sources in the j = 54 line yield a mean diameter of 1.0 pc .
several estimates of the mass of dense gas were made for the sources for which we had sufficient information .
the mean virial mass is 3800 m@xmath1 .
the mean ratio of bolometric luminosity to virial mass ( @xmath2 ) is 190 , about 50 times higher than estimates using co emission , suggesting that star formation is much more efficient in the dense gas probed in this study .
the gas depletion time for the dense gas is roughly @xmath3 yr , comparable to the timescale for gas dispersal around open clusters and ob associations .
we find no statistically significant linewidth size or density size relationships in our data .
instead , both linewidth and density are larger for a given size than would be predicted by the usual relationships .
we find that the linewidth * increases * with density , the opposite of what would be predicted by the usual arguments .
we estimate that the luminosity of our galaxy ( excluding the inner 400 pc ) in the cs j = 54 transition is 15 to 23 l@xmath1 , considerably less than the luminosity in this line within the central 100 pc of ngc 253 and m82 .
in addition , the ratio of far - infrared luminosity to cs luminosity is higher in m82 than in any cloud in our sample . |
let @xmath6 be the set of polynomials of degree @xmath1 with integer coefficients and all zeros in the closed unit disk @xmath7 .
we denote the subset of @xmath6 with simple zeros by @xmath8 . given @xmath9 , we write @xmath10 if @xmath11 and @xmath12 ( respectively @xmath13 if @xmath11 and @xmath14 ) .
schur @xcite studied the limiting behavior of the arithmetic means @xmath0 of zeros for polynomials from @xmath15 as @xmath16 where @xmath9 is an arbitrary fixed number .
he showed that @xmath17 and remarked that this @xmath18 is equal to @xmath19 for _ monic _ polynomials from @xmath6 by kronecker s theorem @xcite .
we prove that @xmath20 for any sequence of polynomials from schur s class @xmath21 this result is obtained as a consequence of the asymptotic equidistribution of zeros near the unit circle .
namely , if @xmath22 are the zeros of @xmath23 , we define the counting measure @xmath24 , where @xmath25 is the unit point mass at @xmath26 .
consider the normalized arclength measure @xmath27 on the unit circumference @xmath28 , with @xmath29 if the @xmath30 converge weakly to @xmath27 as @xmath31 ( @xmath32 ) then @xmath33 thus schur s problem is solved by the following result .
[ thm1.1 ] if @xmath34 satisfy @xmath35 , then @xmath32 as @xmath36 ideas on the equidistribution of zeros date back to jentzsch and szeg , cf .
they were developed further by erds and turn @xcite , and many others ; see @xcite for history and additional references .
more recently , this topic received renewed attention in number theory , e.g. in the work of bilu @xcite .
if the leading coefficients of polynomials are bounded , then we can allow even certain multiple zeros .
define the multiplicity of an irreducible factor @xmath37 of @xmath23 as the integer @xmath38 such that @xmath39 divides @xmath23 , but @xmath40 does not divide @xmath23 . if a factor @xmath37 occurs infinitely often in a sequence @xmath41 then @xmath42 means @xmath43 if @xmath37 is present only in finitely many @xmath23 , then @xmath42 by definition .
[ thm1.2 ] assume that @xmath44 . if every irreducible factor in the sequence of polynomials @xmath23 has multiplicity @xmath45 , then @xmath32 as @xmath36 [ cor1.3 ] if @xmath46 satisfy the assumptions of theorem [ thm1.1 ] or [ thm1.2 ] , then @xmath47 we also show that the norms @xmath48 have at most subexponential growth . [ cor1.4 ] if @xmath41 satisfy the assumptions of theorem [ thm1.1 ] or theorem [ thm1.2 ] , then @xmath49 this result is somewhat unexpected , as we have no direct control of the norm or coefficients ( except for the leading one ) .
for example , @xmath50 has norm @xmath51 .
we now consider quantitative aspects of the convergence @xmath32 . as an application
, we obtain estimates of the convergence rate of @xmath0 to @xmath19 in schur s problem . a classical result on
the distribution of zeros is due to erds and turn @xcite . for @xmath52 with @xmath53 let @xmath54 be the number of zeros in the sector @xmath55 where @xmath56 erds and turn @xcite proved that @xmath57 the constant @xmath58 was improved by ganelius , and @xmath59 was replaced by weaker integral norms by amoroso and mignotte ; see @xcite for more history and references .
our main difficulty in applying to schur s problem is the absence of an effective estimate for @xmath60 .
we prove a new discrepancy " estimate via energy considerations from potential theory .
these ideas originated in part in the work of kleiner , and were developed by sjgren and hsing , see ( * ? ? ?
* ch . 5 ) .
we also use the mahler measure of a polynomial @xmath61 , defined by @xmath62 note that @xmath63 , where @xmath64 .
jensen s formula readily gives @xmath65 @xcite .
hence @xmath66 for any @xmath67 [ thm2.1 ] let @xmath68 satisfy @xmath69 and @xmath70 if @xmath61 is a polynomial with integer coefficients and simple zeros , then @xmath71 this theorem is related to recent results of favre and rivera - letelier @xcite , obtained in a different setting .
choosing @xmath72 appropriately , we obtain an estimate of the means @xmath0 in schur s problem .
[ cor2.2 ] if @xmath73 then @xmath74 we also have an improvement of corollary [ cor1.4 ] for schur s class @xmath15 .
[ cor2.3 ] if @xmath75 then there is some @xmath76 such that @xmath77 as @xmath36 the proof of theorem [ thm2.1 ] gives a result for arbitrary polynomials with simple zeros , and for any continuous @xmath72 with finite dirichlet integral @xmath78=\iint(\phi_x^2 + \phi_y^2)\,da$ ] .
moreover , all arguments may be extended to general sets of logarithmic capacity @xmath79 , e.g. to @xmath80 $ ] .
using the characteristic function @xmath81 , we can prove general discrepancy estimates on arbitrary sets , and obtain an erds - turn - type theorem .
our results have a number of applications to the problems on integer polynomials considered in @xcite .
_ proof of theorem [ thm1.1 ] .
_ observe that the discriminant @xmath82 is an integer , as a symmetric form in the zeros of @xmath23 . since @xmath23 has simple roots ,
we have @xmath83 and @xmath84 using weak compactness , we assume that @xmath85 where @xmath86 is a probability measure on @xmath7 . let @xmath87 since @xmath88 we obtain for the energy of @xmath86 that @xmath89 & : = -\iint \log|x - t|\,d\tau(x)\,d\tau(t ) = \lim_{m\to\infty } \left ( \lim_{n\to\infty } \iint k_m(x , t)\ , d\tau_n(x)\,d\tau_n(t ) \right ) \\ & = \lim_{m\to\infty } \left ( \lim_{n\to\infty } \left ( \frac{1}{n^2 } \sum_{j\neq k } k_m(\alpha_j,\alpha_k ) + \frac{m}{n } \right ) \right ) \le \lim_{m\to\infty } \left ( \liminf_{n\to\infty } \frac{1}{n^2 } \sum_{j\neq k } \log\frac{1}{|\alpha_j-\alpha_k| } \right ) \\ & = \liminf_{n\to\infty } \frac{1}{n^2 } \log\frac{|a_n|^{2n-2}}{\delta(p_n ) } \le \liminf_{n\to\infty } \frac{1}{n^2}\log|a_n|^{2n-2 } = 0.\end{aligned}\ ] ] thus @xmath90\le 0 $ ] . but @xmath91>0 $ ] for any probability measure @xmath92 on @xmath7 , except for @xmath27 @xcite .
hence @xmath93 _ proof of theorem [ thm1.2 ] .
_ let @xmath94 note that for any @xmath95 there are finitely many irreducible factors @xmath37 in the sequence @xmath23 such that @xmath96 where @xmath97 is the zero counting measure for @xmath37 .
indeed , if we have an infinite sequence of such @xmath98 , then @xmath99 , as there are only finitely many @xmath100 of bounded degree .
hence @xmath101 by theorem [ thm1.1 ] .
let the number of such exceptional factors @xmath98 be @xmath102 .
then we have @xmath103 hence @xmath104 and @xmath105 after letting @xmath106 let @xmath120 be the potential of a measure @xmath121 a direct evaluation gives that @xmath122 and @xmath123 @xcite .
consider @xmath124 one computes ( or see @xcite ) that @xmath125 where @xmath126 is the arclength on @xmath127 , and @xmath128 are the inner and the outer normals .
we now use green s identity @xmath129 with @xmath130 and @xmath131 in each component @xmath132 of @xmath133 since @xmath134 in @xmath132 , adding the identities for all @xmath132 , we obtain that @xmath135}\,\sqrt{d[p_{\sigma}]},\ ] ] where @xmath78=\iint(\phi_x^2 + \phi_y^2)\,da$ ] is the dirichlet integral of @xmath136 it is known that @xmath137=2\pi i[\sigma]$ ] ( * ? ? ?
* thm 1.20 ) , where @xmath138=-\iint \log|z - t|\,d\sigma(z)\,d\sigma(t ) = \int p_{\sigma}\,d\sigma$ ] . since @xmath123
, we observe that @xmath139 so that @xmath138=\int p_{\tau_n^r}\,d\tau_n^r - 2\int p_{\mu}\,d\tau_n^r.$ ] further , @xmath140 we also have that @xmath141 we next combine the energy estimates to obtain @xmath142
\le \frac{2}{n}\log m(p_n ) -
\frac{1}{n^2}\log\left|a_n^2 \delta(p_n)\right| - \frac{1}{n}\log{r } + 4r.\ ] ] collecting all estimates , we proceed with @xmath143}\,\sqrt{d[p_{\sigma}]}/(2\pi ) = \omega_{\phi}(r ) + \sqrt{d[\phi]}\,\sqrt{i[\sigma]/(2\pi)}$ ]
. thus we arrive at the main inequality : @xmath144}{2\pi}}\,\left(\frac{2}{n}\log m(p_n ) -
\frac{1}{n^2}\log\left|a_n^2 \delta(p_n)\right| - \frac{1}{n}\log{r } + 4r\right)^{1/2}.\end{aligned}\ ] ] note that @xmath78\le 2\pi r^2 a^2,$ ] as @xmath145 and @xmath146 a.e . in @xmath147
also , @xmath148 since @xmath149 and @xmath150 , we have @xmath151 .
hence follows from by letting @xmath152 _ proof of corollary [ cor2.2 ] . _ since @xmath23 has real coefficients , we have that @xmath153 we let @xmath154 and @xmath155 an elementary computation shows that @xmath156 and @xmath157 for all @xmath158 the mean value theorem gives @xmath159 hence we can use theorem [ thm2.1 ] with @xmath160 and @xmath161 _ proof of corollary [ cor2.3 ] .
_ note that @xmath162 for @xmath163 we let @xmath164 and @xmath165 then @xmath166 and the same estimates hold for @xmath167 hence @xmath78 = o\left(\iint_{|w|\le 1 } |z - w|^{-2 } da(w)\right ) = o\left(\int_{1/n}^1 r^{-1}dr \right ) = o(\log{n}),$ ] and @xmath168 let @xmath169 and use to obtain @xmath170 v. v. andrievskii and h .-
blatt , discrepancy of signed measures and polynomial approximation , springer - verlag , new york , 2002 .
y. bilu , limit distribution of small points on algebraic tori , duke math .
j. 89 ( 1997 ) , 465 - 476 .
p. borwein , computational excursions in analysis and number theory , springer - verlag , new york , 2002 .
p. erds and p. turn , on the distribution of roots of polynomials , ann . math .
51 ( 1950 ) , 105 - 119 .
c. favre and j. rivera - letelier , equidistribution quantitative des points de petite hauteur sur la droite projective , math .
335 ( 2006 ) , 311 - 361 ; corrigendum in math .
( 2007 ) , 799 - 801 .
l. kronecker , zwei stze ber gleichungen mit ganzzahligen cofficienten , j. reine ang .
53 ( 1857 ) , 173 - 175 .
n. s. landkof , foundations of modern potential theory , springer - verlag , new york , 1972 .
e. b. saff and v. totik , logarithmic potentials with external fields , springer - verlag , berlin , 1997 .
i. schur , ber die verteilung der wurzeln bei gewissen algebraischen gleichungen mit ganzzahligen koeffizienten , math .
zeit . 1 ( 1918 ) , 377 - 402 . | schur studied limits of the arithmetic means @xmath0 of zeros for polynomials of degree @xmath1 with integer coefficients and simple zeros in the closed unit disk .
if the leading coefficients are bounded , schur proved that @xmath2 we show that @xmath3 , and estimate the rate of convergence by generalizing the erds - turn theorem on the distribution of zeros . _ to cite this article : i. e. pritsker , c. r. acad .
sci .
paris , ser .
i 336 ( 2003 ) .
_ 0.5 * rsum * 0.5*moyennes de nombres algbriques dans le disque unit . * schur a tudi les limites des moyennes arithmtiques @xmath0 des zros pour les polynmes coefficients entiers de degr @xmath1 ayant des zros simples dans le disque unit ferm .
lorsque les coefficients dominants restent borns , schur a dmontr que @xmath4 .
nous prouvons que @xmath3 .
nous donnons une estimation du taux de convergence , gr@xmath5ce une gnralisation dun thorme de erds - turn sur la distribution des zros . _
pour citer cet article : i. e. pritsker , c. r. acad .
sci .
paris , ser .
i 336 ( 2003 ) . |
two - dimensional oscillator with potential @xmath0 is the simplest system with confinement .
we select this particular case @xmath1 because all formulas greatly simplify , for example , energy levels and wave function values in the origin are @xmath2 we will consider the regular quasi - perturbative method of sum rules ( sr ) to determine energy @xmath3 and @xmath4 of the ground state .
we follow here partially the lectures @xcite .
the general scheme of the sr method @xcite can be easily understood on the example of a correlator @xmath5 , which has the spectral expansion : @xmath6 suppose that we can construct the perturbative expansion of this correlator : @xmath7 where @xmath8 corresponds to `` free movement '' and has the spectral representation : @xmath9 the sum rule
it is simply @xmath10 usually it appears that higher state contributions can be well approximated by `` free states '' outside interval @xmath11 . as a result we have sr in the form : @xmath12 _ our aim _ : to determine @xmath4 and @xmath3 from this sr by calculating spectral density @xmath13 of `` free particle '' and coefficients @xmath14 by demanding stability of this sr in variable @xmath15}$ ] with appropriate value of @xmath16 . in order to select @xmath17 with these properties for our 2dqho
let us consider 2-time green function for all states with @xmath18 only @xmath19-states contribute to this sum . ]
@xmath20 which is the probability amplitude for the transition @xmath21 . to get @xmath5 we put @xmath22 , @xmath23 : @xmath24 in our case @xmath25 , so we have the _ exact result _ for our @xmath17 : @xmath26 how fast is convergence of spectral expansion ( [ eq : g.00.0mu ] ) for @xmath17 ?
let us estimate it for @xmath27 .
exact result ( [ eq : m.ho.exact ] ) gives us @xmath28 , whereas numerically @xmath29 ground state contributes @xmath30 , first excitation
@xmath31 , while the second only @xmath32 .
now we turn to the perturbative expansion of @xmath17 in powers of @xmath33 : @xmath34 here @xmath35 corresponds to the green function of free particle : @xmath36 numerically at @xmath37 we have @xmath38 first correction specifies free result by @xmath39 , while the second by @xmath40 .
this perturbative expansion can be rewritten @xmath41 that means _
asymptotic freedom _ : @xmath42 behaves like @xmath43 at large @xmath44 ! and it is violated by power corrections of the type @xmath45 .
we see in fig .
[ fig : ex.grst.pert2 ] , ( [ eq : m.ho.exact ] ) , is shown by the red solid line , ground state contribution only , @xmath46 , by the green dashed line , and @xmath47 by the blue dashed line .
right panel : ground @xmath48 1-st excited state contribution only , @xmath49 , is shown by the green dashed line , whereas @xmath50 by the blue dashed line .
[ fig : ex.grst.pert2],title="fig:",scaledwidth=49.0% ] , ( [ eq : m.ho.exact ] ) , is shown by the red solid line , ground state contribution only , @xmath46 , by the green dashed line , and @xmath47 by the blue dashed line .
right panel : ground @xmath48 1-st excited state contribution only , @xmath49 , is shown by the green dashed line , whereas @xmath50 by the blue dashed line .
[ fig : ex.grst.pert2],title="fig:",scaledwidth=49.0% ] that for large @xmath51 asymptotic freedom works well : @xmath52 , but we need more and more resonances to saturate @xmath53 . for small @xmath51 in spectral
part survives only ground state @xmath54 , but the perturbation expansion breaks down . in order to model higher resonances let us consider the spectral representation of our correlator @xmath17 : @xmath55 here the spectral density is just the sum of @xmath56-functions : @xmath57 analogously we have integral representation for free correlator : @xmath58 where @xmath59 .
asymptotic freedom dictates _
global duality _ of these two densities ( term `` global '' is related with integration over the whole spectrum ) @xmath60 at first glance these spectral densities have completely different behaviour , see fig .
[ fig : dual.sp.dens ] .
, ( [ eq : sp.den.osc ] ) , is shown by red solid vertical lines , imitating @xmath56-functions , whereas free spectral density , @xmath61 , by the blue solid line .
* right panel : * the phenomenological model @xmath62 , ( [ eq : sp.den.model ] ) , is shown by red solid vertical line , imitating @xmath56-function , and by the blue solid line , corresponding to @xmath61 and starting from the threshold @xmath16 . [
fig : dual.sp.dens],title="fig:",scaledwidth=49.0% ] , ( [ eq : sp.den.osc ] ) , is shown by red solid vertical lines , imitating @xmath56-functions , whereas free spectral density , @xmath61 , by the blue solid line .
* right panel : * the phenomenological model @xmath62 , ( [ eq : sp.den.model ] ) , is shown by red solid vertical line , imitating @xmath56-function , and by the blue solid line , corresponding to @xmath61 and starting from the threshold @xmath16 . [
fig : dual.sp.dens],title="fig:",scaledwidth=49.0% ] but we have very interesting relations between @xmath63-partial integral moments of this dual densities , namely , @xmath64 : @xmath65 for @xmath66 we have approximate relation @xmath67 $ ] .
so , we have duality between each excited resonance in oscillator and free particle in some spectral domain .
that means _ local duality_.
now we can model higher state contributions by `` higher states '' = `` free states '' outside interval @xmath11 or : @xmath68 and this gives us @xmath69 after all we have the following sr : @xmath70 or , equivalently , with @xmath71 : @xmath72 we also have a daughter sr produced from ( [ eq : sr1 ] ) by applying @xmath73 : @xmath74 if we divide ( [ eq : sr2 ] ) by ( [ eq : sr1 ] ) we obtain sr for @xmath75 @xmath76 the strategy of processing these srs is : * to determine @xmath77 by minimal sensitivity to variation of @xmath78 $ ] at appropriate @xmath79 ; * to determine @xmath80 by minimal sensitivity to variation of @xmath51 at appropriate @xmath16 .
how we should determine the _ fidelity window _
@xmath81 $ ] ?
power corrections are of the type @xmath82 and they are huge at @xmath83 .
we demand : @xmath84 higher states are not suppressed by @xmath85 at large @xmath86 .
we demand : @xmath87 then the fidelity window is @xmath88 : only for @xmath89 inside it is reasonable to demand minimal sensitivity of srs to variations in @xmath89 ! let us first consider _ sr setup with fixed @xmath90 _ : we fix the energy of the ground state to the exact value , @xmath91 , and obtain the following fidelity window , @xmath92 and @xmath93 , see fig .
[ fig : sr1.e0.fixed ] , left panel . the result @xmath94
is obtained with only 2 power corrections included and for @xmath95 ( the exact value is @xmath96 ) , see fig .
[ fig : sr1.e0.fixed ] , right panel . ) and ( [ eq : delta.hs.33 ] ) .
* right panel : * we show here the l. h. s. of eq .
( [ eq : sr1 ] ) with @xmath75 fixed at the exact value @xmath91 as a function of @xmath51 ( red solid line ) .
black dashed line corresponds to the position of the exact value of @xmath97 .
positions of @xmath98 and @xmath99 are shown by green vertical arrows in both panels and @xmath51 is displayed in units of @xmath100.[fig : sr1.e0.fixed],title="fig:",scaledwidth=49.0% ] ) and ( [ eq : delta.hs.33 ] ) .
* right panel : * we show here the l. h. s. of eq .
( [ eq : sr1 ] ) with @xmath75 fixed at the exact value @xmath91 as a function of @xmath51 ( red solid line ) .
black dashed line corresponds to the position of the exact value of @xmath97 .
positions of @xmath98 and @xmath99 are shown by green vertical arrows in both panels and @xmath51 is displayed in units of @xmath100.[fig : sr1.e0.fixed],title="fig:",scaledwidth=49.0% ] now we consider _ sr in the complete setup _ , that means that we determine the energy of the ground state from the daughter sr ( [ eq : sr.e0 ] ) .
we take into account 3 power corrections and obtain the fidelity window @xmath101}$ ] and @xmath102 for @xmath103 , see fig .
[ fig : sr12.comp ] , right panel . ) , @xmath104 , as a function of @xmath51 in units of @xmath100 ( red solid line ) .
black dashed line corresponds to the position of exact value of @xmath75 .
* right panel : * the l. h. s. of eq .
( [ eq : sr1 ] ) is shown as a function of @xmath51 in units of @xmath100 ( red solid line ) .
black dashed line corresponds to the position of exact value of @xmath97 .
[ fig : sr12.comp],title="fig:",scaledwidth=49.0% ] ) , @xmath104 , as a function of @xmath51 in units of @xmath100 ( red solid line ) .
black dashed line corresponds to the position of exact value of @xmath75 .
* right panel : * the l. h. s. of eq .
( [ eq : sr1 ] ) is shown as a function of @xmath51 in units of @xmath100 ( red solid line ) .
black dashed line corresponds to the position of exact value of @xmath97 .
[ fig : sr12.comp],title="fig:",scaledwidth=49.0% ] our conclusions about srs in quantum mechanics can be summarized as follows : * srs give @xmath3 and @xmath4 with accuracy _ not worse than 10% _ ; * main source of the error _ crude model _ for the spectral density of higher states : even taking into account 10 power corrections we obtain @xmath105 , @xmath106 , and @xmath107 ; * _ but _ : if we know @xmath108 exactly ( say , from particle data group ) , then accuracy can be twice higher : with taking into account 2 power corrections we obtain @xmath109 and @xmath110 ! * in qcd spectral density more close to perturbative !
in qcd we have a big problem : nobody knows how to analyze bound states .
the method of qcd srs allows us to calculate properties of hadrons ( masses , decay constants , magnetic moments ) without considering hadronization or confinement issues .
it was invented in 1977 by shifman , vainshtein & zakharov ( itep ) @xcite in order to describe @xmath111-meson , the @xmath112-system , discovered in 1974 in @xmath113-annihilation at spear ( slac ) and , in parallel , in @xmath114-collisions at bnl . in 1979
this method was applied to describe light hadrons in massless qcd @xcite .
_ main idea _ : to calculate correlators of hadron currents @xmath115|0}\rangle}$ ] by two approaches and to obtain the sr as the result of the matching .
we start with the fourier transformed correlator of two hadron currents @xmath116|0}\rangle}\right]\,dx\,,\end{aligned}\ ] ] where lorentz@xmath117 includes all lorentz structures , so that @xmath118 is scalar , and use the dispersion integral representation @xmath119 then we apply borel transform defined as @xmath120_{q^2=n m^2}\,.\end{aligned}\ ] ] here we list the most important examples : @xmath121 [ cols="^,^,^,^,^",options="header " , ] in the left panel of fig .
[ fig : lat.ren.cleo ] we demonstrate the @xmath122- , @xmath123- and @xmath124-contours ( solid , dotted and dashed contours around the best - fit point ( ) ) , which have been obtained for values of the twist-4 scale parameter @xmath125~\text{gev}^2 $ ] . as one sees from the blue dashed line within the hatched band , corresponding in this figure to the mean value of @xmath126 and its error bars , the nonlocal qcd sum - rules result with its error bars appears to be in good agreement with the cleo - constraints on @xmath127 at the @xmath122-level
. moreover , the estimate @xmath128 is close to @xmath129 , obtained in the data analysis of the electromagnetic pion form factor within the framework of a different lcsr method in @xcite .
these three independent estimates are in good agreement to each other , giving firm support that the cleo data processing , on one hand , and the theoretical calculations , on the other , are mutually consistent .
another possibility , suggested in @xcite , to obtain constraints on the pion da in the lcsr analysis of the cleo data to use for the twist-4 contribution renormalon - based model , relating it then to parameters @xmath130 and @xmath131 of the pion da . using this method we obtain @xcite the renormalon - based constraints for the parameters @xmath130 and @xmath131 , shown in the right panel of fig .
[ fig : lat.ren.cleo ] in a form of @xmath122-ellipses ( dashed contour ) .
_ new high - precision lattice measurements _ of the the pion da second moment @xmath132 appeared rather recently @xcite .
both groups extracted from their respective simulations , values of @xmath130 at the schmedding
yakovlev scale @xmath133 around @xmath134 , but with different error bars .
it is remarkable that these lattice results are in striking agreement with the estimates of @xmath130 both from nlc qcd srs @xcite and also from the cleo - data analyses based on lcsr@xcite , as illustrated in the right panel of fig .
[ fig : lat.ren.cleo ] , where the lattice results of @xcite are shown in the form of a vertical strip , containing the central value with associated errors , being smaller than in @xcite .
noteworthily , the value of @xmath130 of the displayed lattice measurements ( middle line of the strip ) is very close to the cleo best fit in @xcite ( ) .
it is worth to mention here the results of our analysis of the pion electromagnetic form factor using nlc dictated pion da and analytic perturbative qcd @xcite .
these results are in excellent agreement with cebaf data on pion form factor , shown as diamonds in the fig .
[ fig : ff.pi ] , where the green strip includes both the nlc qcd srs uncertainties , generated by our bunch of the allowed pion da , and by the scale - setting ambiguities at the nlo level . , scaledwidth=49.0% ] from the phenomenological point of view , the most interesting result here is that the bms pion da @xcite ( out of a `` bunch '' of similar doubly - peaked endpoint - suppressed pion das ) yields to predictions for the electromagnetic form factor very close to those obtained with the asymptotic pion da .
conversely , we see that a small deviation of the prediction for the pion form factor from that obtained with the asymptotic pion da does not necessarily imply that the underlying pion da has to be close to the asymptotic profile .
much more important is the behavior of the pion da in the endpoint region @xmath135 .
let me conclude with the following observations : * nlc qcd sr method for the pion da gives us the admissible bunches of das for each value of @xmath136 .
* nlo lcsr method produces new constraints on the pion da parameters ( @xmath130 and @xmath131 ) in conjunction with the cleo data . *
comparing results of the nlc srs with new cleo constraints allows to fix the value of qcd vacuum nonlocality : @xmath137 .
* this bunch of pion das agrees well with recent lattice data and with jlab f(pi ) data on the pion form factor .
i also suggest to the reader to look in the very interesting discussion of the qcd sr approach and its developments written by one of its creators @xcite .
this investigation was supported in part by the bogoliubov infeld programme , grant 2006 , by the heisenberg
landau programme , grant 2006 , and the russian foundation for fundamental research , grant no .
06 - 02 - 16215 .
1 a. v. radyushkin , in _ strong interactions at low and intermediate energies : proceedings of the 13th annual hugs at cebaf ( hugs 98 ) , 26 may2 jun 1998 , newport news , virginia _ , edited by j. l. goity ( world scientific , singapore , 2000 ) , pp .
. m. a. shifman , a. i. vainshtein , and v. i. zakharov , nucl .
phys . * b147 * , 385 ( 1979 ) ; ibid .
448 ; ibid . 519 .
v. a. novikov _
et al . _ ,
* 41 * , 1 ( 1978 ) .
b. l. ioffe and k. n. zyablyuk , nucl . phys . * a687 * , 437 ( 2001 ) ; + b. v. geshkenbein , b. l. ioffe , and k. n. zyablyuk , phys . rev . * d64 * , 093009 ( 2001 ) .
a. v. radyushkin , ubna preprint p2 - 10717 , 1977 [ hep - ph/0410276 ] .
s. v. mikhailov and a. v. radyushkin , nucl .
* b273 * , 297 ( 1986 ) .
e. p. kadantseva , s. v. mikhailov , and a. v. radyushkin , sov
. j. nucl .
phys . * 44 * , 326 ( 1986 ) .
d. mller , phys . rev . *
d49 * , 2525 ( 1994 ) ; ibid . *
d51 * , 3855 ( 1995 ) .
a. p. bakulev and n. g. stefanis , nucl .
b721 * , 50 ( 2005 ) .
s. v. mikhailov and a. v. radyushkin , jetp lett .
* 43 * , 712 ( 1986 ) ; sov .
. phys . * 49 * , 494 ( 1989 ) ; phys .
* d45 * , 1754 ( 1992 ) .
a. p. bakulev and s. v. mikhailov , phys . lett .
* b436 * , 351 ( 1998 ) . v. m. belyaev and b. l. ioffe , zhetf * 83 * , 876 ( 1982 ) .
a. a. ovchinnikov and a. a. pivovarov , sov
. phys . * 48 * , 721 ( 1988 ) .
a. a. pivovarov , bull .
. inst . * 5 * , 1 ( 1991 ) .
m. delia , a. i giacomo , and e. meggiolaro , phys .
rev . * d59 * , 054503 ( 1999 ) .
a. p. bakulev and s. v. mikhailov , phys . rev . * d65 * , 114511 ( 2002 ) .
a. p. bakulev , s. v. mikhailov , and n. g. stefanis , phys . lett . *
b508 * , 279 ( 2001 ) ; in _ proceedings of the 36th rencontres de moriond on qcd and hadronic interactions , 1724 mar 2001 , les arcs , france _
, edited by j. t. t. van ( world scientific , singapore , 2002 ) , pp .
133136 . v. l. chernyak and a. r. zhitnitsky , nucl . phys . *
b201 * , 492 ( 1982 ) ; ibid . *
b214 * , 547(e ) ( 1983 ) .
a. v. radyushkin and r. ruskov , nucl . phys . *
b481 * , 625 ( 1996 ) .
a. khodjamirian , eur .
j. * c6 * , 477 ( 1999 ) .
a. schmedding and o. yakovlev , phys . rev .
* d62 * , 116002 ( 2000 ) .
a. p. bakulev , s. v. mikhailov , and n. g. stefanis , phys . rev . * d67 * , 074012 ( 2003 ) ; phys
b578 * , 91 ( 2004 ) .
m. praszalowicz and a. rostworowski , phys . rev .
* d64 * , 074003 ( 2001 ) .
v. m. braun , a. khodjamirian , and m. maul , phys . rev . *
d61 * , 073004 ( 2000 ) .
j. bijnens and a. khodjamirian , eur .
j. * c26 * , 67 ( 2002 ) .
s. s. agaev , phys .
* d72 * , 114010 ( 2005 ) .
a. p. bakulev , s. v. mikhailov , and n. g. stefanis , phys . rev . * d73 * , 056002 ( 2006 ) .
l. del debbio , few body syst .
* 36 * , 77 ( 2005 ) .
m. gckeler _
et al . _ , talk in the workshop on light - cone qcd and nonperturbative hadron physics 2005 ( lc 2005 ) , cairns , queensland , australia , 715 jul 2005 [ hep - lat/0510089 ] .
a. p. bakulev _
et al . _ ,
phys . rev . * d70 * , 033014 ( 2004 ) .
j. volmer _
et al . _ ,
lett . * 86 * , 1713 ( 2001 ) . c. n. brown _ et al .
_ , phys . rev . *
d8 * , 92 ( 1973 ) . c. j. bebek _
et al . _ ,
phys . rev . *
d13 * , 25 ( 1976 ) . m. shifman , `` snapshots of hadrons or the story of how the vacuum medium determines the properties of the classical mesons which are produced , live and die in the qcd vacuum '' , lecture given at the 1997 yukawa international seminar on non - perturbative qcd - structure of the qcd vacuum , kyoto , december 212 , 1997 [ hep - ph/9802214 ] . | we illustrate the general scheme of the sum rule ( sr ) method using 2d quantum harmonic oscillator ( 2dqho ) as a toy model .
we introduce correlator , related to green function of 2dqho , and describe the property of asymptotic freedom for 2dqho .
we explain how the duality conception allows one to describe excited states
. finally we present numerical results and extract some lessons to learn from our exposition .
then we switch to the qcd and show that qcd srs supply us the method to study hadrons in non - perturbative qcd . here
the main emphasis is put on the pion and its distribution amplitude and form factors . |
1 . insertion energies @xmath17 for evaluating the equation of state of a 1d non - isodesmic lg model from the pdt ( [ pdt ] ) .
single - hatched cells denote the constraint that a cell be empty , while double - hatched cells are constrained to be occupied .
the insertion cell is constrained to be empty but is identified by leaving it unshaded .
all remaining cells ( not shown ) have no constraints . factors of 2x arise from 1d symmetry about the insertion cell .
2 . insertion energies @xmath17 required when using the pdt ( [ pdt ] ) to evaluate probabilities of occupancy directly to the right of constrained cells or walls , defining the cluster distribution of a 1d non - isodesmic lg model .
( a ) empty wall : @xmath18 ; ( b ) monomer wall : @xmath20 ; ( c ) cluster wall : @xmath22 .
single - hatched cells denote the constraint that a cell be empty , while double - hatched cells are constrained to be occupied .
the insertion cell is constrained to be empty but is identified by leaving it unshaded .
all remaining cells ( not shown ) have no constraints .
@xmath57 + broken isodesmic symmetry can be an important aspect of linear aggregation in solution .
it is shown that experimental data on the self - assembly of peptide tapes can be mapped to exactly solvable one - dimensional models that incorporate this broken symmetry . | exactly solvable models of linear aggregation have been known since ising s seminal one - dimensional model .
this model is defined by a unique nearest - neighbour bond strength that is independent of the length of the cluster ; known as isodesmic symmetry .
linear aggregation in real systems has often been associated with broken isodesmic symmetry . here
we show that important examples can be mapped to a class of one - dimensional models that are also exactly solvable .
[ [ section ] ] a model has isodesmic symmetry if every member of the infinite set of cluster equilibria @xmath0+[n-1]\leftrightarrow [ n]$ ] define the same equilibrium ` constant ' , as is appropriate to linear aggregation ( self - assembly ) where the energy required to break a chain aggregate is independent of the position of the break and the length of the chain .
the one dimensional ( 1d ) ising @xcite and equivalent lattice - gas ( lg ) models are isodesmic everywhere in phase - space .
we shall consider the generalisation where the first member of the set only , @xmath0+[1]\leftrightarrow [ 2]$ ] , has a different equilibrium constant to all the rest . when treated phenomenologically with the `` law of mass action '' as an implicit solvent model
this is a well - known problem in the physical chemistry of aggregation @xcite and is also the basis of current understanding of the self - assembly of peptide tapes involved in protein miss - folding diseases such as alzheimer s dementia @xcite .
+ let us specify a 1d lg hamiltonian with two energies ; namely , an energy @xmath1 that an isolated monomer @xmath0 $ ] must take up in the process of bonding to a cluster , and an attractive nearest - neighbour energy @xmath2 for each bond within a cluster of length @xmath3 . then
, when a monomer joins a cluster @xmath4 $ ] the change in energy involved is @xmath5 , but when two monomers combine to form a dimer the gain in attractive energy is only @xmath6 .
let us use the exponential factor @xmath7 to represent a temperature field ( @xmath8 denotes inverse temperature in units of boltzmann s constant ) and it is convenient to define @xmath9 .
a direct route to the exact equation of state and cluster distribution is to apply widom s potential distribution theorem ( pdt ) , @xcite : @xmath10 where @xmath11 denotes the monomer activity ( @xmath12 is chemical potential ) , @xmath13 is the density in lg units ( probability of occupancy of a single cell ) , the subscript @xmath14 indicates a conditional average ( there is an empty cell at position @xmath15 enforced by the @xmath16 ) factor on the left side of the equation ) and @xmath17 denotes the change in energy that would arise if a monomer were to be inserted into this empty cell .
the evaluation of this average ( or ratio of partition functions ) can be broken down into insertions into a finite set of specific clusters ; let us use @xmath18 , @xmath19 for occupancy next to an empty cell , @xmath20 , @xmath21 for occupancy next to a monomer isolated on the other side and @xmath22 , @xmath23 for occupancy next to the end of an fixed aggregate ( needed below to evaluate the cluster distribution ) .
figure 1 enumerates all the possibilities required for our model , together with the corresponding values of @xmath17 .
one can then use conditional probability to evaluate ( [ pdt ] ) by inspection ( this can be done in any order but here i shall start at the insertion cell and move right and left ) : @xmath24\right . \nonumber\\ & + & \tau^{\gamma}\left[\rho_w(1-\rho_1)\tau^{1-\gamma}+\rho_w\rho_1\tau\right]^2 \left.\right\ } \,\ , .
\label{eos1}\end{aligned}\ ] ] to evaluate @xmath19 , @xmath21 ( and @xmath23 ) , one simply applies the pdt to insertions in the presence of the fixed clusters at the left of the insertion cells depicted in fig . 2 ;
each case is associated with three non - degenerate states . from fig .
2a we have @xmath25\right\}\\ \,\ , , \label{fw}\ ] ] while from fig .
2b and fig .
2c , respectively , @xmath26 inserting ( [ fw ] ) into ( [ eos1 ] ) leads to a simplification : @xmath27 ^ 2\,\ , .
\label{eos}\ ] ] to evaluate @xmath28 and hence also obtain @xmath29 , @xmath21 and @xmath23 , one can solve ( [ fw]-[f1fa ] ) as a pair of simultaneous equations ; e.g. @xmath30 the solution yields @xmath31 together with a cubic equation of state for @xmath19 : @xmath32 although the general mathematical solution for one - dimensional models is invariably a cubic , for short - range models in equilibrium there must never be more than one physical solution in order to preserve the absence of phase transitions in one - dimension @xcite . for our non - isodesmic model
we see that this arises because ( [ cubic ] ) is also a quadratic for the activity , which only ever has one physical solution : @xmath33 \,\,.\label{soln}\ ] ] in this version of the equation of state the variables @xmath34 and @xmath19 are a field and a density , respectively .
one can transform the equation of state into @xmath35 by using @xmath19 as an implicit variable varying between zero and infinity in both equations ( [ soln ] ) and ( [ eos ] ) .
the occupation variables @xmath21 and @xmath23 similarly follow from ( [ fwf1 ] ) and ( [ fa ] ) , respectively .
note that there is no region of phase space associated with imaginary solutions contained within the above analysis .
+ the entire cluster distribution is also defined by the above occupation probabilities : @xmath36 & = & ( 1-\rho)\rho_w(1-\rho_1 ) \nonumber\\ & = & ( 1-\rho)(1-\rho_w)\rho_a\tau^{-1 } \,\,,\label{one}\\ \left[n>1\right ] & = & ( 1-\rho)\rho_w\rho_1\rho_a^{n-2}(1-\rho_a)\nonumber\\ & = & [ 1]\tau^{-\gamma}\rho_a^{n-1}\,\ , .
\label{cdis}\end{aligned}\ ] ] here , i have chosen to build the clusters in strict order of left to right and note that an isolated monomer is actually a cluster of two empty sites enclosing an occupied site .
note also that to obtain the second line of ( [ one ] ) one requires eqn .
( [ f1fw ] ) .
it is straightforward to directly check the self - consistency of the pdt by showing that ( [ eos ] ) is also the solution of the sum rule @xmath37 + \sum_{n>1}\,\,n[n ] \,\ , .
\label{srrho}\ ] ] the clusters @xmath4 $ ] satisfy the isodesmic exponential distribution @xmath4=[1]\tau^{-\gamma}\rho_a^{n-1}$ ] , renormalized by a factor @xmath38 , and thus have an average cluster length @xmath39}{\sum_{n>1}[n ] } = 2+f_a = 2+f_1\tau^{\gamma}\,\ , .
\label{nbar}\ ] ] the only effect of the broken isodesmic symmetry is to cause the aggregation distribution to step down a factor of @xmath38 between isolated monomers ( @xmath40 ) and dimers ( @xmath41 ) . in the case of peptide tape formation ,
the broken isodesmic symmetry ( @xmath42 ) results from the energy barrier that must be overcome by a helical monomer when unwinding in the process of joining a beta - sheet tape .
experimental procedures exist that can directly measure the percentage of peptide in a beta - sheet confirmation , @xcite .
this defines the order parameter @xmath43 = \frac{2f_1+f_1 ^ 2\tau^{\gamma}}{1 + 2f_1+f_1 ^ 2\tau^{\gamma}}\,\ , .
\label{phit}\ ] ] these results are essentially identical to the free - energy minimization proposed by aggeli et al @xcite .
the only mathematical difference is that the lg model has an additional factor of @xmath44 multiplying each cluster probability density @xmath45 $ ] , to allow for arbitrarily high concentration .
from the geometry of typical peptide tapes @xcite one notes that a monomer concentration of @xmath46 molar maps roughly to @xmath47 in lg units . since the experimental data for @xmath48 in fig .
6b of aggeli et al is at concentrations less than 500 micromolar , this factor is completely irrelevant .
accordingly , we have identified an exactly solvable lg model with identical aggregation statistics to a known phenomenological analysis that has been successfully fitted to experimental data .
the data imply a value of @xmath49 around 0.2 .
there is however one physical difference with the free - energy analysis adopted in ref.@xcite ; namely , instead of normalising by the volume of a monomer ( as is automatic in the context of a 1d lg model ) the authors choose a volume approximately 1000 times smaller . from ( [ one],[cdis ] ) this is equivalent to increasing the values of both @xmath34 and @xmath50 by 1000 , which is not sufficiently inconsistent at @xmath51 to noticeably affect the fit to the data .
however , it does result in a significant renormalization of the predicted binding energy @xmath52 associated with the self - assembly .
aggeli et al describe their small volume as characterising the movement of monomers within a tape and thus might be regarded as a rough correction for the entropy cost associated with forming a linear aggregate in a 3d solution , in contrast to a 1d model .
+ by breaking the isodesmic symmetry with a non - zero value of @xmath49 the self - assembly of dimers is suppressed to higher concentrations , after which the rest of the cluster distribution kicks in . this induces a sigmoidal shape to @xmath48 just before the percentage of beta - sheet self - assembly reaches 50% : @xmath53 there is no discontinuous jump in the order parameter @xmath54 ( no van der waals ` loop ' in @xmath55 ) , since as already discussed there is no region of two - phase coexistence for these short - range 1d models .
this behaviour is reminiscent of micelle formation ; since the ` micelles ' are tapes , the concentration at which this rapid increase in linear aggregation takes place can be denoted a ` ctc ' .
it is therefore of some interest that we have identified a class of exactly solvable 1d models that display the phenomenon of a ctc .
these broken isodesmic lg models are defined explicitly by the hamiltonian @xmath56 a non - zero value of @xmath49 is therefore mathematically equivalent to the introduction of a three - body interaction .
99 e. ising , z. physik * 31 * , 253 ( 1925 ) .
r. b. martin , chemical reviews * 96 * , 3043 ( 1996 ) .
one argument is that the self - assembly of a dimer involves twice the loss of orientational entropy than when a monomer joins a cluster .
a. aggeli , m. bell , n. boden , j. n. keen , t. c. b. mcleish , i. nyrkova , s. e. radford and a. semenov , j. mater
. chem .
* 7 * , 1135 ( 1997 ) .
b. widom , j. chem .
phys . * 39 * , 2808 ( 1963 ) . in the current context ( monomer insertion )
this sum rule is best derived from enforcing detailed balance between the system of interest and an ideal - gas reservoir of monomers at the same temperature and chemical potential .
l. d. landau and e. m. lifshitz , _ statistical physics _
3rd ed .
part i , ( pergamon , new york , 1980 ) ; final page . c. whitehouse , j. fang , a. aggeli , m. bell , r. brydson , c. w. g. fishwick , j. r. henderson , c. m. knobler , r. w. owens , n. h. thomson , d. a. smith , and n. boden , angew .
chem .
int . ed .
* 44 * , 1965 ( 2005 ) ; see fig .
2 . |
Subsets and Splits
No saved queries yet
Save your SQL queries to embed, download, and access them later. Queries will appear here once saved.